Update tflite_support to v0.3.1

Quite a few patches needed, see the README for more details.

Sadly this does come with a binary size increase because of additional
files needed from tflite. This update is important for continued
code stability and security.

Bug: 1248206
Binary-Size: Size increase is unavoidable (see above).
Change-Id: I17d553da036617d51897dad3a370cc09e7cd9eba
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/3345972
Reviewed-by: Dirk Pranke <dpranke@google.com>
Reviewed-by: Josh Simmons <jds@google.com>
Reviewed-by: Tommy Nyquist <nyquist@chromium.org>
Reviewed-by: Michael Crouse <mcrouse@chromium.org>
Reviewed-by: Ravjit Uppal <ravjit@chromium.org>
Commit-Queue: Robert Ogden <robertogden@chromium.org>
Cr-Commit-Position: refs/heads/main@{#956539}

components/optimization_guide/core/bert_model_executor.cc

@@ -7,7 +7,7 @@
 #include "base/trace_event/trace_event.h"
 #include "components/optimization_guide/core/model_util.h"
 #include "components/optimization_guide/core/tflite_op_resolver.h"
-#include "third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/bert_nl_classifier.h"
+#include "third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/bert_nl_classifier.h"
 
 namespace optimization_guide {
 
@@ -28,18 +28,21 @@ BertModelExecutor::Execute(ModelExecutionTask* execution_task,
                GetStringNameForOptimizationTarget(optimization_target_),
                "input_length", input.size());
   *out_status = ExecutionStatus::kSuccess;
-  return static_cast<tflite::task::text::nlclassifier::BertNLClassifier*>(
-             execution_task)
+  return static_cast<tflite::task::text::BertNLClassifier*>(execution_task)
       ->Classify(input);
 }
 
 std::unique_ptr<BertModelExecutor::ModelExecutionTask>
 BertModelExecutor::BuildModelExecutionTask(base::MemoryMappedFile* model_file,
                                            ExecutionStatus* out_status) {
+  tflite::task::text::BertNLClassifierOptions options;
+  *options.mutable_base_options()
+       ->mutable_model_file()
+       ->mutable_file_content() = std::string(
+      reinterpret_cast<const char*>(model_file->data()), model_file->length());
   auto maybe_nl_classifier =
-      tflite::task::text::nlclassifier::BertNLClassifier::CreateFromBuffer(
-          reinterpret_cast<const char*>(model_file->data()),
-          model_file->length(), std::make_unique<TFLiteOpResolver>());
+      tflite::task::text::BertNLClassifier::CreateFromOptions(
+          std::move(options), std::make_unique<TFLiteOpResolver>());
   if (maybe_nl_classifier.ok())
     return std::move(maybe_nl_classifier.value());
   *out_status = ExecutionStatus::kErrorModelFileNotValid;

components/optimization_guide/core/model_validator.cc

@@ -64,7 +64,12 @@ absl::Status ModelValidatorExecutor::Preprocess(
 float ModelValidatorExecutor::Postprocess(
     const std::vector<const TfLiteTensor*>& output_tensors) {
   std::vector<float> data;
-  tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  absl::Status status =
+      tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  if (!status.ok()) {
+    NOTREACHED();
+    return -1;
+  }
   return data[0];
 }
 

components/optimization_guide/core/test_tflite_model_executor.cc

@@ -11,14 +11,15 @@ namespace optimization_guide {
 absl::Status TestTFLiteModelExecutor::Preprocess(
     const std::vector<TfLiteTensor*>& input_tensors,
     const std::vector<float>& input) {
-  tflite::task::core::PopulateTensor<float>(input, input_tensors[0]);
-  return absl::OkStatus();
+  return tflite::task::core::PopulateTensor<float>(input, input_tensors[0]);
 }
 
 std::vector<float> TestTFLiteModelExecutor::Postprocess(
     const std::vector<const TfLiteTensor*>& output_tensors) {
   std::vector<float> data;
-  tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  absl::Status status =
+      tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  DCHECK(status.ok());
   return data;
 }
 

components/permissions/prediction_service/prediction_model_executor.cc

@@ -28,43 +28,90 @@ absl::Status PredictionModelExecutor::Preprocess(
       NOTREACHED();
   }
 
-  tflite::task::core::PopulateTensor<float>(
+  absl::Status status = tflite::task::core::PopulateTensor<float>(
       input.client_features().client_stats().avg_deny_rate(), input_tensors[0]);
-  tflite::task::core::PopulateTensor<float>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<float>(
       input.client_features().client_stats().avg_dismiss_rate(),
       input_tensors[1]);
-  tflite::task::core::PopulateTensor<float>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<float>(
       input.client_features().client_stats().avg_grant_rate(),
       input_tensors[2]);
-  tflite::task::core::PopulateTensor<float>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<float>(
       input.client_features().client_stats().avg_ignore_rate(),
       input_tensors[3]);
-  tflite::task::core::PopulateTensor<float>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<float>(
       input.permission_features()[0].permission_stats().avg_deny_rate(),
       input_tensors[4]);
-  tflite::task::core::PopulateTensor<float>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<float>(
       input.permission_features()[0].permission_stats().avg_dismiss_rate(),
       input_tensors[5]);
-  tflite::task::core::PopulateTensor<float>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<float>(
       input.permission_features()[0].permission_stats().avg_grant_rate(),
       input_tensors[6]);
-  tflite::task::core::PopulateTensor<float>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<float>(
       input.permission_features()[0].permission_stats().avg_ignore_rate(),
       input_tensors[7]);
-  tflite::task::core::PopulateTensor<int64_t>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<int64_t>(
       static_cast<int64_t>(
           input.permission_features()[0].permission_stats().prompts_count()),
       input_tensors[8]);
-  tflite::task::core::PopulateTensor<int64_t>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<int64_t>(
       static_cast<int64_t>(
           input.client_features().client_stats().prompts_count()),
       input_tensors[9]);
-  tflite::task::core::PopulateTensor<int64_t>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<int64_t>(
       static_cast<int64_t>(input.client_features().gesture_enum()),
       input_tensors[10]);
-  tflite::task::core::PopulateTensor<int64_t>(
+  if (!status.ok()) {
+    return status;
+  }
+
+  status = tflite::task::core::PopulateTensor<int64_t>(
       static_cast<int64_t>(input.client_features().platform_enum()),
       input_tensors[11]);
+  if (!status.ok()) {
+    return status;
+  }
 
   return absl::OkStatus();
 }
@@ -73,7 +120,10 @@ GeneratePredictionsResponse PredictionModelExecutor::Postprocess(
   DCHECK(request_type_ == RequestType::kNotifications ||
          request_type_ == RequestType::kGeolocation);
   std::vector<float> data;
-  tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  absl::Status status =
+      tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  DCHECK(status.ok());
+
   GeneratePredictionsResponse response;
   float threshold = request_type_ == RequestType::kNotifications
                         ? kNotificationPredictionsThreshold

components/segmentation_platform/internal/execution/segmentation_model_executor.cc

@@ -31,8 +31,7 @@ absl::Status SegmentationModelExecutor::Preprocess(
         "length of input data does not match length of tensor");
   }
 
-  tflite::task::core::PopulateTensor<float>(input, input_tensors[0]);
-  return absl::OkStatus();
+  return tflite::task::core::PopulateTensor<float>(input, input_tensors[0]);
 }
 
 float SegmentationModelExecutor::Postprocess(
@@ -43,7 +42,12 @@ float SegmentationModelExecutor::Postprocess(
   DCHECK_EQ(1u, output_tensors[0]->bytes / sizeof(output_tensors[0]->type));
 
   std::vector<float> data;
-  tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  absl::Status status =
+      tflite::task::core::PopulateVector<float>(output_tensors[0], &data);
+  if (!status.ok()) {
+    NOTREACHED();
+    return -1;
+  }
   DCHECK_EQ(1u, data.size());
   return data[0];
 }

components/translate/core/language_detection/language_detection_model.cc

@@ -4,7 +4,7 @@
 
 #include "components/translate/core/language_detection/language_detection_model.h"
 
-#include "base/files/memory_mapped_file.h"
+#include "base/cxx17_backports.h"
 #include "base/metrics/histogram_macros.h"
 #include "base/metrics/histogram_macros_local.h"
 #include "base/strings/utf_string_conversions.h"
@@ -73,19 +73,27 @@ void LanguageDetectionModel::UpdateWithFile(base::File model_file) {
   if (!model_file.IsValid())
     return;
 
-  if (!model_fb_.Initialize(std::move(model_file)))
-    return;
-
   recorder.set_state(
       LanguageDetectionModelState::kModelFileValidAndMemoryMapped);
 
-  auto statusor_classifier = tflite::task::text::nlclassifier::NLClassifier::
-      CreateFromBufferAndOptions(
-          reinterpret_cast<const char*>(model_fb_.data()), model_fb_.length(),
-          {.input_tensor_index = 0,
-           .output_score_tensor_index = 0,
-           .output_label_tensor_index = 2},
-          CreateLangIdResolver());
+  tflite::task::text::NLClassifierOptions options;
+  options.set_input_tensor_index(0);
+  options.set_output_score_tensor_index(0);
+  options.set_output_label_tensor_index(2);
+
+  std::string file_content(model_file.GetLength(), '\0');
+  int bytes_read =
+      model_file.Read(0, base::data(file_content), model_file.GetLength());
+  if (bytes_read != model_file.GetLength()) {
+    return;
+  }
+  *options.mutable_base_options()
+       ->mutable_model_file()
+       ->mutable_file_content() = std::move(file_content);
+
+  auto statusor_classifier =
+      tflite::task::text::nlclassifier::NLClassifier::CreateFromOptions(
+          options, CreateLangIdResolver());
   if (!statusor_classifier.ok()) {
     LOCAL_HISTOGRAM_BOOLEAN("LanguageDetection.TFLiteModel.InvalidModelFile",
                             true);

components/translate/core/language_detection/language_detection_model.h

@@ -6,7 +6,8 @@
 #define COMPONENTS_TRANSLATE_CORE_LANGUAGE_DETECTION_LANGUAGE_DETECTION_MODEL_H_
 
 #include <string>
-#include "base/files/memory_mapped_file.h"
+
+#include "base/files/file.h"
 
 namespace tflite {
 namespace task {
@@ -70,11 +71,6 @@ class LanguageDetectionModel {
   std::pair<std::string, float> DetectTopLanguage(
       const std::string& sampled_str) const;
 
-  // A memory-mapped file that contains the TFLite model used for
-  // determining the language of a page. This must be valid in order
-  // to evaluate the model owned by |this|.
-  base::MemoryMappedFile model_fb_;
-
   // The tflite classifier that can determine the language of text.
   std::unique_ptr<tflite::task::text::nlclassifier::NLClassifier>
       lang_detection_model_;

infra/inclusive_language_presubmit_exempt_dirs.txt

@@ -734,15 +734,30 @@ third_party/tensorflow-text/src/tensorflow_text/python/benchmarks/test_data/unca
 third_party/tensorflow-text/src/tensorflow_text/python/metrics 1 1
 third_party/tensorflow-text/src/tensorflow_text/python/ops/test_data 1 1
 third_party/test_fonts 15 1
+third_party/tflite_support/patches 8 2
+third_party/tflite_support/src 1 1
+third_party/tflite_support/src/tensorflow_lite_support/acceleration 3 1
+third_party/tflite_support/src/tensorflow_lite_support/c 2 2
+third_party/tflite_support/src/tensorflow_lite_support/cc/port/default 1 1
 third_party/tflite_support/src/tensorflow_lite_support/cc/task 4 1
-third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision 9 4
-third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/proto 4 2
+third_party/tflite_support/src/tensorflow_lite_support/cc/task/audio 1 1
+third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/proto 2 1
+third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision 13 5
+third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/proto 12 4
+third_party/tflite_support/src/tensorflow_lite_support/cc/test/testdata/task/text 5 3
 third_party/tflite_support/src/tensorflow_lite_support/codegen 1 1
 third_party/tflite_support/src/tensorflow_lite_support/custom_ops/kernel/sentencepiece 2 1
 third_party/tflite_support/src/tensorflow_lite_support/custom_ops/kernel/sentencepiece/testdata 1 1
 third_party/tflite_support/src/tensorflow_lite_support/custom_ops/python 2 1
-third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop 4 2
+third_party/tflite_support/src/tensorflow_lite_support/custom_ops/testdata 5 1
+third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop 8 3
+third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python 12 3
+third_party/tflite_support/src/tensorflow_lite_support/java/src/java/org/tensorflow/lite/support/image/ops 1 1
+third_party/tflite_support/src/tensorflow_lite_support/java/src/java/org/tensorflow/lite/task/audio/classifier 1 1
+third_party/tflite_support/src/tensorflow_lite_support/metadata 5 1
 third_party/tflite_support/src/tensorflow_lite_support/metadata/java/src/java/org/tensorflow/lite/support/metadata 3 1
+third_party/tflite_support/src/tensorflow_lite_support/metadata/python/tests/testdata/nl_classifier 2 1
+third_party/tflite_support/src/tensorflow_lite_support/odml/ios/image 1 1
 third_party/tlslite 1 1
 third_party/tlslite/patches 7 3
 third_party/tlslite/tlslite 3 2

third_party/tflite/BUILD.gn

@@ -414,8 +414,18 @@ static_library("tflite") {
     "src/tensorflow/lite/core/api/tensor_utils.cc",
     "src/tensorflow/lite/core/api/tensor_utils.h",
     "src/tensorflow/lite/core/subgraph.cc",
+    "src/tensorflow/lite/delegates/interpreter_utils.cc",
+    "src/tensorflow/lite/delegates/interpreter_utils.h",
     "src/tensorflow/lite/delegates/nnapi/nnapi_delegate.h",
     "src/tensorflow/lite/delegates/nnapi/nnapi_delegate_disabled.cc",
+    "src/tensorflow/lite/experimental/acceleration/configuration/delegate_registry.cc",
+    "src/tensorflow/lite/experimental/acceleration/configuration/delegate_registry.h",
+    "src/tensorflow/lite/experimental/acceleration/configuration/flatbuffer_to_proto.cc",
+    "src/tensorflow/lite/experimental/acceleration/configuration/flatbuffer_to_proto.h",
+    "src/tensorflow/lite/experimental/acceleration/configuration/proto_to_flatbuffer.cc",
+    "src/tensorflow/lite/experimental/acceleration/configuration/proto_to_flatbuffer.h",
+    "src/tensorflow/lite/experimental/acceleration/mini_benchmark/mini_benchmark.cc",
+    "src/tensorflow/lite/experimental/acceleration/mini_benchmark/mini_benchmark.h",
     "src/tensorflow/lite/experimental/resource/initialization_status.cc",
     "src/tensorflow/lite/experimental/resource/initialization_status.h",
     "src/tensorflow/lite/experimental/resource/lookup_interfaces.h",

third_party/tflite_support/BUILD.gn

@@ -16,13 +16,19 @@ config("tflite_support_config") {
 proto_library("tflite_support_proto") {
   proto_in_dir = "src"
   sources = [
+    "src/tensorflow_lite_support/cc/task/core/proto/base_options.proto",
     "src/tensorflow_lite_support/cc/task/core/proto/external_file.proto",
+    "src/tensorflow_lite_support/cc/task/text/proto/bert_nl_classifier_options.proto",
+    "src/tensorflow_lite_support/cc/task/text/proto/nl_classifier_options.proto",
     "src/tensorflow_lite_support/cc/task/vision/proto/bounding_box.proto",
     "src/tensorflow_lite_support/cc/task/vision/proto/class.proto",
     "src/tensorflow_lite_support/cc/task/vision/proto/classifications.proto",
     "src/tensorflow_lite_support/cc/task/vision/proto/image_classifier_options.proto",
   ]
   cc_generator_options = "lite=true:"
+
+  import_dirs = [ "//third_party/tflite/src" ]
+  proto_deps = [ "//third_party/tflite:tflite-config-proto" ]
 }
 
 config("tflite_support_flags") {
@@ -31,6 +37,9 @@ config("tflite_support_flags") {
     "-Wno-extern-c-compat",
     "-Wno-implicit-function-declaration",
     "-Wno-sign-compare",
+    "-Wno-ignored-attributes",
+    "-Wno-deprecated-declarations",
+    "-Wno-unused-variable",
   ]
   if (!is_win) {
     cflags_cc = [ "-frtti" ]
@@ -51,15 +60,14 @@ static_library("tflite_support") {
   sources = [
     "src/tensorflow_lite_support/cc/common.cc",
     "src/tensorflow_lite_support/cc/common.h",
-    "src/tensorflow_lite_support/cc/port/default/statusor.cc",
-    "src/tensorflow_lite_support/cc/port/default/statusor.h",
-    "src/tensorflow_lite_support/cc/port/default/statusor_internals.h",
     "src/tensorflow_lite_support/cc/port/default/tflite_wrapper.cc",
     "src/tensorflow_lite_support/cc/port/default/tflite_wrapper.h",
     "src/tensorflow_lite_support/cc/port/status_macros.h",
     "src/tensorflow_lite_support/cc/port/statusor.h",
     "src/tensorflow_lite_support/cc/task/core/base_task_api.h",
     "src/tensorflow_lite_support/cc/task/core/category.h",
+    "src/tensorflow_lite_support/cc/task/core/error_reporter.cc",
+    "src/tensorflow_lite_support/cc/task/core/error_reporter.h",
     "src/tensorflow_lite_support/cc/task/core/external_file_handler.cc",
     "src/tensorflow_lite_support/cc/task/core/external_file_handler.h",
     "src/tensorflow_lite_support/cc/task/core/proto/external_file_proto_inc.h",
@@ -68,10 +76,15 @@ static_library("tflite_support") {
     "src/tensorflow_lite_support/cc/task/core/task_utils.h",
     "src/tensorflow_lite_support/cc/task/core/tflite_engine.cc",
     "src/tensorflow_lite_support/cc/task/core/tflite_engine.h",
-    "src/tensorflow_lite_support/cc/task/text/nlclassifier/bert_nl_classifier.cc",
-    "src/tensorflow_lite_support/cc/task/text/nlclassifier/bert_nl_classifier.h",
+    "src/tensorflow_lite_support/cc/task/processor/image_preprocessor.cc",
+    "src/tensorflow_lite_support/cc/task/processor/image_preprocessor.h",
+    "src/tensorflow_lite_support/cc/task/processor/processor.cc",
+    "src/tensorflow_lite_support/cc/task/processor/processor.h",
+    "src/tensorflow_lite_support/cc/task/text/bert_nl_classifier.cc",
+    "src/tensorflow_lite_support/cc/task/text/bert_nl_classifier.h",
     "src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.cc",
     "src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h",
+    "src/tensorflow_lite_support/cc/task/text/proto/nl_classifier_options_proto_inc.h",
     "src/tensorflow_lite_support/cc/task/vision/core/classification_head.cc",
     "src/tensorflow_lite_support/cc/task/vision/core/classification_head.h",
     "src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.cc",
@@ -126,8 +139,10 @@ static_library("tflite_support") {
   configs -= [ "//build/config/compiler:chromium_code" ]
 
   configs += [
-    ":tflite_support_flags",
     "//build/config/compiler:no_chromium_code",
+
+    # Must be after no_chromium_code for warning flags to be ordered correctly.
+    ":tflite_support_flags",
   ]
 
   public_configs = [ ":tflite_support_config" ]

third_party/tflite_support/README.chromium

@@ -1,8 +1,8 @@
 Name: TensorFlow Lite Support
 Short Name: tflite-support
 URL: https://github.com/tensorflow/tflite-support
-Version: 3faaca9c6a3b22dec4d636b6b092431c9ac409e8
-Date: 2021/01/05
+Version: v0.3.1
+Date: 2021/12/16
 License: Apache 2.0
 License File: LICENSE
 Security Critical: Yes
@@ -13,27 +13,30 @@ TFLite Support is a toolkit that helps users to develop ML and deploy TFLite
 models onto mobile devices.  It works cross-Platform and is supported on
 Java, C++ (WIP), and Swift (WIP).
 
-Modifications:
-- Use chromium's logging utility in place of glog (patches/0001-use-base-logging.patch)
-- Use size_t rather than int for loops according to chromium style (0001-use-size_t.patch)
-- Rely on re::StringPiece instead of absl::string_view (0001-use_StringPiece-for_string_view.patch)
-- Remove unsafe use of conversions between NSString and string by using SysNSStringToUTF8. Note, this
-  is unused code but required for presubmit checks. (0001-use-SysNSStringToUTF8.patch)
-- Remove usage of absl::Cord in tflite::support::CreateStatusWithPayload (0001-no-absl-cord.patch)
-- Use _Exit instead of _exit to work on all platforms (0001-use-exit.patch)
-- Remove external file handlers support for memory mapping files to support Windows
-  (0001-remove-unsupported-memory-map-from-file-handler.patch)
-- Fixes sign compare issues in tflite-support (0001-task-utils-sign-compare.patch)
-- Remove support for sentencepiece tokenizers (0001-no-sentencepiece-tokenizer.patch)
-- Allows for the max sequence used by BERT models to be 512 instead of 128 (0001-bert-max-seq-len.patch)
-- Ensure name field in metadata exists before checking for tflite metadata (0001-add-metadata-name-check.patch)
-
 Third party dependencies:
 - tflite
 - libzip
 - utf
 - tensorflow-text
 
+Modifications:
+01) Use re2::StringPiece instead of absl::string_view in regex_tokenizer.cc
+02) Remove support for sentencepiece tokenization because the required overhead
+isn't worth adding this functionality, esp since no feature team needs it.
+03) [To Be Upstreamed] Remove unused functions.
+04) Remove the ABSL_DEPRECATED annotation from a deprecated struct since this
+is a no-op in chromium builds and upsets clang.
+05) [To Be Upstreamed] Use size_t in for loop in nl_classifier.h
+06) [To Be Upstreamed] Remove unused variable in task_utils.h
+07) Do not use absl::any since it is not supported in chromium
+08) [To Be Upstreamed] Remove unused stl include in tokenizer_jni_lib.h
+09) Remove unbuilt files that triggered checkdeps warnings, and fix file perms.
+10) Remove memory mapped file support in external_file_handler.cc since it is
+only available on POSIX systems.
+
 Update Process:
 1) Clone the tflite-support github repo at the desired commit into src/
 2) Apply each patch listed above residing in patches/ using `git apply patches/$PATCHFILE`
+3) Get the build working.
+4) Record the patches made with `git format-patches HEAD -<number of changes>`
+

third_party/tflite_support/patches/0001-Fix-signed-comparison-in-base_vision_task_api.h.patch

@@ -1,34 +0,0 @@
-From c8bdfe3f6b3ce087c36b551d668b97101f620bdc Mon Sep 17 00:00:00 2001
-From: Daniel Rubery <drubery@chromium.org>
-Date: Thu, 6 May 2021 11:45:48 -0700
-Subject: [PATCH] Fix signed comparison in base_vision_task_api.h
-
----
- .../cc/task/vision/core/base_vision_task_api.h                | 4 ++--
- 1 file changed, 2 insertions(+), 2 deletions(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/base_vision_task_api.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/base_vision_task_api.h
-index 3d1359685f3f..c787876bec33 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/base_vision_task_api.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/base_vision_task_api.h
-@@ -204,7 +204,7 @@ class BaseVisionTaskApi
-         if (normalization_options.num_values == 1) {
-           float mean_value = normalization_options.mean_values[0];
-           float inv_std_value = (1.0f / normalization_options.std_values[0]);
--          for (int i = 0; i < input_data_byte_size / sizeof(uint8);
-+          for (size_t i = 0; i < input_data_byte_size / sizeof(uint8);
-                i++, input_data++, normalized_input_data++) {
-             *normalized_input_data =
-                 inv_std_value * (static_cast<float>(*input_data) - mean_value);
-@@ -214,7 +214,7 @@ class BaseVisionTaskApi
-               1.0f / normalization_options.std_values[0],
-               1.0f / normalization_options.std_values[1],
-               1.0f / normalization_options.std_values[2]};
--          for (int i = 0; i < input_data_byte_size / sizeof(uint8);
-+          for (size_t i = 0; i < input_data_byte_size / sizeof(uint8);
-                i++, input_data++, normalized_input_data++) {
-             *normalized_input_data = inv_std_values[i % 3] *
-                                      (static_cast<float>(*input_data) -
--- 
-2.31.1.607.g51e8a6a459-goog
-

third_party/tflite_support/patches/0001-Remove-signed-comparison-in-frame_buffer.h.patch

@@ -1,25 +0,0 @@
-From 368b317061ba7deb1f42c52c5443c261bb6c03ea Mon Sep 17 00:00:00 2001
-From: Daniel Rubery <drubery@chromium.org>
-Date: Thu, 6 May 2021 11:40:37 -0700
-Subject: [PATCH] Remove signed comparison in frame_buffer.h
-
----
- .../tensorflow_lite_support/cc/task/vision/core/frame_buffer.h  | 2 +-
- 1 file changed, 1 insertion(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-index 22f63fc34d36..42ac080c4749 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-@@ -246,7 +246,7 @@ class FrameBuffer {
- 
-   // Returns plane indexed by the input `index`.
-   const Plane plane(int index) const {
--    if (index > -1 && index < planes_.size()) {
-+    if (index > -1 && static_cast<size_t>(index) < planes_.size()) {
-       return planes_[index];
-     }
-     return {};
--- 
-2.31.1.607.g51e8a6a459-goog
-

third_party/tflite_support/patches/0001-Remove-unused-qualifiers-in-frame_buffer.h.patch

@@ -1,38 +0,0 @@
-From b23fcde4753dbf5e4adc325e9ded16800f1d1bc5 Mon Sep 17 00:00:00 2001
-From: Daniel Rubery <drubery@chromium.org>
-Date: Thu, 6 May 2021 11:38:06 -0700
-Subject: [PATCH] Remove unused qualifiers in frame_buffer.h
-
----
- .../cc/task/vision/core/frame_buffer.h                      | 6 +++---
- 1 file changed, 3 insertions(+), 3 deletions(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-index 1556b7dfabef..22f63fc34d36 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-@@ -242,7 +242,7 @@ class FrameBuffer {
-         timestamp_(timestamp) {}
- 
-   // Returns number of planes.
--  const int plane_count() const { return planes_.size(); }
-+  int plane_count() const { return planes_.size(); }
- 
-   // Returns plane indexed by the input `index`.
-   const Plane plane(int index) const {
-@@ -256,10 +256,10 @@ class FrameBuffer {
-   const Dimension dimension() const { return dimension_; }
- 
-   // Returns FrameBuffer format.
--  const Format format() const { return format_; }
-+  Format format() const { return format_; }
- 
-   // Returns FrameBuffer orientation.
--  const Orientation orientation() const { return orientation_; }
-+  Orientation orientation() const { return orientation_; }
- 
-   // Returns FrameBuffer timestamp.
-   const absl::Time timestamp() const { return timestamp_; }
--- 
-2.31.1.607.g51e8a6a459-goog
-

third_party/tflite_support/patches/0001-Use-third_party-libyuv.patch

@@ -1,30 +0,0 @@
-From 226d36a5d12ca3080b1d0d9b450be949e418e318 Mon Sep 17 00:00:00 2001
-From: Daniel Rubery <drubery@chromium.org>
-Date: Thu, 6 May 2021 11:26:23 -0700
-Subject: [PATCH] Use third_party/libyuv
-
----
- .../cc/task/vision/utils/libyuv_frame_buffer_utils.cc           | 2 +-
- 1 file changed, 1 insertion(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc
-index f3cd0c70fe1b..b50b500bb5a4 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc
-@@ -23,12 +23,12 @@ limitations under the License.
- #include "absl/status/status.h"
- #include "absl/strings/str_cat.h"
- #include "absl/strings/str_format.h"
--#include "include/libyuv.h"
- #include "tensorflow_lite_support/cc/common.h"
- #include "tensorflow_lite_support/cc/port/integral_types.h"
- #include "tensorflow_lite_support/cc/port/status_macros.h"
- #include "tensorflow_lite_support/cc/port/statusor.h"
- #include "tensorflow_lite_support/cc/task/vision/utils/frame_buffer_common_utils.h"
-+#include "third_party/libyuv/include/libyuv.h"
- 
- namespace tflite {
- namespace task {
--- 
-2.31.1.607.g51e8a6a459-goog
-

third_party/tflite_support/patches/0001-add-metadata-name-check.patch

@@ -1,28 +0,0 @@
-From 5e3e4b63a6bfd871afa16f8d27f2daa8b99d84e9 Mon Sep 17 00:00:00 2001
-From: mcrouse <mcrouse@google.com>
-Date: Thu, 19 Aug 2021 11:31:28 -0700
-Subject: [PATCH] add metadata name check
-
----
- .../metadata/cc/metadata_extractor.cc                        | 5 ++++-
- 1 file changed, 4 insertions(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/metadata/cc/metadata_extractor.cc b/third_party/tflite-support/src/tensorflow_lite_support/metadata/cc/metadata_extractor.cc
-index ad5df76f1c27b..42f2a7c13a516 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/metadata/cc/metadata_extractor.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/metadata/cc/metadata_extractor.cc
-@@ -159,7 +159,10 @@ absl::Status ModelMetadataExtractor::InitFromModelBuffer(
-   // Look for the "TFLITE_METADATA" field, if any.
-   for (int i = 0; i < model_->metadata()->size(); ++i) {
-     const auto metadata = model_->metadata()->Get(i);
--    if (metadata->name() && metadata->name()->str() != kMetadataBufferName) {
-+    if (!metadata->name()) {
-+      continue;
-+    }
-+    if (metadata->name()->str() != kMetadataBufferName) {
-       continue;
-     }
-     const auto buffer_index = metadata->buffer();
--- 
-2.33.0.rc2.250.ged5fa647cd-goog
-

third_party/tflite_support/patches/0001-bert-max-seq-len.patch

@@ -1,25 +0,0 @@
-From 49cd597b3c1fbfef2e3772682aa98575654131ba Mon Sep 17 00:00:00 2001
-From: Sophie Chang <sophiechang@chromium.org>
-Date: Mon, 1 Mar 2021 19:33:21 +0000
-Subject: [PATCH] allow for more tokens
-
----
- .../cc/task/text/nlclassifier/bert_nl_classifier.h              | 2 +-
- 1 file changed, 1 insertion(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/bert_nl_classifier.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/bert_nl_classifier.h
-index cd5c5a3ade03..e78085d98761 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/bert_nl_classifier.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/bert_nl_classifier.h
-@@ -52,7 +52,7 @@ class BertNLClassifier : public NLClassifier {
-  public:
-   using NLClassifier::NLClassifier;
-   // Max number of tokens to pass to the model.
--  static constexpr int kMaxSeqLen = 128;
-+  static constexpr int kMaxSeqLen = 512;
- 
-   // Factory function to create a BertNLClassifier from TFLite model with
-   // metadata.
--- 
-2.30.1.766.gb4fecdf3b7-goog
-

third_party/tflite_support/patches/0001-no-absl-cord.patch

@@ -1,33 +0,0 @@
-From 61fb20a08d2325d03759a5b9394c033901fc0a7f Mon Sep 17 00:00:00 2001
-From: Sophie Chang <sophiechang@chromium.org>
-Date: Wed, 3 Feb 2021 04:21:19 +0000
-Subject: [PATCH] do not use cord in tflite status payload
-
----
- .../tflite-support/src/tensorflow_lite_support/cc/common.cc    | 3 ---
- 1 file changed, 3 deletions(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/common.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/common.cc
-index 47dd3bcc6581..ed373e96d555 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/common.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/common.cc
-@@ -15,7 +15,6 @@ limitations under the License.
- 
- #include "tensorflow_lite_support/cc/common.h"
- 
--#include "absl/strings/cord.h"
- #include "absl/strings/str_cat.h"
- 
- namespace tflite {
-@@ -26,8 +25,6 @@ absl::Status CreateStatusWithPayload(absl::StatusCode canonical_code,
-                                      TfLiteSupportStatus tfls_code) {
-   // NOTE: Ignores `message` if the canonical code is ok.
-   absl::Status status = absl::Status(canonical_code, message);
--  // NOTE: Does nothing if the canonical code is ok.
--  status.SetPayload(kTfLiteSupportPayload, absl::Cord(absl::StrCat(tfls_code)));
-   return status;
- }
- 
--- 
-2.30.0.365.g02bc693789-goog
-

third_party/tflite_support/patches/0001-task-utils-sign-compare.patch

@@ -1,34 +0,0 @@
-From f84b50f175efff54ee6a6ef795703907245260cd Mon Sep 17 00:00:00 2001
-From: Sophie Chang <sophiechang@chromium.org>
-Date: Wed, 10 Feb 2021 17:55:30 +0000
-Subject: [PATCH] fix sign issues
-
----
- .../src/tensorflow_lite_support/cc/task/core/task_utils.h     | 4 ++--
- 1 file changed, 2 insertions(+), 2 deletions(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/task_utils.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/task_utils.h
-index 744dbbfb0f80..ced3dbcae9e4 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/task_utils.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/task_utils.h
-@@ -119,7 +119,7 @@ inline void PopulateVector(const TfLiteTensor* tensor, std::vector<T>* data) {
-   const T* results = GetTensorData<T>(tensor);
-   size_t num = tensor->bytes / sizeof(tensor->type);
-   data->reserve(num);
--  for (int i = 0; i < num; i++) {
-+  for (size_t i = 0; i < num; i++) {
-     data->emplace_back(results[i]);
-   }
- }
-@@ -169,7 +169,7 @@ static TensorType* FindTensorByName(
-       tensor_metadatas->size() != tensors.size()) {
-     return nullptr;
-   }
--  for (int i = 0; i < tensor_metadatas->size(); i++) {
-+  for (flatbuffers::uoffset_t i = 0; i < tensor_metadatas->size(); i++) {
-     if (strcmp(name.data(), tensor_metadatas->Get(i)->name()->c_str()) == 0) {
-       return tensors[i];
-     }
--- 
-2.30.0.478.g8a0d178c01-goog
-

third_party/tflite_support/patches/0001-use-StringPiece-for-string_view.patch

@@ -1,38 +0,0 @@
-From 81287a62d65139f29c512fed88ed734bef2c33f5 Mon Sep 17 00:00:00 2001
-From: Michael Crouse <mcrouse@chromium.org>
-Date: Tue, 22 Dec 2020 14:25:39 -0800
-Subject: [PATCH] use StringPiece for string_view
-
----
- .../cc/text/tokenizers/regex_tokenizer.cc              | 10 +++++-----
- 1 file changed, 5 insertions(+), 5 deletions(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc
-index 38aff8805b30..44c43b2d5086 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc
-@@ -61,16 +61,16 @@ RegexTokenizer::RegexTokenizer(const std::string& regex_pattern,
- }
- 
- TokenizerResult RegexTokenizer::Tokenize(const std::string& input) {
--  absl::string_view leftover(input.data());
--  absl::string_view last_end = leftover;
-+  re2::StringPiece leftover(input.data());
-+  re2::StringPiece last_end = leftover;
- 
-   TokenizerResult result;
- 
-   // Keep looking for split points until we have reached the end of the input.
--  absl::string_view extracted_delim_token;
-+  re2::StringPiece extracted_delim_token;
-   while (RE2::FindAndConsume(&leftover, delim_re_, &extracted_delim_token)) {
--    absl::string_view token(last_end.data(),
--                            extracted_delim_token.data() - last_end.data());
-+    re2::StringPiece token(last_end.data(),
-+                           extracted_delim_token.data() - last_end.data());
-     bool has_non_empty_token = token.length() > 0;
- 
-     last_end = leftover;
--- 
-2.29.2.729.g45daf8777d-goog
-

third_party/tflite_support/patches/0001-use-SysNSStringToUTF8.patch

@@ -1,29 +0,0 @@
-From e4b8790a56487279b084fb59a2186a8bfd24b838 Mon Sep 17 00:00:00 2001
-From: Michael Crouse <mcrouse@chromium.org>
-Date: Thu, 7 Jan 2021 08:20:06 -0800
-Subject: [PATCH] use SysNSStringToUTF8
-
----
- .../tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm | 3 ++-
- 1 file changed, 2 insertions(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm b/third_party/tflite-support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm
-index 2a11bb673047..b82be34a9ab9 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm
-@@ -12,10 +12,11 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- See the License for the specific language governing permissions and
- limitations under the License.
- ==============================================================================*/
-+#inmport "base/strings/sys_string_conversions.h"
- #import "third_party/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.h"
- 
- std::string MakeString(NSString* str) {
--  return std::string([str UTF8String]);
-+  return SysNSStringToUTF8(str); 
- }
- 
- NSString* MakeNSString(const std::string& str) {
--- 
-2.29.2.729.g45daf8777d-goog
-

third_party/tflite_support/patches/0001-use-base-logging.patch

@@ -1,26 +0,0 @@
-From 5307d81798215dae084b5079e797fd4408040340 Mon Sep 17 00:00:00 2001
-From: Michael Crouse <mcrouse@chromium.org>
-Date: Tue, 22 Dec 2020 14:18:09 -0800
-Subject: [PATCH] use base logging
-
----
- .../src/tensorflow_lite_support/cc/port/default/statusor.cc     | 2 +-
- 1 file changed, 1 insertion(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc
-index 547a79192324..182c37e4aaf6 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc
-@@ -18,8 +18,8 @@ limitations under the License.
- 
- #include <utility>
- 
--#include <glog/logging.h>
- #include "absl/strings/str_cat.h"
-+#include "base/logging.h"
- 
- namespace tflite {
- namespace support {
--- 
-2.29.2.729.g45daf8777d-goog
-

third_party/tflite_support/patches/0001-use-exit.patch

@@ -1,25 +0,0 @@
-From 49de34d7489ba5218a822461a42786844a1e344b Mon Sep 17 00:00:00 2001
-From: Sophie Chang <sophiechang@chromium.org>
-Date: Wed, 3 Feb 2021 04:30:56 +0000
-Subject: [PATCH] use _Exit
-
----
- .../src/tensorflow_lite_support/cc/port/default/statusor.cc     | 2 +-
- 1 file changed, 1 insertion(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc
-index 182c37e4aaf6..058c0070f0da 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/port/default/statusor.cc
-@@ -50,7 +50,7 @@ void Helper::HandleInvalidStatusCtorArg(absl::Status* status) {
- void Helper::Crash(const absl::Status& status) {
-   LOG(FATAL) << "Attempting to fetch value instead of handling error "
-              << status;
--  _exit(1);
-+  _Exit(1);
- }
- 
- void ThrowBadStatusOrAccess(absl::Status status) {
--- 
-2.30.0.365.g02bc693789-goog
-

third_party/tflite_support/patches/0001-use-re2-StringPiece-for-RegexTokenizer-Tokenize.patch

@@ -0,0 +1,36 @@
+From b16b7af6f58ede0718fabf9c0da7495c79400c90 Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Wed, 15 Dec 2021 14:39:48 -0800
+Subject: [PATCH 01/11] use re2 StringPiece for RegexTokenizer::Tokenize
+
+---
+ .../cc/text/tokenizers/regex_tokenizer.cc                 | 8 ++++----
+ 1 file changed, 4 insertions(+), 4 deletions(-)
+
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc b/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc
+index 564f5f63a0584..832f9df42f824 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/regex_tokenizer.cc
+@@ -61,15 +61,15 @@ RegexTokenizer::RegexTokenizer(const std::string& regex_pattern,
+ }
+ 
+ TokenizerResult RegexTokenizer::Tokenize(const std::string& input) {
+-  absl::string_view leftover(input.data());
+-  absl::string_view last_end = leftover;
++  re2::StringPiece leftover(input.data());
++  re2::StringPiece last_end = leftover;
+ 
+   TokenizerResult result;
+ 
+   // Keep looking for split points until we have reached the end of the input.
+-  absl::string_view extracted_delim_token;
++  re2::StringPiece extracted_delim_token;
+   while (RE2::FindAndConsume(&leftover, delim_re_, &extracted_delim_token)) {
+-    absl::string_view token(last_end.data(),
++    re2::StringPiece token(last_end.data(),
+                             extracted_delim_token.data() - last_end.data());
+     bool has_non_empty_token = token.length() > 0;
+ 
+-- 
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0001-use-size_t.patch

@@ -1,25 +0,0 @@
-From ecb535154168358a72de6b51099a9549b970bce5 Mon Sep 17 00:00:00 2001
-From: Michael Crouse <mcrouse@chromium.org>
-Date: Tue, 22 Dec 2020 14:34:12 -0800
-Subject: [PATCH] use size_t
-
----
- .../cc/task/text/nlclassifier/nl_classifier.h                   | 2 +-
- 1 file changed, 1 insertion(+), 1 deletion(-)
-
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
-index d3c3bed2083d..4055a93467d4 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
-@@ -151,7 +151,7 @@ class NLClassifier : public core::BaseTaskApi<std::vector<core::Category>,
-       const std::string& name,
-       int index) {
-     if (metadata_array != nullptr && metadata_array->size() == tensors.size()) {
--      for (int i = 0; i < metadata_array->size(); i++) {
-+      for (size_t i = 0; i < metadata_array->size(); i++) {
-         if (strcmp(name.data(), metadata_array->Get(i)->name()->c_str()) == 0) {
-           return tensors[i];
-         }
--- 
-2.29.2.729.g45daf8777d-goog
-

third_party/tflite_support/patches/0002-sentencepiece-tokenization-not-supported.patch

@@ -1,16 +1,16 @@
-From 7faac3ddcbc05275d797dda64a9b9d7f2279ae1c Mon Sep 17 00:00:00 2001
-From: Sophie Chang <sophiechang@chromium.org>
-Date: Thu, 11 Feb 2021 00:53:47 +0000
-Subject: [PATCH] no sentencepiece tokenizer
+From 9aa45ea43f8d84db1e20674c294f9ab958b12d7e Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Wed, 15 Dec 2021 14:57:16 -0800
+Subject: [PATCH 02/11] sentencepiece tokenization not supported
 
 ---
- .../cc/text/tokenizers/tokenizer_utils.cc             | 11 -----------
- 1 file changed, 11 deletions(-)
+ .../cc/text/tokenizers/tokenizer_utils.cc          | 14 ++++----------
+ 1 file changed, 4 insertions(+), 10 deletions(-)
 
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc
-index 352c4a8c5e4f..46786fd7faf8 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc b/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc
+index 9abca9691f058..28f0137f54278 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_utils.cc
 @@ -20,7 +20,6 @@ limitations under the License.
  #include "tensorflow_lite_support/cc/port/status_macros.h"
  #include "tensorflow_lite_support/cc/text/tokenizers/bert_tokenizer.h"
@@ -19,11 +19,18 @@ index 352c4a8c5e4f..46786fd7faf8 100644
  #include "tensorflow_lite_support/metadata/metadata_schema_generated.h"
  
  namespace tflite {
-@@ -73,16 +72,6 @@ StatusOr<std::unique_ptr<Tokenizer>> CreateTokenizerFromProcessUnit(
-       return absl::make_unique<BertTokenizer>(vocab_buffer.data(),
+@@ -29,7 +28,6 @@ namespace text {
+ namespace tokenizer {
+ 
+ using ::tflite::ProcessUnit;
+-using ::tflite::SentencePieceTokenizerOptions;
+ using ::tflite::support::CreateStatusWithPayload;
+ using ::tflite::support::StatusOr;
+ using ::tflite::support::TfLiteSupportStatus;
+@@ -74,14 +72,10 @@ StatusOr<std::unique_ptr<Tokenizer>> CreateTokenizerFromProcessUnit(
                                                vocab_buffer.size());
      }
--    case ProcessUnitOptions_SentencePieceTokenizerOptions: {
+     case ProcessUnitOptions_SentencePieceTokenizerOptions: {
 -      const tflite::SentencePieceTokenizerOptions* options =
 -          tokenizer_process_unit->options_as<SentencePieceTokenizerOptions>();
 -      ASSIGN_OR_RETURN(absl::string_view model_buffer,
@@ -32,9 +39,13 @@ index 352c4a8c5e4f..46786fd7faf8 100644
 -      // TODO(b/160647204): Extract sentence piece model vocabulary
 -      return absl::make_unique<SentencePieceTokenizer>(model_buffer.data(),
 -                                                       model_buffer.size());
--    }
++      return CreateStatusWithPayload(
++        absl::StatusCode::kInvalidArgument,
++        "Chromium does not support sentencepiece tokenization",
++        TfLiteSupportStatus::kMetadataInvalidTokenizerError);
+     }
      case ProcessUnitOptions_RegexTokenizerOptions: {
        const tflite::RegexTokenizerOptions* options =
-           tokenizer_process_unit->options_as<RegexTokenizerOptions>();
 -- 
-2.30.0.478.g8a0d178c01-goog
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0003-rm-unused-func.patch

@@ -0,0 +1,224 @@
+From 7bd2e5f0e2bb560e55efc3dd86249ff42a10d08c Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Wed, 15 Dec 2021 15:08:59 -0800
+Subject: [PATCH 03/11] rm unused func
+
+---
+ .../vision/utils/libyuv_frame_buffer_utils.cc | 201 +-----------------
+ 1 file changed, 1 insertion(+), 200 deletions(-)
+
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc
+index 0ece48636504e..6fd3ca81c984c 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/utils/libyuv_frame_buffer_utils.cc
+@@ -1326,206 +1326,7 @@ absl::Status CropResize(const FrameBuffer& buffer, int x0, int y0, int x1,
+   }
+ }
+ 
+-// Returns the scaled dimension of the input_size best fit within the
+-// output_size bound while respecting the aspect ratio.
+-FrameBuffer::Dimension GetScaledDimension(FrameBuffer::Dimension input_size,
+-                                          FrameBuffer::Dimension output_size) {
+-  int original_width = input_size.width;
+-  int original_height = input_size.height;
+-  int bound_width = output_size.width;
+-  int bound_height = output_size.height;
+-  int new_width = original_width;
+-  int new_height = original_height;
+-
+-  // Try to fit the width first.
+-  new_width = bound_width;
+-  new_height = (new_width * original_height) / original_width;
+-
+-  // Try to fit the height if needed.
+-  if (new_height > bound_height) {
+-    new_height = bound_height;
+-    new_width = (new_height * original_width) / original_height;
+-  }
+-  return FrameBuffer::Dimension{.width = new_width, .height = new_height};
+-}
+-
+-// This method only supports kGRAY, kRGBA, and kRGB formats.
+-absl::Status UniformCropResizePlane(const FrameBuffer& buffer,
+-                                    std::vector<int> crop_coordinates,
+-                                    FrameBuffer* output_buffer) {
+-  int x0 = 0, y0 = 0;
+-  FrameBuffer::Dimension input_dimension = buffer.dimension();
+-  if (!crop_coordinates.empty()) {
+-    x0 = crop_coordinates[0];
+-    y0 = crop_coordinates[1];
+-    input_dimension =
+-        GetCropDimension(x0, crop_coordinates[2], y0, crop_coordinates[3]);
+-  }
+-  if (input_dimension == output_buffer->dimension()) {
+-    // Cropping only case.
+-    return CropPlane(buffer, x0, y0, crop_coordinates[2], crop_coordinates[3],
+-                     output_buffer);
+-  }
+-
+-  // Cropping is achieved by adjusting origin to (x0, y0).
+-  ASSIGN_OR_RETURN(int pixel_stride, GetPixelStrides(buffer.format()));
+-  int adjusted_offset =
+-      buffer.plane(0).stride.row_stride_bytes * y0 + x0 * pixel_stride;
+-  FrameBuffer::Plane plane = {
+-      /*buffer=*/buffer.plane(0).buffer + adjusted_offset,
+-      /*stride=*/{buffer.plane(0).stride.row_stride_bytes, pixel_stride}};
+-  auto adjusted_buffer =
+-      FrameBuffer::Create({plane}, input_dimension, buffer.format(),
+-                          buffer.orientation(), buffer.timestamp());
+-
+-  // Uniform resize is achieved by adjusting the resize dimension to fit the
+-  // output_buffer and respect the input aspect ratio at the same time. We
+-  // create an intermediate output buffer with adjusted dimension and point its
+-  // backing buffer to the output_buffer. Note the stride information on the
+-  // adjusted_output_buffer is not used in the Resize* methods.
+-  FrameBuffer::Dimension adjusted_dimension =
+-      GetScaledDimension(input_dimension, output_buffer->dimension());
+-  FrameBuffer::Plane output_plane = {/*buffer=*/output_buffer->plane(0).buffer,
+-                                     /*stride=*/output_buffer->plane(0).stride};
+-  auto adjusted_output_buffer = FrameBuffer::Create(
+-      {output_plane}, adjusted_dimension, output_buffer->format(),
+-      output_buffer->orientation(), output_buffer->timestamp());
+-
+-  switch (buffer.format()) {
+-    case FrameBuffer::Format::kRGB:
+-      return ResizeRgb(*adjusted_buffer, adjusted_output_buffer.get());
+-    case FrameBuffer::Format::kRGBA:
+-      return ResizeRgba(*adjusted_buffer, adjusted_output_buffer.get());
+-    case FrameBuffer::Format::kGRAY:
+-      return ResizeGray(*adjusted_buffer, adjusted_output_buffer.get());
+-    default:
+-      return CreateStatusWithPayload(
+-          StatusCode::kInternal,
+-          absl::StrFormat("Format %i is not supported.", buffer.format()),
+-          TfLiteSupportStatus::kImageProcessingError);
+-  }
+-}
+-
+-absl::Status UniformCropResizeYuv(const FrameBuffer& buffer,
+-                                  std::vector<int> crop_coordinates,
+-                                  FrameBuffer* output_buffer) {
+-  int x0 = 0, y0 = 0;
+-  FrameBuffer::Dimension input_dimension = buffer.dimension();
+-  if (!crop_coordinates.empty()) {
+-    x0 = crop_coordinates[0];
+-    y0 = crop_coordinates[1];
+-    input_dimension =
+-        GetCropDimension(x0, crop_coordinates[2], y0, crop_coordinates[3]);
+-  }
+-  if (input_dimension == output_buffer->dimension()) {
+-    // Cropping only case.
+-    int x1 = crop_coordinates[2];
+-    int y1 = crop_coordinates[3];
+-    switch (buffer.format()) {
+-      case FrameBuffer::Format::kNV12:
+-      case FrameBuffer::Format::kNV21:
+-        return CropNv(buffer, x0, y0, x1, y1, output_buffer);
+-      case FrameBuffer::Format::kYV12:
+-      case FrameBuffer::Format::kYV21:
+-        return CropYv(buffer, x0, y0, x1, y1, output_buffer);
+-      default:
+-        return CreateStatusWithPayload(
+-            StatusCode::kInternal,
+-            absl::StrFormat("Format %i is not supported.", buffer.format()),
+-            TfLiteSupportStatus::kImageProcessingError);
+-    }
+-  }
+-
+-  // Cropping is achieved by adjusting origin to (x0, y0).
+-  ASSIGN_OR_RETURN(FrameBuffer::YuvData input_data,
+-                   FrameBuffer::GetYuvDataFromFrameBuffer(buffer));
+-  // Cropping YUV planes by offsetting the origins of each plane.
+-  // TODO(b/152629712): Investigate the impact of color shifting caused by the
+-  // bounding box with odd X or Y starting positions.
+-  const int plane_y_offset = input_data.y_row_stride * y0 + x0;
+-  const int plane_uv_offset = input_data.uv_row_stride * (y0 / 2) +
+-                              input_data.uv_pixel_stride * (x0 / 2);
+-  FrameBuffer::Plane adjusted_plane_y = {
+-      /*buffer=*/input_data.y_buffer + plane_y_offset,
+-      /*stride=*/{input_data.y_row_stride, /*pixel_stride_bytes=*/1}};
+-  FrameBuffer::Plane adjusted_plane_u = {
+-      /*buffer=*/input_data.u_buffer + plane_uv_offset,
+-      /*stride=*/{input_data.uv_row_stride, input_data.uv_pixel_stride}};
+-  FrameBuffer::Plane adjusted_plane_v = {
+-      /*buffer=*/input_data.v_buffer + plane_uv_offset,
+-      /*stride=*/{input_data.uv_row_stride, input_data.uv_pixel_stride}};
+-
+-  // Uniform resize is achieved by adjusting the resize dimension to fit the
+-  // output_buffer and respect the input aspect ratio at the same time. For
+-  // YUV formats, we need access to the actual output dimension to get the
+-  // correct address of each plane. For this, we are not calling ResizeNv or
+-  // ResizeYv but the libyuv scale methods directly.
+-  FrameBuffer::Dimension adjusted_dimension =
+-      GetScaledDimension(input_dimension, output_buffer->dimension());
+-  ASSIGN_OR_RETURN(FrameBuffer::YuvData output_data,
+-                   FrameBuffer::GetYuvDataFromFrameBuffer(*output_buffer));
+-
+-  switch (buffer.format()) {
+-    case FrameBuffer::Format::kNV12: {
+-      int ret = libyuv::NV12Scale(
+-          adjusted_plane_y.buffer, adjusted_plane_y.stride.row_stride_bytes,
+-          adjusted_plane_u.buffer, adjusted_plane_u.stride.row_stride_bytes,
+-          input_dimension.width, input_dimension.height,
+-          const_cast<uint8_t*>(output_data.y_buffer), output_data.y_row_stride,
+-          const_cast<uint8_t*>(output_data.u_buffer), output_data.uv_row_stride,
+-          adjusted_dimension.width, adjusted_dimension.height,
+-          libyuv::FilterMode::kFilterBilinear);
+-      if (ret != 0) {
+-        return CreateStatusWithPayload(
+-            StatusCode::kUnknown, "Libyuv NV12Scale operation failed.",
+-            TfLiteSupportStatus::kImageProcessingBackendError);
+-      }
+-      return absl::OkStatus();
+-    }
+-    case FrameBuffer::Format::kNV21: {
+-      int ret = libyuv::NV12Scale(
+-          adjusted_plane_y.buffer, adjusted_plane_y.stride.row_stride_bytes,
+-          adjusted_plane_v.buffer, adjusted_plane_v.stride.row_stride_bytes,
+-          input_dimension.width, input_dimension.height,
+-          const_cast<uint8_t*>(output_data.y_buffer), output_data.y_row_stride,
+-          const_cast<uint8_t*>(output_data.v_buffer), output_data.uv_row_stride,
+-          adjusted_dimension.width, adjusted_dimension.height,
+-          libyuv::FilterMode::kFilterBilinear);
+-      if (ret != 0) {
+-        return CreateStatusWithPayload(
+-            StatusCode::kUnknown, "Libyuv NV12Scale operation failed.",
+-            TfLiteSupportStatus::kImageProcessingBackendError);
+-      }
+-      return absl::OkStatus();
+-    }
+-    case FrameBuffer::Format::kYV12:
+-    case FrameBuffer::Format::kYV21: {
+-      int ret = libyuv::I420Scale(
+-          adjusted_plane_y.buffer, adjusted_plane_y.stride.row_stride_bytes,
+-          adjusted_plane_u.buffer, adjusted_plane_u.stride.row_stride_bytes,
+-          adjusted_plane_v.buffer, adjusted_plane_v.stride.row_stride_bytes,
+-          input_dimension.width, input_dimension.height,
+-          const_cast<uint8_t*>(output_data.y_buffer), output_data.y_row_stride,
+-          const_cast<uint8_t*>(output_data.u_buffer), output_data.uv_row_stride,
+-          const_cast<uint8_t*>(output_data.v_buffer), output_data.uv_row_stride,
+-          adjusted_dimension.width, adjusted_dimension.height,
+-          libyuv::FilterMode::kFilterBilinear);
+-      if (ret != 0) {
+-        return CreateStatusWithPayload(
+-            StatusCode::kUnknown, "Libyuv I420Scale operation failed.",
+-            TfLiteSupportStatus::kImageProcessingBackendError);
+-      }
+-      return absl::OkStatus();
+-    }
+-    default:
+-      return CreateStatusWithPayload(
+-          StatusCode::kInternal,
+-          absl::StrFormat("Format %i is not supported.", buffer.format()),
+-          TfLiteSupportStatus::kImageProcessingError);
+-  }
+-  return absl::OkStatus();
+-}
+-}  // namespace
++} // namespace
+ 
+ absl::Status LibyuvFrameBufferUtils::Crop(const FrameBuffer& buffer, int x0,
+                                           int y0, int x1, int y1,
+-- 
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0004-rm-noop-deprecated-attribute.patch

@@ -0,0 +1,26 @@
+From 243aadd7dcea9be980aa89d183bf2dea7cba202b Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Wed, 15 Dec 2021 15:49:40 -0800
+Subject: [PATCH 04/11] rm noop deprecated attribute
+
+---
+ .../cc/task/text/nlclassifier/nl_classifier.h                  | 3 ---
+ 1 file changed, 3 deletions(-)
+
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
+index d5b49dfd75277..ac12536355db4 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
+@@ -43,9 +43,6 @@ namespace text {
+ namespace nlclassifier {
+ 
+ // Options to identify input and output tensors of the model
+-ABSL_DEPRECATED(
+-    "Prefer using `tflite::task::text::NLClassifierOptions` and "
+-    "`CreateFromOptions`")
+ struct NLClassifierOptions {
+   int input_tensor_index = 0;
+   int output_score_tensor_index = 0;
+-- 
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0005-use-size_t-in-for-loop.patch

@@ -0,0 +1,25 @@
+From 5d67933d8d4440816a02f7a319d7323041c3f7bf Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Wed, 15 Dec 2021 15:51:22 -0800
+Subject: [PATCH 05/11] use size_t in for loop
+
+---
+ .../cc/task/text/nlclassifier/nl_classifier.h                   | 2 +-
+ 1 file changed, 1 insertion(+), 1 deletion(-)
+
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
+index ac12536355db4..2adafba8f2fa9 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h
+@@ -179,7 +179,7 @@ class NLClassifier : public core::BaseTaskApi<std::vector<core::Category>,
+           metadata_array,
+       const std::string& name, int index) {
+     if (metadata_array != nullptr && metadata_array->size() == tensors.size()) {
+-      for (int i = 0; i < metadata_array->size(); i++) {
++      for (size_t i = 0; i < metadata_array->size(); i++) {
+         if (strcmp(name.data(), metadata_array->Get(i)->name()->c_str()) == 0) {
+           return tensors[i];
+         }
+-- 
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0006-unused-variable.patch

@@ -0,0 +1,29 @@
+From b2d06daf8ab5cff8748489407b6ad10ea600948d Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Thu, 16 Dec 2021 08:35:07 -0800
+Subject: [PATCH 06/11] unused variable
+
+---
+ .../src/tensorflow_lite_support/cc/task/core/task_utils.h   | 6 ++++--
+ 1 file changed, 4 insertions(+), 2 deletions(-)
+
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/task_utils.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/task_utils.h
+index 03ef9ade9af41..e95ea73a4a812 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/task_utils.h
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/task_utils.h
+@@ -144,8 +144,10 @@ inline absl::Status PopulateVector(const TfLiteTensor* tensor,
+ template <>
+ inline absl::Status PopulateVector<std::string>(
+     const TfLiteTensor* tensor, std::vector<std::string>* data) {
+-  std::string* v;
+-  ASSIGN_OR_RETURN(v, AssertAndReturnTypedTensor<std::string>(tensor));
++  if (tensor->type != typeToTfLiteType<std::string>()) {
++    return absl::InvalidArgumentError("not of type string");
++  }
++
+   int num = GetStringCount(tensor);
+   data->reserve(num);
+   for (int i = 0; i < num; i++) {
+-- 
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0007-do-not-use-absl-any.patch

@@ -1,25 +1,25 @@
-From 670dfffa386fd0ff28e66cfe1238af43b4e587ce Mon Sep 17 00:00:00 2001
-From: Daniel Rubery <drubery@chromium.org>
-Date: Thu, 6 May 2021 11:22:13 -0700
-Subject: [PATCH] Remove use of banned absl::any
+From d3d4385132632282fc91c735875ebfc90697b067 Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Thu, 16 Dec 2021 13:28:16 -0800
+Subject: [PATCH 07/11] do not use absl any
 
 ---
  .../cc/task/vision/core/frame_buffer.h        | 27 -------------------
  1 file changed, 27 deletions(-)
 
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-index 2bea92883c4d..1556b7dfabef 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
+index c1289673cb82b..1668447393e9e 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/vision/core/frame_buffer.h
 @@ -27,7 +27,6 @@ limitations under the License.
- #include "absl/strings/str_cat.h"
- #include "absl/time/clock.h"
- #include "absl/time/time.h"
--#include "absl/types/any.h"
- #include "absl/types/optional.h"
+ #include "absl/strings/str_cat.h"  // from @com_google_absl
+ #include "absl/time/clock.h"  // from @com_google_absl
+ #include "absl/time/time.h"  // from @com_google_absl
+-#include "absl/types/any.h"  // from @com_google_absl
+ #include "absl/types/optional.h"  // from @com_google_absl
  #include "tensorflow_lite_support/cc/port/integral_types.h"
  #include "tensorflow_lite_support/cc/port/statusor.h"
-@@ -253,31 +252,6 @@ class FrameBuffer {
+@@ -250,31 +249,6 @@ class FrameBuffer {
      return {};
    }
  
@@ -51,7 +51,7 @@ index 2bea92883c4d..1556b7dfabef 100644
    // Returns FrameBuffer dimension.
    const Dimension dimension() const { return dimension_; }
  
-@@ -292,7 +266,6 @@ class FrameBuffer {
+@@ -289,7 +263,6 @@ class FrameBuffer {
  
   private:
    std::vector<Plane> planes_;
@@ -60,5 +60,5 @@ index 2bea92883c4d..1556b7dfabef 100644
    Format format_;
    Orientation orientation_;
 -- 
-2.31.1.607.g51e8a6a459-goog
+2.34.1.307.g9b7440fafd-goog
 

third_party/tflite_support/patches/0008-unused-string-include.patch

@@ -0,0 +1,25 @@
+From 5feffc2cdd8c970490fadd812401be4eb57174d5 Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Thu, 16 Dec 2021 13:43:57 -0800
+Subject: [PATCH 08/11] unused string include
+
+---
+ .../cc/text/tokenizers/tokenizer_jni_lib.h                      | 2 --
+ 1 file changed, 2 deletions(-)
+
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_jni_lib.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_jni_lib.h
+index fc7285c6807b0..33677d305a853 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_jni_lib.h
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/text/tokenizers/tokenizer_jni_lib.h
+@@ -17,8 +17,6 @@ limitations under the License.
+ 
+ #include <jni.h>
+ 
+-#include <string>
+-
+ #include "tensorflow_lite_support/cc/text/tokenizers/tokenizer.h"
+ #include "tensorflow_lite_support/cc/utils/jni_utils.h"
+ 
+-- 
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0009-remove-unbuilt-files-and-change-exec-bit-where-neede.patch

@@ -0,0 +1,666 @@
+From 515f1ef8496e5c73318aa41f6295bbfbefb6bbae Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Thu, 16 Dec 2021 13:56:14 -0800
+Subject: [PATCH 09/11] remove unbuilt files and change exec bit where needed
+
+---
+ .../cc/port/benchmark.h                       |  21 ---
+ .../cc/port/default/status_matchers.h         |  55 -------
+ .../tensorflow_lite_support/cc/port/gmock.h   |  21 ---
+ .../tensorflow_lite_support/cc/port/gtest.h   |  21 ---
+ .../tensorflow_lite_support/cc/port/proto2.h  |  32 ----
+ .../examples/task/audio/desktop/python/BUILD  |   0
+ .../task/audio/desktop/python/README.md       |   0
+ .../desktop/python/audio_classifier_demo.py   |   0
+ .../examples/task/vision/desktop/python/BUILD |   0
+ .../desktop/python/image_classifier_demo.py   |   0
+ .../desktop/python/image_segmenter_demo.py    |   0
+ .../desktop/python/object_detector_demo.py    |   0
+ .../ios/utils/Sources/TFLStringUtil.mm        |  23 ---
+ .../metadata/cc/metadata_populator.cc         | 150 ------------------
+ .../metadata/cc/utils/zip_mem_file.cc         | 124 ---------------
+ .../metadata/cc/utils/zip_mem_file.h          |  71 ---------
+ .../odml/ios/image/resources/grace_hopper.jpg | Bin
+ .../tools/ci_build/build_all.sh               |   0
+ .../ci_build/builds/build_ios_framework.sh    |   0
+ .../tools/ci_build/builds/pip_smoke_test.sh   |   0
+ .../tools/ci_build/common.sh                  |   0
+ .../tools/ci_build/common_win.bat             |   0
+ 22 files changed, 518 deletions(-)
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/cc/port/benchmark.h
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/status_matchers.h
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/cc/port/gmock.h
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/cc/port/gtest.h
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/cc/port/proto2.h
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/BUILD
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/README.md
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/audio_classifier_demo.py
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/BUILD
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/image_classifier_demo.py
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/image_segmenter_demo.py
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/object_detector_demo.py
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/metadata_populator.cc
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.cc
+ delete mode 100644 third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.h
+ mode change 100755 => 100644 third_party/tflite_support/src/tensorflow_lite_support/odml/ios/image/resources/grace_hopper.jpg
+ mode change 100644 => 100755 third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/build_all.sh
+ mode change 100644 => 100755 third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/builds/build_ios_framework.sh
+ mode change 100644 => 100755 third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/builds/pip_smoke_test.sh
+ mode change 100644 => 100755 third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/common.sh
+ mode change 100644 => 100755 third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/common_win.bat
+
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/benchmark.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/port/benchmark.h
+deleted file mode 100644
+index 74bc1a6857664..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/benchmark.h
++++ /dev/null
+@@ -1,21 +0,0 @@
+-/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-
+-#ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_BENCHMARK_H_
+-#define TENSORFLOW_LITE_SUPPORT_CC_PORT_BENCHMARK_H_
+-
+-#include "gtest/benchmark.h"
+-
+-#endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_BENCHMARK_H_
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/status_matchers.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/status_matchers.h
+deleted file mode 100644
+index 6d9668043c183..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/status_matchers.h
++++ /dev/null
+@@ -1,55 +0,0 @@
+-/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-
+-#ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUS_MATCHERS_H_
+-#define TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUS_MATCHERS_H_
+-
+-#include "gmock/gmock.h"
+-#include "gtest/gtest.h"
+-
+-#define SUPPORT_STATUS_MACROS_IMPL_CONCAT_INNER_(x, y) x##y
+-#define SUPPORT_STATUS_MACROS_IMPL_CONCAT_(x, y) \
+-  SUPPORT_STATUS_MACROS_IMPL_CONCAT_INNER_(x, y)
+-
+-#undef SUPPORT_ASSERT_OK
+-#define SUPPORT_ASSERT_OK(expr) \
+-  SUPPORT_ASSERT_OK_IMPL_(      \
+-      SUPPORT_STATUS_MACROS_IMPL_CONCAT_(_status, __LINE__), expr)
+-
+-#define SUPPORT_ASSERT_OK_IMPL_(status, expr) \
+-  auto status = (expr);                       \
+-  ASSERT_TRUE(status.ok());
+-
+-#undef SUPPORT_EXPECT_OK
+-#define SUPPORT_EXPECT_OK(expr) \
+-  SUPPORT_EXPECT_OK_IMPL_(      \
+-      SUPPORT_STATUS_MACROS_IMPL_CONCAT_(_status, __LINE__), expr)
+-
+-#define SUPPORT_EXPECT_OK_IMPL_(status, expr) \
+-  auto status = (expr);                       \
+-  EXPECT_TRUE(status.ok());
+-
+-#undef SUPPORT_ASSERT_OK_AND_ASSIGN
+-#define SUPPORT_ASSERT_OK_AND_ASSIGN(lhs, rexpr)                           \
+-  SUPPORT_ASSERT_OK_AND_ASSIGN_IMPL_(                                      \
+-      SUPPORT_STATUS_MACROS_IMPL_CONCAT_(_status_or_value, __LINE__), lhs, \
+-      rexpr)
+-
+-#define SUPPORT_ASSERT_OK_AND_ASSIGN_IMPL_(statusor, lhs, rexpr) \
+-  auto statusor = (rexpr);                                       \
+-  ASSERT_TRUE(statusor.ok());                                    \
+-  lhs = std::move(statusor.value())
+-
+-#endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUS_MATCHERS_H_
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/gmock.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/port/gmock.h
+deleted file mode 100644
+index 5e4334db323d6..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/gmock.h
++++ /dev/null
+@@ -1,21 +0,0 @@
+-/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-
+-#ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_GMOCK_H_
+-#define TENSORFLOW_LITE_SUPPORT_CC_PORT_GMOCK_H_
+-
+-#include "gmock/gmock.h"
+-
+-#endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_GMOCK_H_
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/gtest.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/port/gtest.h
+deleted file mode 100644
+index dbe2e5e6f9d7c..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/gtest.h
++++ /dev/null
+@@ -1,21 +0,0 @@
+-/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-
+-#ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_GTEST_H_
+-#define TENSORFLOW_LITE_SUPPORT_CC_PORT_GTEST_H_
+-
+-#include "gtest/gtest.h"
+-
+-#endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_GTEST_H_
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/proto2.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/port/proto2.h
+deleted file mode 100644
+index 3cde2ab81d6ee..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/port/proto2.h
++++ /dev/null
+@@ -1,32 +0,0 @@
+-/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-#ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_PROTO_NS_H_
+-#define TENSORFLOW_LITE_SUPPORT_CC_PORT_PROTO_NS_H_
+-
+-#include "google/protobuf/message_lite.h"
+-#include "google/protobuf/text_format.h"
+-
+-namespace tflite {
+-namespace support {
+-namespace proto {
+-
+-using TextFormat = ::google::protobuf::TextFormat;
+-using MessageLite = ::google::protobuf::MessageLite;
+-
+-}  // namespace proto
+-}  // namespace support
+-}  // namespace tflite
+-
+-#endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_PROTO_NS_H_
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/BUILD b/third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/BUILD
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/README.md b/third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/README.md
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/audio_classifier_demo.py b/third_party/tflite_support/src/tensorflow_lite_support/examples/task/audio/desktop/python/audio_classifier_demo.py
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/BUILD b/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/BUILD
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/image_classifier_demo.py b/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/image_classifier_demo.py
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/image_segmenter_demo.py b/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/image_segmenter_demo.py
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/object_detector_demo.py b/third_party/tflite_support/src/tensorflow_lite_support/examples/task/vision/desktop/python/object_detector_demo.py
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm b/third_party/tflite_support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm
+deleted file mode 100644
+index 6e9cf23802427..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.mm
++++ /dev/null
+@@ -1,23 +0,0 @@
+-/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-#import "third_party/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.h"
+-
+-std::string MakeString(NSString* str) { return std::string([str UTF8String]); }
+-
+-NSString* MakeNSString(const std::string& str) {
+-  return [[NSString alloc] initWithBytes:const_cast<void*>(static_cast<const void*>(str.data()))
+-                                  length:str.length()
+-                                encoding:NSUTF8StringEncoding];
+-}
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/metadata_populator.cc b/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/metadata_populator.cc
+deleted file mode 100644
+index e21d426369e2e..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/metadata_populator.cc
++++ /dev/null
+@@ -1,150 +0,0 @@
+-/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-
+-#include "tensorflow_lite_support/metadata/cc/metadata_populator.h"
+-
+-#include <cstdlib>
+-#include <cstring>
+-#include <functional>
+-
+-#include "flatbuffers/flatbuffers.h"  // from @flatbuffers
+-#include "contrib/minizip/ioapi.h"
+-#include "contrib/minizip/zip.h"
+-#include "tensorflow/lite/schema/schema_generated.h"
+-#include "tensorflow_lite_support/cc/common.h"
+-#include "tensorflow_lite_support/cc/port/status_macros.h"
+-#include "tensorflow_lite_support/cc/port/statusor.h"
+-#include "tensorflow_lite_support/metadata/cc/utils/zip_mem_file.h"
+-#include "tensorflow_lite_support/metadata/metadata_schema_generated.h"
+-
+-namespace tflite {
+-namespace metadata {
+-
+-namespace {
+-constexpr char kMetadataBufferName[] = "TFLITE_METADATA";
+-
+-using ::absl::StatusCode;
+-using ::tflite::support::CreateStatusWithPayload;
+-using ::tflite::support::TfLiteSupportStatus;
+-
+-}  // namespace
+-
+-ModelMetadataPopulator::ModelMetadataPopulator(const tflite::Model& model) {
+-  model.UnPackTo(&model_t_);
+-}
+-
+-/* static */
+-tflite::support::StatusOr<std::unique_ptr<ModelMetadataPopulator>>
+-ModelMetadataPopulator::CreateFromModelBuffer(const char* buffer_data,
+-                                              size_t buffer_size) {
+-  // Rely on the simplest, base flatbuffers verifier. Here is not the place to
+-  // e.g. use an OpResolver: we just want to make sure the buffer is valid to
+-  // access the metadata.
+-  flatbuffers::Verifier verifier = flatbuffers::Verifier(
+-      reinterpret_cast<const uint8_t*>(buffer_data), buffer_size);
+-  if (!tflite::VerifyModelBuffer(verifier)) {
+-    return CreateStatusWithPayload(
+-        StatusCode::kInvalidArgument,
+-        "The model is not a valid FlatBuffer buffer.",
+-        TfLiteSupportStatus::kInvalidFlatBufferError);
+-  }
+-  // Use absl::WrapUnique() to call private constructor:
+-  // https://abseil.io/tips/126.
+-  return absl::WrapUnique(
+-      new ModelMetadataPopulator(*tflite::GetModel(buffer_data)));
+-}
+-
+-void ModelMetadataPopulator::LoadMetadata(const char* metadata_buffer_data,
+-                                          size_t metadata_buffer_size) {
+-  // Pack the model metadata in a buffer.
+-  auto model_metadata_buffer = std::make_unique<tflite::BufferT>();
+-  model_metadata_buffer->data = {metadata_buffer_data,
+-                                 metadata_buffer_data + metadata_buffer_size};
+-  // Check if the model already has metadata. If so, just override the buffer
+-  // and exit.
+-  for (const auto& metadata_t : model_t_.metadata) {
+-    if (metadata_t->name == kMetadataBufferName) {
+-      model_t_.buffers[metadata_t->buffer] = std::move(model_metadata_buffer);
+-      return;
+-    }
+-  }
+-  // Model doesn't already have metadata: add metadata buffer and pointer to the
+-  // buffer in the model metadata section.
+-  model_t_.buffers.push_back(std::move(model_metadata_buffer));
+-  auto metadata_t = std::make_unique<tflite::MetadataT>();
+-  metadata_t->name = kMetadataBufferName;
+-  metadata_t->buffer = model_t_.buffers.size() - 1;
+-  model_t_.metadata.push_back(std::move(metadata_t));
+-}
+-
+-void ModelMetadataPopulator::LoadAssociatedFiles(
+-    const absl::flat_hash_map<std::string, std::string>& associated_files) {
+-  associated_files_ = associated_files;
+-}
+-
+-tflite::support::StatusOr<std::string>
+-ModelMetadataPopulator::AppendAssociatedFiles(const char* model_buffer_data,
+-                                              size_t model_buffer_size) {
+-  // Create in-memory zip file.
+-  ZipMemFile mem_file = ZipMemFile(model_buffer_data, model_buffer_size);
+-  // Open zip.
+-  zipFile zf = zipOpen2(/*pathname=*/nullptr, APPEND_STATUS_CREATEAFTER,
+-                        /*globalcomment=*/nullptr, &mem_file.GetFileFuncDef());
+-  if (zf == nullptr) {
+-    return CreateStatusWithPayload(
+-        StatusCode::kUnknown, "Unable to open zip archive",
+-        TfLiteSupportStatus::kMetadataAssociatedFileZipError);
+-  }
+-  // Write associated files.
+-  for (const auto& [name, contents] : associated_files_) {
+-    if ((zipOpenNewFileInZip(zf, name.c_str(),
+-                             /*zipfi=*/nullptr,
+-                             /*extrafield_local=*/nullptr,
+-                             /*size_extrafield_local=*/0,
+-                             /*extrafield_global=*/nullptr,
+-                             /*size_extrafield_global=*/0,
+-                             /*comment=*/nullptr,
+-                             /*method=*/0,
+-                             /*level=*/Z_DEFAULT_COMPRESSION) != ZIP_OK) ||
+-        (zipWriteInFileInZip(zf, contents.data(), contents.length()) !=
+-         ZIP_OK) ||
+-        (zipCloseFileInZip(zf) != ZIP_OK)) {
+-      return CreateStatusWithPayload(
+-          StatusCode::kUnknown, "Unable to write file to zip archive",
+-          TfLiteSupportStatus::kMetadataAssociatedFileZipError);
+-    }
+-  }
+-  // Close zip.
+-  if (zipClose(zf, /*global_comment=*/nullptr) != ZIP_OK) {
+-    return CreateStatusWithPayload(
+-        StatusCode::kUnknown, "Unable to close zip archive",
+-        TfLiteSupportStatus::kMetadataAssociatedFileZipError);
+-  }
+-  // Return as a string.
+-  return std::string(mem_file.GetFileContent());
+-}
+-
+-tflite::support::StatusOr<std::string> ModelMetadataPopulator::Populate() {
+-  // Build model.
+-  flatbuffers::FlatBufferBuilder model_fbb;
+-  model_fbb.Finish(tflite::Model::Pack(model_fbb, &model_t_),
+-                   tflite::ModelIdentifier());
+-  return AppendAssociatedFiles(
+-      reinterpret_cast<char*>(model_fbb.GetBufferPointer()),
+-      model_fbb.GetSize());
+-}
+-
+-}  // namespace metadata
+-}  // namespace tflite
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.cc b/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.cc
+deleted file mode 100644
+index 2e4d9107c8c31..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.cc
++++ /dev/null
+@@ -1,124 +0,0 @@
+-/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-
+-#include "tensorflow_lite_support/metadata/cc/utils/zip_mem_file.h"
+-
+-#include <algorithm>
+-#include <cstdio>
+-
+-#include "absl/strings/string_view.h"  // from @com_google_absl
+-#include "contrib/minizip/ioapi.h"
+-
+-namespace tflite {
+-namespace metadata {
+-
+-ZipMemFile::ZipMemFile(const char* buffer, size_t size)
+-    : data_(buffer, size), offset_(0) {
+-  zlib_filefunc_def_.zopen_file = OpenFile;
+-  zlib_filefunc_def_.zread_file = ReadFile;
+-  zlib_filefunc_def_.zwrite_file = WriteFile;
+-  zlib_filefunc_def_.ztell_file = TellFile;
+-  zlib_filefunc_def_.zseek_file = SeekFile;
+-  zlib_filefunc_def_.zclose_file = CloseFile;
+-  zlib_filefunc_def_.zerror_file = ErrorFile;
+-  zlib_filefunc_def_.opaque = this;
+-}
+-
+-zlib_filefunc_def& ZipMemFile::GetFileFuncDef() { return zlib_filefunc_def_; }
+-
+-absl::string_view ZipMemFile::GetFileContent() const { return data_; }
+-
+-/* static */
+-voidpf ZipMemFile::OpenFile(voidpf opaque, const char* filename, int mode) {
+-  // Result is never used, but needs to be non-null for `zipOpen2` not to fail.
+-  return opaque;
+-}
+-
+-/* static */
+-size_t ZipMemFile::ReadFile(voidpf opaque, voidpf stream, void* buf,
+-                            size_t size) {
+-  auto* mem_file = static_cast<ZipMemFile*>(opaque);
+-  if (mem_file->offset_ < 0 || mem_file->Size() < mem_file->offset_) {
+-    return 0;
+-  }
+-  if (mem_file->offset_ + size > mem_file->Size()) {
+-    size = mem_file->Size() - mem_file->offset_;
+-  }
+-  memcpy(buf,
+-         static_cast<const char*>(mem_file->data_.c_str()) + mem_file->offset_,
+-         size);
+-  mem_file->offset_ += size;
+-  return size;
+-}
+-
+-/* static */
+-size_t ZipMemFile::WriteFile(voidpf opaque, voidpf stream, const void* buf,
+-                             size_t size) {
+-  auto* mem_file = static_cast<ZipMemFile*>(opaque);
+-  if (mem_file->offset_ + size > mem_file->Size()) {
+-    mem_file->data_.resize(mem_file->offset_ + size);
+-  }
+-  mem_file->data_.replace(mem_file->offset_, size,
+-                          static_cast<const char*>(buf), size);
+-  mem_file->offset_ += size;
+-  return size;
+-}
+-
+-/* static */
+-ptrdiff_t ZipMemFile::TellFile(voidpf opaque, voidpf stream) {
+-  return static_cast<ZipMemFile*>(opaque)->offset_;
+-}
+-
+-/* static */
+-ptrdiff_t ZipMemFile::SeekFile(voidpf opaque, voidpf stream, size_t offset,
+-                               int origin) {
+-  auto* mem_file = static_cast<ZipMemFile*>(opaque);
+-  switch (origin) {
+-    case SEEK_SET:
+-      mem_file->offset_ = offset;
+-      return 0;
+-    case SEEK_CUR:
+-      if (mem_file->offset_ + offset < 0 ||
+-          mem_file->offset_ + offset > mem_file->Size()) {
+-        return -1;
+-      }
+-      mem_file->offset_ += offset;
+-      return 0;
+-    case SEEK_END:
+-      if (mem_file->Size() - offset < 0 ||
+-          mem_file->Size() - offset > mem_file->Size()) {
+-        return -1;
+-      }
+-      mem_file->offset_ = offset + mem_file->Size();
+-      return 0;
+-    default:
+-      return -1;
+-  }
+-}
+-
+-/* static */
+-int ZipMemFile::CloseFile(voidpf opaque, voidpf stream) {
+-  // Nothing to do.
+-  return 0;
+-}
+-
+-/* static */
+-int ZipMemFile::ErrorFile(voidpf opaque, voidpf stream) {
+-  // Unused.
+-  return 0;
+-}
+-
+-}  // namespace metadata
+-}  // namespace tflite
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.h b/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.h
+deleted file mode 100644
+index ef7843d70cff6..0000000000000
+--- a/third_party/tflite_support/src/tensorflow_lite_support/metadata/cc/utils/zip_mem_file.h
++++ /dev/null
+@@ -1,71 +0,0 @@
+-/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+-
+-Licensed under the Apache License, Version 2.0 (the "License");
+-you may not use this file except in compliance with the License.
+-You may obtain a copy of the License at
+-
+-    http://www.apache.org/licenses/LICENSE-2.0
+-
+-Unless required by applicable law or agreed to in writing, software
+-distributed under the License is distributed on an "AS IS" BASIS,
+-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+-See the License for the specific language governing permissions and
+-limitations under the License.
+-==============================================================================*/
+-
+-#ifndef TENSORFLOW_LITE_SUPPORT_METADATA_CC_UTILS_ZIP_MEM_FILE_H_
+-#define TENSORFLOW_LITE_SUPPORT_METADATA_CC_UTILS_ZIP_MEM_FILE_H_
+-
+-#include <cstdlib>
+-
+-#include "absl/strings/string_view.h"  // from @com_google_absl
+-#include "contrib/minizip/ioapi.h"
+-
+-namespace tflite {
+-namespace metadata {
+-
+-// In-memory zip file implementation.
+-//
+-// Adapted from [1], with a few key differences:
+-// * backed by an `std::string` instead of malloc-ed C buffers,
+-// * supports opening the file for writing through `zipOpen2`.
+-//
+-// [1]:
+-// https://github.com/google/libkml/blob/master/third_party/zlib-1.2.3/contrib/minizip/iomem_simple.c
+-class ZipMemFile {
+- public:
+-  // Constructs an in-memory zip file from a buffer.
+-  ZipMemFile(const char* buffer, size_t size);
+-  // Provides access to the `zlib_filefunc_def` implementation for the in-memory
+-  // zip file.
+-  zlib_filefunc_def& GetFileFuncDef();
+-  // Provides access to the file contents.
+-  absl::string_view GetFileContent() const;
+-
+- private:
+-  // The string backing the in-memory file.
+-  std::string data_;
+-  // The current offset in the file.
+-  size_t offset_;
+-  // The `zlib_filefunc_def` implementation for this in-memory zip file.
+-  zlib_filefunc_def zlib_filefunc_def_;
+-
+-  // Convenience function to access the current data size.
+-  size_t Size() const { return data_.size(); }
+-
+-  // The file function implementations used in the `zlib_filefunc_def`.
+-  static voidpf OpenFile(voidpf opaque, const char* filename, int mode);
+-  static size_t ReadFile(voidpf opaque, voidpf stream, void* buf, size_t size);
+-  static size_t WriteFile(voidpf opaque, voidpf stream, const void* buf,
+-                          size_t size);
+-  static ptrdiff_t TellFile(voidpf opaque, voidpf stream);
+-  static ptrdiff_t SeekFile(voidpf opaque, voidpf stream, size_t offset,
+-                            int origin);
+-  static int CloseFile(voidpf opaque, voidpf stream);
+-  static int ErrorFile(voidpf opaque, voidpf stream);
+-};
+-
+-}  // namespace metadata
+-}  // namespace tflite
+-
+-#endif  // TENSORFLOW_LITE_SUPPORT_METADATA_CC_UTILS_ZIP_MEM_FILE_H_
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/odml/ios/image/resources/grace_hopper.jpg b/third_party/tflite_support/src/tensorflow_lite_support/odml/ios/image/resources/grace_hopper.jpg
+old mode 100755
+new mode 100644
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/build_all.sh b/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/build_all.sh
+old mode 100644
+new mode 100755
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/builds/build_ios_framework.sh b/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/builds/build_ios_framework.sh
+old mode 100644
+new mode 100755
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/builds/pip_smoke_test.sh b/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/builds/pip_smoke_test.sh
+old mode 100644
+new mode 100755
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/common.sh b/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/common.sh
+old mode 100644
+new mode 100755
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/common_win.bat b/third_party/tflite_support/src/tensorflow_lite_support/tools/ci_build/common_win.bat
+old mode 100644
+new mode 100755
+-- 
+2.34.1.307.g9b7440fafd-goog
+

third_party/tflite_support/patches/0010-only-support-model-file-passed-in-from-mem.patch

@@ -1,22 +1,25 @@
-From 8c5a37f7324b4a03f123c6faecedb4abc8eb0066 Mon Sep 17 00:00:00 2001
-From: mcrouse <mcrouse@google.com>
-Date: Fri, 5 Feb 2021 15:30:25 +0000
-Subject: [PATCH] remove unsupported memory map from file handler
+From 4b7e971e2f2f6ef3fd394858d975b64479047872 Mon Sep 17 00:00:00 2001
+From: Robert Ogden <robertogden@chromium.org>
+Date: Mon, 20 Dec 2021 08:50:35 -0800
+Subject: [PATCH 10/11] only support model file passed in from mem
 
 ---
- .../cc/task/core/external_file_handler.cc     | 126 +-----------------
- .../cc/task/core/external_file_handler.h      |   7 -
+ .../cc/task/core/external_file_handler.cc     | 143 ++----------------
+ .../cc/task/core/external_file_handler.h      |  23 +--
  .../cc/task/core/tflite_engine.cc             |   2 -
  .../cc/task/core/tflite_engine.h              |   2 -
- 4 files changed, 6 insertions(+), 131 deletions(-)
+ 4 files changed, 10 insertions(+), 160 deletions(-)
 
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc
-index ee689e41c6e5..55b662f0926f 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc
-@@ -18,9 +18,6 @@ limitations under the License.
- #include <errno.h>
- #include <fcntl.h>
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc
+index dcde0c926c653..e91a54fb7d11a 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.cc
+@@ -15,45 +15,25 @@ limitations under the License.
+ 
+ #include "tensorflow_lite_support/cc/task/core/external_file_handler.h"
+ 
+-#include <errno.h>
+-#include <fcntl.h>
  #include <stddef.h>
 -#include <sys/mman.h>
 -#include <unistd.h>
@@ -24,7 +27,21 @@ index ee689e41c6e5..55b662f0926f 100644
  #include <memory>
  #include <string>
  
-@@ -40,18 +37,6 @@ using ::tflite::support::CreateStatusWithPayload;
+-#include "absl/memory/memory.h"  // from @com_google_absl
++#include "absl/memory/memory.h"       // from @com_google_absl
+ #include "absl/strings/str_format.h"  // from @com_google_absl
+ #include "tensorflow_lite_support/cc/common.h"
+-#include "tensorflow_lite_support/cc/port/statusor.h"
+ #include "tensorflow_lite_support/cc/port/status_macros.h"
++#include "tensorflow_lite_support/cc/port/statusor.h"
+ 
+ namespace tflite {
+ namespace task {
+ namespace core {
+-namespace {
+ 
+ using ::absl::StatusCode;
+ using ::tflite::support::CreateStatusWithPayload;
  using ::tflite::support::StatusOr;
  using ::tflite::support::TfLiteSupportStatus;
  
@@ -40,10 +57,12 @@ index ee689e41c6e5..55b662f0926f 100644
 -  return aligned_offset;
 -}
 -
- }  // namespace
- 
+-}  // namespace
+-
  /* static */
-@@ -71,103 +56,11 @@ absl::Status ExternalFileHandler::MapExternalFile() {
+ StatusOr<std::unique_ptr<ExternalFileHandler>>
+ ExternalFileHandler::CreateFromExternalFile(const ExternalFile* external_file) {
+@@ -71,123 +51,18 @@ absl::Status ExternalFileHandler::MapExternalFile() {
    if (!external_file_.file_content().empty()) {
      return absl::OkStatus();
    }
@@ -144,16 +163,22 @@ index ee689e41c6e5..55b662f0926f 100644
 -        TfLiteSupportStatus::kFileMmapError);
 -  }
 -  return absl::OkStatus();
-+  return CreateStatusWithPayload(
-+      StatusCode::kInvalidArgument,
-+      "ExternalFile must have 'file_content' set, loading from"
-+      "'file_name' is not supported.",
-+      TfLiteSupportStatus::kInvalidArgumentError);
++
++  return CreateStatusWithPayload(StatusCode::kInvalidArgument,
++                                 "ExternalFile must specify 'file_content' "
++                                 "to be compatible with Chromium.",
++                                 TfLiteSupportStatus::kInvalidArgumentError);
  }
  
  absl::string_view ExternalFileHandler::GetFileContent() {
-@@ -180,14 +73,7 @@ absl::string_view ExternalFileHandler::GetFileContent() {
-   }
+-  if (!external_file_.file_content().empty()) {
+-    return external_file_.file_content();
+-  } else {
+-    return absl::string_view(static_cast<const char*>(buffer_) +
+-                                 buffer_offset_ - buffer_aligned_offset_,
+-                             buffer_size_);
+-  }
++  return external_file_.file_content();
  }
  
 -ExternalFileHandler::~ExternalFileHandler() {
@@ -168,48 +193,64 @@ index ee689e41c6e5..55b662f0926f 100644
  
  }  // namespace core
  }  // namespace task
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h
-index 236d90347698..ad292dcc3702 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h
-@@ -65,10 +65,6 @@ class ExternalFileHandler {
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h
+index cf0bdf0b48037..48c62813e212e 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/external_file_handler.h
+@@ -18,7 +18,7 @@ limitations under the License.
+ 
+ #include <memory>
+ 
+-#include "absl/status/status.h"  // from @com_google_absl
++#include "absl/status/status.h"        // from @com_google_absl
+ #include "absl/strings/string_view.h"  // from @com_google_absl
+ #include "tensorflow_lite_support/cc/port/integral_types.h"
+ #include "tensorflow_lite_support/cc/port/statusor.h"
+@@ -64,27 +64,6 @@ class ExternalFileHandler {
+ 
    // Reference to the input ExternalFile.
    const ExternalFile& external_file_;
- 
+-
 -  // The file descriptor of the ExternalFile if provided by path, as it is
 -  // opened and owned by this class. Set to -1 otherwise.
 -  int owned_fd_{-1};
 -
-   // Points to the memory buffer mapped from the file descriptor of the
-   // ExternalFile, if provided by path or file descriptor.
-   void* buffer_{};
-@@ -82,9 +78,6 @@ class ExternalFileHandler {
- 
-   // The aligned mapped memory buffer offset, if any.
-   int64 buffer_aligned_offset_{};
+-  // Points to the memory buffer mapped from the file descriptor of the
+-  // ExternalFile, if provided by path or file descriptor.
+-  void* buffer_{};
+-
+-  // The mapped memory buffer offset, if any.
+-  int64 buffer_offset_{};
+-  // The size in bytes of the mapped memory buffer, if any.
+-  int64 buffer_size_{};
+-
+-  // As mmap(2) requires the offset to be a multiple of sysconf(_SC_PAGE_SIZE):
+-
+-  // The aligned mapped memory buffer offset, if any.
+-  int64 buffer_aligned_offset_{};
 -  // The aligned mapped memory buffer size in bytes taking into account the
 -  // offset shift introduced by buffer_aligned_memory_offset_, if any.
 -  int64 buffer_aligned_size_{};
  };
  
  }  // namespace core
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc
-index 0317d5f8ea34..6230e5c645c0 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc
+index 8cd4585161df7..484b9a099ecdc 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.cc
 @@ -15,8 +15,6 @@ limitations under the License.
  
  #include "tensorflow_lite_support/cc/task/core/tflite_engine.h"
  
 -#include <unistd.h>
 -
- #include "absl/strings/match.h"
- #include "absl/strings/str_cat.h"
- #include "tensorflow/lite/builtin_ops.h"
-diff --git a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h
-index 6a7e97dd264e..bc55f6b0fe72 100644
---- a/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h
-+++ b/third_party/tflite-support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h
+ #include <memory>
+ 
+ #include "absl/strings/match.h"  // from @com_google_absl
+diff --git a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h
+index 9b44c6e5c022a..53dabdc4841d7 100644
+--- a/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h
++++ b/third_party/tflite_support/src/tensorflow_lite_support/cc/task/core/tflite_engine.h
 @@ -16,8 +16,6 @@ limitations under the License.
  #ifndef TENSORFLOW_LITE_SUPPORT_CC_TASK_CORE_TFLITE_ENGINE_H_
  #define TENSORFLOW_LITE_SUPPORT_CC_TASK_CORE_TFLITE_ENGINE_H_
@@ -218,7 +259,7 @@ index 6a7e97dd264e..bc55f6b0fe72 100644
 -
  #include <memory>
  
- #include "absl/memory/memory.h"
+ #include "absl/memory/memory.h"  // from @com_google_absl
 -- 
-2.30.0.365.g02bc693789-goog
+2.34.1.307.g9b7440fafd-goog
 

third_party/tflite_support/src/.bazelrc

@@ -0,0 +1,170 @@
+# This file is based on tensorflow's (v2.2.0) .bazelrc found here:
+# https://github.com/tensorflow/tensorflow/blob/v2.2.0/.bazelrc
+
+# Sets the default Apple platform to macOS.
+build:macos --apple_platform_type=macos
+
+# Flag to enable remote config. Required starting from TF 2.2.
+common --experimental_repo_remote_exec
+
+# For workaround https://github.com/bazelbuild/bazel/issues/8772 with Bazel >= 0.29.1
+build --java_toolchain=//third_party/toolchains/java:tf_java_toolchain
+build --host_java_toolchain=//third_party/toolchains/java:tf_java_toolchain
+
+# Suppress C++ compiler warnings, otherwise build logs become 10s of MBs.
+build:android --copt=-w
+build:linux --copt=-w
+build:macos --copt=-w
+build:windows --copt=/w
+
+# Android workspace configurations. Should be replaced by an interative configure in the future.
+build --action_env ANDROID_NDK_HOME
+build --action_env ANDROID_NDK_API_LEVEL
+build --action_env ANDROID_BUILD_TOOLS_VERSION
+build --action_env ANDROID_SDK_API_LEVEL
+build --action_env ANDROID_SDK_HOME
+
+# Android configs. Bazel needs to have --cpu and --fat_apk_cpu both set to the
+# target CPU to build transient dependencies correctly. See
+# https://docs.bazel.build/versions/master/user-manual.html#flag--fat_apk_cpu
+
+build:android --crosstool_top=//external:android/crosstool
+build:android --host_crosstool_top=@bazel_tools//tools/cpp:toolchain
+build:android_arm --config=android
+build:android_arm --cpu=armeabi-v7a
+build:android_arm --fat_apk_cpu=armeabi-v7a
+build:android_arm64 --config=android
+build:android_arm64 --cpu=arm64-v8a
+build:android_arm64 --fat_apk_cpu=arm64-v8a
+build:android_x86 --config=android
+build:android_x86 --cpu=x86
+build:android_x86 --fat_apk_cpu=x86
+build:android_x86_64 --config=android
+build:android_x86_64 --cpu=x86_64
+build:android_x86_64 --fat_apk_cpu=x86_64
+
+# iOS configs for each architecture and the fat binary builds.
+build:ios --apple_platform_type=ios
+build:ios --apple_bitcode=embedded --copt=-fembed-bitcode
+build:ios --copt=-Wno-c++11-narrowing
+build:ios_armv7 --config=ios
+build:ios_armv7 --cpu=ios_armv7
+build:ios_arm64 --config=ios
+build:ios_arm64 --cpu=ios_arm64
+build:ios_x86_64 --config=ios
+build:ios_x86_64 --cpu=ios_x86_64
+build:ios_fat --config=ios
+build:ios_fat --ios_multi_cpus=armv7,arm64,x86_64
+
+# TFLite build configs for generic embedded Linux
+build:elinux --crosstool_top=@local_config_embedded_arm//:toolchain
+build:elinux --host_crosstool_top=@bazel_tools//tools/cpp:toolchain
+build:elinux_aarch64 --config=elinux
+build:elinux_aarch64 --cpu=aarch64
+build:elinux_aarch64 --distinct_host_configuration=true
+build:elinux_armhf --config=elinux
+build:elinux_armhf --cpu=armhf
+build:elinux_armhf --distinct_host_configuration=true
+
+# By default, build TF in C++ 14 mode.
+build:android --cxxopt=-std=c++14
+build:android --host_cxxopt=-std=c++14
+build:ios --cxxopt=-std=c++14
+build:ios --host_cxxopt=-std=c++14
+build:linux --cxxopt=-std=c++14
+build:linux --host_cxxopt=-std=c++14
+build:macos --cxxopt=-std=c++14
+build:macos --host_cxxopt=-std=c++14
+build:windows --cxxopt=/std:c++14
+build:windows --host_cxxopt=/std:c++14
+
+# Config to use a mostly-static build and disable modular op registration
+# support (this will revert to loading TensorFlow with RTLD_GLOBAL in Python).
+# By default, TensorFlow will build with a dependence on
+# //tensorflow:libtensorflow_framework.so.
+build:monolithic --define framework_shared_object=false
+
+# For projects which use TensorFlow as part of a Bazel build process, putting
+# nothing in a bazelrc will default to a monolithic build. The following line
+# opts in to modular op registration support by default.
+build --define framework_shared_object=true
+
+# ASAN build
+build:asan --strip=never
+build:asan --copt -fsanitize=address
+build:asan --copt -DADDRESS_SANITIZER
+build:asan --copt -O1
+build:asan --copt -g
+build:asan --copt -fno-omit-frame-pointer
+build:asan --linkopt -fsanitize=address
+
+# dbg config, as a shorthand for '--config=opt -c dbg'
+build:dbg --config=opt -c dbg
+# for now, disable arm_neon. see: https://github.com/tensorflow/tensorflow/issues/33360
+build:dbg --cxxopt -DTF_LITE_DISABLE_X86_NEON
+# AWS SDK must be compiled in release mode. see: https://github.com/tensorflow/tensorflow/issues/37498
+build:dbg --copt -DDEBUG_BUILD
+
+build --define=use_fast_cpp_protos=true
+build --define=allow_oversize_protos=true
+
+# TF uses `standalone`, which is deprecated.
+build --spawn_strategy=local
+build -c opt
+
+# Adding "--cxxopt=-D_GLIBCXX_USE_CXX11_ABI=0" creates parity with TF
+# compilation options. It also addresses memory use due to
+# copy-on-write semantics of std::strings of the older ABI.
+build --cxxopt=-D_GLIBCXX_USE_CXX11_ABI=0
+
+# Make Bazel print out all options from rc files.
+build --announce_rc
+
+# Other build flags.
+build --define=grpc_no_ares=true
+
+# See https://github.com/bazelbuild/bazel/issues/7362 for information on what
+# --incompatible_remove_legacy_whole_archive flag does.
+# This flag is set to true in Bazel 1.0 and newer versions. We tried to migrate
+# Tensorflow to the default, however test coverage wasn't enough to catch the
+# errors.
+# There is ongoing work on Bazel team's side to provide support for transitive
+# shared libraries. As part of migrating to transitive shared libraries, we
+# hope to provide a better mechanism for control over symbol exporting, and
+# then tackle this issue again.
+#
+# TODO: Remove this line once TF doesn't depend on Bazel wrapping all library
+# archives in -whole_archive -no_whole_archive.
+build --noincompatible_remove_legacy_whole_archive
+
+# These are bazel 2.0's incompatible flags. Tensorflow needs to use bazel 2.0.0
+# to use cc_shared_library, as part of the Tensorflow Build Improvements RFC:
+# https://github.com/tensorflow/community/pull/179
+build --noincompatible_prohibit_aapt1
+
+# Build TF with C++ 17 features.
+build:c++17 --cxxopt=-std=c++1z
+build:c++17 --cxxopt=-stdlib=libc++
+build:c++1z --config=c++17
+
+# Enable using platform specific build settings, except when cross-compiling for
+# mobile platforms.
+build --enable_platform_specific_config
+build:android --noenable_platform_specific_config
+build:ios --noenable_platform_specific_config
+
+# Suppress all warning messages.
+build:short_logs --output_filter=DONT_MATCH_ANYTHING
+build:verbose_logs --output_filter=
+build --config=short_logs
+
+# Options to build TensorFlow 1.x or 2.x.
+build:v1 --define=tf_api_version=1
+build:v2 --define=tf_api_version=2
+build:v1 --action_env=TF2_BEHAVIOR=0
+build:v2 --action_env=TF2_BEHAVIOR=1
+build --config=v2
+test --config=v2
+
+# Put user-specific options in .bazelrc.user
+try-import %workspace%/.bazelrc.user

third_party/tflite_support/src/.bazelversion

@@ -0,0 +1 @@
+3.7.2

third_party/tflite_support/src/README.md

@@ -5,8 +5,8 @@ models onto mobile devices. It works cross-Platform and is supported on Java,
 C++ (WIP), and Swift (WIP). The TFLite Support project consists of the following
 major components:
 
-*   **TFLite Support Library**: a cross-platform library that helps to
-    deploy TFLite models onto mobile devices.
+*   **TFLite Support Library**: a cross-platform library that helps to deploy
+    TFLite models onto mobile devices.
 *   **TFLite Model Metadata**: (metadata populator and metadata extractor
     library): includes both human and machine readable information about what a
     model does and how to use the model.
@@ -55,6 +55,11 @@ Utils, you need to set up following env variables correctly:
 *   `ANDROID_SDK_API_LEVEL`
 *   `ANDROID_BUILD_TOOLS_VERSION`
 
+## How to contribute
+
+Please issue a pull request and assign @xunkai55 or @lu-wang-g for a code
+review.
+
 ## Contact us
 
 Let us know what you think about TFLite Support by creating a

third_party/tflite_support/src/WORKSPACE

@@ -1,9 +1,58 @@
 workspace(name = "org_tensorflow_lite_support")
 
 load("@bazel_tools//tools/build_defs/repo:java.bzl", "java_import_external")
-load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
+load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive", "http_file")
 load("@//third_party/py:python_configure.bzl", "python_configure")
 
+http_file(
+    name = "mobilebert_float",
+    sha256 = "883bf5d40f0b0ae435326bb21ed0f4c9004b22c3fd1539383fd16d68623696dd",
+    urls = ["https://tfhub.dev/tensorflow/lite-model/mobilebert/1/default/1?lite-format=tflite"],
+)
+
+http_file(
+    name = "mobilebert_with_metadata",
+    sha256 = "e79d3c70108bbdee02da657b679349cab46dbb859a05b599c76b53d98e82f272",
+    urls = ["https://tfhub.dev/tensorflow/lite-model/mobilebert/1/metadata/1?lite-format=tflite"],
+)
+
+http_file(
+    name = "30k-clean",
+    sha256 = "fefb02b667a6c5c2fe27602d28e5fb3428f66ab89c7d6f388e7c8d44a02d0336",
+    urls = ["https://storage.googleapis.com/download.tensorflow.org/models/tflite_support/bert_qa/30k-clean.model"],
+)
+
+http_file(
+    name = "mobilebert_vocab",
+    sha256 = "07eced375cec144d27c900241f3e339478dec958f92fddbc551f295c992038a3",
+    urls = ["https://storage.googleapis.com/download.tensorflow.org/models/tflite_support/bert_qa/mobilebert_vocab.txt"],
+)
+
+
+http_file(
+    name = "albert",
+    sha256 = "4a29c7063c518925960229f49dd03e8da5d6682001cf73037815dcd98afd728a",
+    urls = ["https://tfhub.dev/tensorflow/lite-model/albert_lite_base/squadv1/1?lite-format=tflite"],
+)
+
+http_file(
+    name = "albert_with_metadata",
+    sha256 = "8a8a91856b94b945e4a9f22f0332bbf105c3b6b878bb23abfc97eb89d3e8436a",
+    urls = ["https://tfhub.dev/tensorflow/lite-model/albert_lite_base/squadv1/metadata/1?lite-format=tflite"],
+)
+
+http_file(
+    name = "bert_nl_classifier",
+    sha256 = "1e5a550c09bff0a13e61858bcfac7654d7fcc6d42106b4f15e11117695069600",
+    urls = ["https://storage.googleapis.com/download.tensorflow.org/models/tflite_support/bert_nl_classifier/bert_nl_classifier.tflite"],
+)
+
+http_file(
+    name = "bert_nl_classifier_no_metadata",
+    sha256 = "9b4554f6e28a72a3f40511964eed1ccf4e74cc074f81543cacca4faf169a173e",
+    urls = ["https://storage.googleapis.com/download.tensorflow.org/models/tflite_support/bert_nl_classifier/bert_nl_classifier_no_metadata.tflite"],
+)
+
 http_archive(
     name = "io_bazel_rules_closure",
     sha256 = "5b00383d08dd71f28503736db0500b6fb4dda47489ff5fc6bed42557c07c6ba9",
@@ -14,6 +63,14 @@ http_archive(
     ],
 )
 
+# GoogleTest/GoogleMock framework. Used by most unit-tests.
+http_archive(
+    name = "com_google_googletest",
+    urls = ["https://github.com/google/googletest/archive/4ec4cd23f486bf70efcc5d2caa40f24368f752e3.zip"],
+    strip_prefix = "googletest-4ec4cd23f486bf70efcc5d2caa40f24368f752e3",
+    sha256 = "de682ea824bfffba05b4e33b67431c247397d6175962534305136aa06f92e049",
+)
+
 # Apple and Swift rules.
 # https://github.com/bazelbuild/rules_apple/releases
 http_archive(
@@ -37,15 +94,21 @@ http_archive(
     ],
 )
 
-# tf-nightly-20200810
+# TF on 2021-11-09.
+TENSORFLOW_COMMIT = "6a144e7763914d3f6141a7cdc9cb116cc23425f9"
+TENSORFLOW_SHA256 = "cec9a514c09d2b171ad447f3413151b25a6c3d88d048148cced1e85db81f3617"
 http_archive(
     name = "org_tensorflow",
-    sha256 = "fc6d7c57cd9427e695a38ad00fb6ecc3f623bac792dd44ad73a3f85b338b68be",
-    strip_prefix = "tensorflow-8a4ffe2e1ae722cff5306778df0cfca8b7f503fe",
+    sha256 = TENSORFLOW_SHA256,
+    strip_prefix = "tensorflow-" + TENSORFLOW_COMMIT,
     urls = [
-        "https://github.com/tensorflow/tensorflow/archive/8a4ffe2e1ae722cff5306778df0cfca8b7f503fe.tar.gz",
+        "https://github.com/tensorflow/tensorflow/archive/" + TENSORFLOW_COMMIT
+        + ".tar.gz",
+    ],
+    patches = [
+        # We need to rename lite/ios/BUILD.apple to lite/ios/BUILD.
+        "@//third_party:tensorflow_lite_ios_build.patch",
     ],
-    patches = ["@//third_party:tensorflow_lite_ios_build.patch"],
     patch_args = ["-p1"],
 )
 
@@ -60,11 +123,11 @@ gflags()
 third_party_http_archive(
     name = "pybind11",
     urls = [
-        "https://storage.googleapis.com/mirror.tensorflow.org/github.com/pybind/pybind11/archive/v2.4.3.tar.gz",
-        "https://github.com/pybind/pybind11/archive/v2.4.3.tar.gz",
+        "https://storage.googleapis.com/mirror.tensorflow.org/github.com/pybind/pybind11/archive/v2.6.0.tar.gz",
+        "https://github.com/pybind/pybind11/archive/v2.6.0.tar.gz",
     ],
-    sha256 = "1eed57bc6863190e35637290f97a20c81cfe4d9090ac0a24f3bbf08f265eb71d",
-    strip_prefix = "pybind11-2.4.3",
+    sha256 = "90b705137b69ee3b5fc655eaca66d0dc9862ea1759226f7ccd3098425ae69571",
+    strip_prefix = "pybind11-2.6.0",
     build_file = "//third_party:pybind11.BUILD",
 )
 
@@ -119,13 +182,14 @@ http_archive(
     ],
 )
 
-# ABSL cpp library lts_2020_02_25
-# Needed for absl/status
+# ABSL cpp library lts_2021_03_24 Patch2
+# See https://github.com/abseil/abseil-cpp/releases for details.
+# Needed for absl/status and absl/status:statusor
 http_archive(
     name = "com_google_absl",
     build_file = "//third_party:com_google_absl.BUILD",
     urls = [
-        "https://github.com/abseil/abseil-cpp/archive/20200225.tar.gz",
+        "https://github.com/abseil/abseil-cpp/archive/20210324.2.tar.gz",
     ],
     # Remove after https://github.com/abseil/abseil-cpp/issues/326 is solved.
     patches = [
@@ -134,8 +198,8 @@ http_archive(
     patch_args = [
         "-p1",
     ],
-    strip_prefix = "abseil-cpp-20200225",
-    sha256 = "728a813291bdec2aa46eab8356ace9f75ac2ed9dfe2df5ab603c4e6c09f1c353"
+    strip_prefix = "abseil-cpp-20210324.2",
+    sha256 = "59b862f50e710277f8ede96f083a5bb8d7c9595376146838b9580be90374ee1f"
 )
 
 http_archive(
@@ -175,12 +239,12 @@ http_archive(
 
 http_archive(
     name = "libyuv",
-    urls = ["https://chromium.googlesource.com/libyuv/libyuv/+archive/6d603ec3f57dafddc424ef895e5d903915e94ba6.tar.gz"],
-    # Adding the constrain of sha256 and strip_prefix will cause failure.
-    # It seems that the downloaded libyuv was different every time, so that
-    # the specified sha256 and strip_prefix cannot match.
-    # sha256 = "ce196c72858456baa8022fa4a0dc18b77d619265dbc0e3d58e25ad15ca402522",
-    # strip_prefix = "libyuv-6d603ec3f57dafddc424ef895e5d903915e94ba6",
+    urls = ["https://chromium.googlesource.com/libyuv/libyuv/+archive/39240f7149cffde62e3620344d222c8ab2c21178.tar.gz"],
+    # Adding the constrain of sha256 and strip_prefix will cause failure as of
+    # Jan 2021. It seems that the downloaded libyuv was different every time,
+    # so that the specified sha256 and strip_prefix cannot match.
+    # sha256 = "01c2e30eb8e83880f9ba382f6bece9c38cd5b07f9cadae46ef1d5a69e07fafaf",
+    # strip_prefix = "libyuv-39240f7149cffde62e3620344d222c8ab2c21178",
     build_file = "//third_party:libyuv.BUILD",
 )
 
@@ -243,18 +307,18 @@ http_archive(
 )
 
 http_archive(
-    name = "com_google_protobuf",
-    sha256 = "a79d19dcdf9139fa4b81206e318e33d245c4c9da1ffed21c87288ed4380426f9",
-    strip_prefix = "protobuf-3.11.4",
-    urls = ["https://github.com/protocolbuffers/protobuf/archive/v3.11.4.tar.gz"],
-    patches = [
-        "@//third_party:com_google_protobuf_fixes.diff"
-    ],
-    patch_args = [
-        "-p1",
-    ],
+  name = "libedgetpu",
+  sha256 = "a179016a5874c58db969a5edd3fecf57610604e751b5c4d6d82ad58c383ffd64",
+  strip_prefix = "libedgetpu-ea1eaddbddece0c9ca1166e868f8fd03f4a3199e",
+  urls = [
+    "https://github.com/google-coral/libedgetpu/archive/ea1eaddbddece0c9ca1166e868f8fd03f4a3199e.tar.gz"
+  ],
 )
 
+# Set up TensorFlow version for Coral.
+load("@libedgetpu//:workspace.bzl", "libedgetpu_dependencies")
+libedgetpu_dependencies(TENSORFLOW_COMMIT, TENSORFLOW_SHA256)
+
 # AutoValue 1.6+ shades Guava, Auto Common, and JavaPoet. That's OK
 # because none of these jars become runtime dependencies.
 java_import_external(
@@ -317,12 +381,38 @@ java_import_external(
     default_visibility = ["@com_google_auto_value//:__pkg__"],
 )
 
+http_archive(
+    name = "robolectric",
+    urls = ["https://github.com/robolectric/robolectric-bazel/archive/4.4.tar.gz"],
+    strip_prefix = "robolectric-bazel-4.4",
+)
+load("@robolectric//bazel:robolectric.bzl", "robolectric_repositories")
+robolectric_repositories()
+
 load("//third_party/flatbuffers:workspace.bzl", flatbuffers = "repo")
 
 flatbuffers()
+
+RULES_JVM_EXTERNAL_TAG = "3.2"
+
+http_archive(
+    name = "rules_jvm_external",
+    strip_prefix = "rules_jvm_external-%s" % RULES_JVM_EXTERNAL_TAG,
+    sha256 = "82262ff4223c5fda6fb7ff8bd63db8131b51b413d26eb49e3131037e79e324af",
+    url = "https://github.com/bazelbuild/rules_jvm_external/archive/%s.zip" % RULES_JVM_EXTERNAL_TAG,
+)
+
+load("@rules_jvm_external//:defs.bzl", "maven_install")
+
 # Set up TF.
-load("@org_tensorflow//tensorflow:workspace.bzl", "tf_workspace")
-tf_workspace(tf_repo_name="@org_tensorflow")
+load("@org_tensorflow//tensorflow:workspace3.bzl", "workspace")
+workspace()
+load("@org_tensorflow//tensorflow:workspace2.bzl", "workspace")  # buildifier: disable=load
+workspace()
+load("@org_tensorflow//tensorflow:workspace1.bzl", "workspace")  # buildifier: disable=load
+workspace()
+load("@org_tensorflow//tensorflow:workspace0.bzl", "workspace")  # buildifier: disable=load
+workspace()
 
 load("//third_party/tensorflow:tf_configure.bzl", "tf_configure")
 tf_configure(name = "local_config_tf")
@@ -346,35 +436,38 @@ apple_support_dependencies()
 load("@upb//bazel:repository_defs.bzl", "bazel_version_repository")
 bazel_version_repository(name = "bazel_version")
 
-
-# Set up Android.
-load("//third_party/android:android_configure.bzl", "android_configure")
-android_configure(name="local_config_android")
-load("@local_config_android//:android.bzl", "android_workspace")
-android_workspace()
-
 python_configure(name = "local_config_python")
 
+ATS_TAG = "androidx-test-1.3.0"
+http_archive(
+    name = "android_test_support",
+    strip_prefix = "android-test-%s" % ATS_TAG,
+    urls = ["https://github.com/android/android-test/archive/%s.tar.gz" % ATS_TAG],
+)
+load("@android_test_support//:repo.bzl", "android_test_repositories")
+android_test_repositories()
 
 # Maven dependencies.
 
-RULES_JVM_EXTERNAL_TAG = "3.2"
-
-http_archive(
-    name = "rules_jvm_external",
-    strip_prefix = "rules_jvm_external-%s" % RULES_JVM_EXTERNAL_TAG,
-    sha256 = "82262ff4223c5fda6fb7ff8bd63db8131b51b413d26eb49e3131037e79e324af",
-    url = "https://github.com/bazelbuild/rules_jvm_external/archive/%s.zip" % RULES_JVM_EXTERNAL_TAG,
-)
-
-load("@rules_jvm_external//:defs.bzl", "maven_install")
-
 maven_install(
     artifacts = [
         "androidx.annotation:annotation:aar:1.1.0",
+        "androidx.annotation:annotation-experimental:1.1.0",
+        "androidx.multidex:multidex:jar:2.0.1",
+        "androidx.test:core:jar:1.3.0",
+        "androidx.test.ext:junit:jar:1.1.2",
+        "androidx.test:runner:jar:1.3.0",
+        "com.google.android.odml:image:aar:1.0.0-beta1",
+        "com.google.truth:truth:jar:1.1",
+        "commons-io:commons-io:jar:2.8.0",
+        # Mockito >= 3.4.6 cannot pass bazel desugar.
+        "org.mockito:mockito-android:jar:3.0.0",
+        "org.mockito:mockito-core:jar:3.0.0",
+        "org.mockito:mockito-inline:jar:3.0.0",
+        "org.robolectric:robolectric:jar:4.4",
+        "junit:junit:jar:4.13",
     ],
     repositories = [
-        "https://jcenter.bintray.com",
         "https://maven.google.com",
         "https://dl.google.com/dl/android/maven2",
         "https://repo1.maven.org/maven2",
@@ -382,3 +475,23 @@ maven_install(
     fetch_sources = True,
     version_conflict_policy = "pinned",
 )
+
+http_archive(
+    name = "tf_toolchains",
+    sha256 = "d72b2e52baf0592f5b94347b128ef75422fc22f63dfcf2d5fd46bc732cab052b",
+    strip_prefix = "toolchains-1.3.0",
+    urls = [
+        "http://mirror.tensorflow.org/github.com/tensorflow/toolchains/archive/v1.3.0.tar.gz",
+        "https://github.com/tensorflow/toolchains/archive/v1.3.0.tar.gz",
+    ],
+)
+
+load("@tf_toolchains//toolchains/embedded/arm-linux:arm_linux_toolchain_configure.bzl", "arm_linux_toolchain_configure")
+
+# TFLite crossbuild toolchain for embeddeds Linux
+arm_linux_toolchain_configure(
+    name = "local_config_embedded_arm",
+    build_file = "@tf_toolchains//toolchains/embedded/arm-linux:BUILD",
+    aarch64_repo = "../aarch64_linux_toolchain",
+    armhf_repo = "../armhf_linux_toolchain",
+)

third_party/tflite_support/src/tensorflow_lite_support/BUILD

@@ -8,17 +8,21 @@ package(
 
 exports_files(["LICENSE"])
 
-# LINT.IfChange
 package_group(
     name = "users",
+    includes = [
+        ":internal",
+    ],
     packages = [
-        "//tensorflow_lite_support/...",
-        "//third_party/py/tensorflow_examples/...",
-        "//third_party/tensorflow_models/...",
     ],
 )
-# Remove internal path from tensorflow_lite_support:users in the copybara file.
-# LINT.ThenChange(//tensorflow_lite_support/copy.bara.sky)
+
+package_group(
+    name = "internal",
+    packages = [
+        "//tensorflow_lite_support/...",
+    ],
+)
 
 # Config setting for determining if we are building for Android.
 config_setting(

third_party/tflite_support/src/tensorflow_lite_support/acceleration/README.md

@@ -0,0 +1,18 @@
+# Acceleration allowlisting
+
+A complementary directory for the work of
+[accelerator allowlisting](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/lite/experimental/acceleration)
+in TensorFlow Lite.
+
+## Coral Edge TPU plugin
+
+The Coral Edge TPU delegate plugin used in the
+[acceleration library](https://github.com/tensorflow/tflite-support/blob/master/tensorflow_lite_support/cc/port/default/tflite_wrapper.h).
+See
+[CoralSettings](https://github.com/tensorflow/tensorflow/blob/896fecee319ffeb4af2a3c0b5436f3a55ab058fa/tensorflow/lite/experimental/acceleration/configuration/configuration.proto#L323)
+about how to configure the Coral Edge TPU plugin. You can use the acceleration
+library together with
+[Task Library](https://github.com/tensorflow/tflite-support/tree/master/tensorflow_lite_support/cc/task).
+Configure your desired accelerator, including the Coral plugin through the
+options of each task, i.e.
+[image_classifier_options](https://github.com/tensorflow/tflite-support/blob/43f1267b99f1dbc27c7c5b2e1111e1ff6b9121ea/tensorflow_lite_support/cc/task/vision/proto/image_classifier_options.proto#L79).

third_party/tflite_support/src/tensorflow_lite_support/acceleration/configuration/BUILD

@@ -0,0 +1,136 @@
+load("@org_tensorflow//tensorflow/lite/java:aar_with_jni.bzl", "aar_with_jni")
+load("@build_bazel_rules_android//android:rules.bzl", "android_library")
+load("@org_tensorflow//tensorflow/lite:build_def.bzl", "tflite_jni_binary")
+
+package(
+    default_visibility = [
+        "//tensorflow_lite_support:internal",
+    ],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library(
+    name = "gpu_plugin",
+    visibility = [
+        "//visibility:public",
+    ],
+    deps = [
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:gpu_plugin",
+    ],
+    alwayslink = 1,  # For registration to always run.
+)
+
+cc_library(
+    name = "nnapi_plugin",
+    visibility = [
+        "//visibility:public",
+    ],
+    deps = [
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:nnapi_plugin",
+    ],
+    alwayslink = 1,  # For registration to always run.
+)
+
+cc_library(
+    name = "hexagon_plugin",
+    visibility = [
+        "//visibility:public",
+    ],
+    deps = [
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:hexagon_plugin",
+    ],
+    alwayslink = 1,  # For registration to always run.
+)
+
+cc_library(
+    name = "xnnpack_plugin",
+    visibility = [
+        "//visibility:public",
+    ],
+    deps = [
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:xnnpack_plugin",
+    ],
+    alwayslink = 1,  # For registration to always run.
+)
+
+# To use the edgetpu_coral_plugin externally, add the following flags to the bazel command:
+# --define darwinn_portable=1
+cc_library(
+    name = "edgetpu_coral_plugin",
+    srcs = ["edgetpu_coral_plugin.cc"],
+    visibility = [
+        "//visibility:public",
+    ],
+    deps = [
+        "@com_google_absl//absl/container:node_hash_map",
+        "@com_google_absl//absl/memory",
+        "@com_google_absl//absl/strings",
+        "@com_google_glog//:glog",
+        "@libedgetpu//tflite/public:edgetpu_c",
+        "@libedgetpu//tflite/public:oss_edgetpu_direct_all",  # buildcleaner: keep
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:configuration_fbs",
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:delegate_registry",
+    ],
+    alwayslink = 1,  # For registration to always run.
+)
+
+# To test it externally, plugin a Coral device, and run the following command:
+# bazel test tensorflow_lite_support/acceleration/configuration:edgetpu_coral_plugin_test \
+# --define darwinn_portable=1
+cc_test(
+    name = "edgetpu_coral_plugin_test",
+    srcs = ["edgetpu_coral_plugin_test.cc"],
+    data = [
+        "//tensorflow_lite_support/acceleration/configuration/testdata:test_files",
+    ],
+    tags = [
+        "manual",
+        "notap",  # Requires edge TPU device.
+    ],
+    deps = [
+        ":edgetpu_coral_plugin",
+        "//tensorflow_lite_support/examples/task/vision/desktop/utils:image_utils",
+        "@com_google_googletest//:gtest_main",
+        "@org_tensorflow//tensorflow/lite:framework",
+        "@org_tensorflow//tensorflow/lite/c:common",
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:configuration_fbs",
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:delegate_registry",
+        "@org_tensorflow//tensorflow/lite/kernels:builtin_ops",
+    ],
+)
+
+# Targets for delegate plugin library Maven release.
+
+# GPU delegate plugin library.
+tflite_jni_binary(
+    name = "libgpu_delegate_plugin.so",
+    linkscript = "//tensorflow_lite_support/java:default_version_script.lds",
+    visibility = ["//visibility:private"],
+    deps = [
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration/c:gpu_plugin",
+    ],
+)
+
+cc_library(
+    name = "gpu_delegate_plugin_native",
+    srcs = [
+        ":libgpu_delegate_plugin.so",
+    ],
+    visibility = ["//visibility:private"],
+)
+
+# Android target of Acceleration@Scale GPU plugin.
+# Use this target when GPU delegate is selected in the Task Library Java API.
+android_library(
+    name = "gpu_delegate_plugin_android",
+    visibility = [
+        "//visibility:public",
+    ],
+    exports = [":gpu_delegate_plugin_native"],
+)
+
+# AAR target of Acceleration@Scale GPU acceleration for OSS release.
+aar_with_jni(
+    name = "gpu-delegate-plugin",
+    android_library = ":gpu_delegate_plugin_android",
+)

third_party/tflite_support/src/tensorflow_lite_support/acceleration/configuration/edgetpu_coral_plugin.cc

@@ -0,0 +1,170 @@
+/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#include <memory>
+#include <unordered_map>
+
+#include <glog/logging.h>
+#include "absl/container/node_hash_map.h"  // from @com_google_absl
+#include "absl/memory/memory.h"            // from @com_google_absl
+#include "absl/strings/match.h"            // from @com_google_absl
+#include "absl/strings/numbers.h"          // from @com_google_absl
+#include "tensorflow/lite/experimental/acceleration/configuration/configuration_generated.h"
+#include "tensorflow/lite/experimental/acceleration/configuration/delegate_registry.h"
+#include "tflite/public/edgetpu_c.h"
+
+namespace tflite {
+namespace delegates {
+namespace {
+
+constexpr int kDEFAULT_USB_MAX_BULK_IN_QUEUE_LENGTH = 32;
+constexpr char kUsb[] = "usb";
+constexpr char kPci[] = "pci";
+
+inline std::string ConvertPerformance(
+    const CoralSettings_::Performance& from_performance) {
+  switch (from_performance) {
+    case CoralSettings_::Performance_LOW:
+      return "Low";
+    case CoralSettings_::Performance_MEDIUM:
+      return "Medium";
+    case CoralSettings_::Performance_HIGH:
+      return "High";
+    default:
+      return "Max";
+  }
+}
+
+inline std::string ConvertBool(bool from_bool) {
+  return from_bool ? "True" : "False";
+}
+
+bool MatchDevice(const std::string& device,
+                 const std::string& type,
+                 int* index) {
+  const auto prefix(type + ":");
+  if (!absl::StartsWith(device, prefix))
+    return false;
+  if (!absl::SimpleAtoi(device.substr(prefix.size()), index))
+    return false;
+  if (*index < 0)
+    return false;
+  return true;
+}
+
+// device_index corresponds to specific device type, e.g. "usb:0" means the
+// first USB device or "pci:0" means the first PCIe device.
+TfLiteDelegate* CreateEdgeTpuDelegate(
+    absl::optional<edgetpu_device_type> device_type,
+    absl::optional<int> device_index,
+    const absl::node_hash_map<std::string, std::string>& device_options) {
+  std::vector<edgetpu_option> options(device_options.size());
+  size_t i = 0;
+  for (auto& device_option : device_options) {
+    options[i++] = {device_option.first.c_str(), device_option.second.c_str()};
+  }
+
+  size_t num_devices;
+  std::unique_ptr<edgetpu_device, decltype(&edgetpu_free_devices)> devices(
+      edgetpu_list_devices(&num_devices), &edgetpu_free_devices);
+
+  if (!device_index.has_value()) {
+    return CreateEdgeTpuDelegate(device_type, 0, device_options);
+  } else {
+    const int index = device_index.value();
+    if (device_type.has_value()) {
+      int type_index = 0;
+      for (size_t i = 0; i < num_devices; i++) {
+        const auto& device = devices.get()[i];
+        if (device.type == device_type.value() && type_index++ == index)
+          return edgetpu_create_delegate(device.type, device.path,
+                                         options.data(), options.size());
+      }
+    } else {
+      if (index < num_devices)
+        return edgetpu_create_delegate(devices.get()[index].type,
+                                       devices.get()[index].path,
+                                       options.data(), options.size());
+    }
+    return nullptr;
+  }
+}
+
+TfLiteDelegate* CreateEdgeTpuDelegate(
+    const std::string& device,
+    const absl::node_hash_map<std::string, std::string>& options) {
+  if (device.empty()) {
+    return CreateEdgeTpuDelegate(absl::nullopt, absl::nullopt, options);
+  } else if (device == kUsb) {
+    return CreateEdgeTpuDelegate(EDGETPU_APEX_USB, absl::nullopt, options);
+  } else if (device == kPci) {
+    return CreateEdgeTpuDelegate(EDGETPU_APEX_PCI, absl::nullopt, options);
+  } else {
+    int index;
+    if (MatchDevice(device, "", &index)) {
+      return CreateEdgeTpuDelegate(absl::nullopt, index, options);
+    } else if (MatchDevice(device, kUsb, &index)) {
+      return CreateEdgeTpuDelegate(EDGETPU_APEX_USB, index, options);
+    } else if (MatchDevice(device, kPci, &index)) {
+      return CreateEdgeTpuDelegate(EDGETPU_APEX_PCI, index, options);
+    } else {
+      LOG(ERROR) << "Cannot match the given device string (" << device
+                 << ") with a Coral device.";
+      return nullptr;
+    }
+  }
+}
+
+class EdgeTpuCoralPlugin : public DelegatePluginInterface {
+ public:
+  TfLiteDelegatePtr Create() override {
+    return TfLiteDelegatePtr(CreateEdgeTpuDelegate(device_, options_),
+                             edgetpu_free_delegate);
+  }
+
+  int GetDelegateErrno(TfLiteDelegate* from_delegate) override { return 0; }
+
+  static std::unique_ptr<DelegatePluginInterface> New(
+      const TFLiteSettings& acceleration) {
+    return absl::make_unique<EdgeTpuCoralPlugin>(acceleration);
+  }
+
+  explicit EdgeTpuCoralPlugin(const TFLiteSettings& tflite_settings) {
+    const auto* coral_settings = tflite_settings.coral_settings();
+    if (!coral_settings) {
+      return;
+    }
+
+    device_ = coral_settings->device()->str();
+    options_.insert(
+        {"Performance", ConvertPerformance(coral_settings->performance())});
+    options_.insert(
+        {"Usb.AlwaysDfu", ConvertBool(coral_settings->usb_always_dfu())});
+    options_.insert(
+        {"Usb.MaxBulkInQueueLength",
+         std::to_string(coral_settings->usb_max_bulk_in_queue_length() == 0
+                            ? kDEFAULT_USB_MAX_BULK_IN_QUEUE_LENGTH
+                            : coral_settings->usb_max_bulk_in_queue_length())});
+  }
+
+ private:
+  std::string device_;
+  absl::node_hash_map<std::string, std::string> options_;
+};
+}  // namespace
+
+TFLITE_REGISTER_DELEGATE_FACTORY_FUNCTION(EdgeTpuCoralPlugin,
+                                          EdgeTpuCoralPlugin::New);
+}  // namespace delegates
+}  // namespace tflite

third_party/tflite_support/src/tensorflow_lite_support/acceleration/configuration/edgetpu_coral_plugin_test.cc

@@ -0,0 +1,100 @@
+/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include <gmock/gmock.h>
+#include <gtest/gtest.h>
+#include "tensorflow/lite/c/common.h"
+#include "tensorflow/lite/experimental/acceleration/configuration/configuration_generated.h"
+#include "tensorflow/lite/experimental/acceleration/configuration/delegate_registry.h"
+#include "tensorflow/lite/interpreter.h"
+#include "tensorflow/lite/kernels/register.h"
+#include "tensorflow/lite/model.h"
+#include "tensorflow_lite_support/examples/task/vision/desktop/utils/image_utils.h"
+
+namespace tflite {
+namespace delegates {
+namespace {
+
+constexpr char kEdgeTpuModelFilePath[] =
+    "tensorflow_lite_support/acceleration/configuration/testdata/"
+    "mobilenet_v1_1.0_224_quant_edgetpu.tflite";
+constexpr char kRegularModelFilePath[] =
+    "tensorflow_lite_support/acceleration/configuration/testdata/"
+    "mobilenet_v1_1.0_224_quant.tflite";
+constexpr char kImagePath[] =
+    "tensorflow_lite_support/acceleration/configuration/testdata/"
+    "burger.jpg";
+
+using ::tflite::task::vision::DecodeImageFromFile;
+using ::tflite::task::vision::ImageData;
+using ::tflite::task::vision::ImageDataFree;
+
+using EdgeTpuCoralPluginTest = testing::TestWithParam<std::string>;
+
+INSTANTIATE_TEST_SUITE_P(CoralPluginTests,
+                         EdgeTpuCoralPluginTest,
+                         testing::Values(kRegularModelFilePath,
+                                         kEdgeTpuModelFilePath));
+
+TEST_P(EdgeTpuCoralPluginTest, CreateEdgeTpuCoralPlugin) {
+  // Create the Coral delegate from the Coral plugin.
+  flatbuffers::FlatBufferBuilder flatbuffer_builder;
+  auto settings = flatbuffers::GetTemporaryPointer(
+      flatbuffer_builder,
+      CreateTFLiteSettings(flatbuffer_builder, tflite::Delegate_EDGETPU_CORAL));
+  auto plugin = ::tflite::delegates::DelegatePluginRegistry::CreateByName(
+      "EdgeTpuCoralPlugin", *settings);
+  auto coral_delegate = plugin->Create();
+
+  // Load the tflite model file.
+  std::unique_ptr<::tflite::FlatBufferModel> tflite_model =
+      ::tflite::FlatBufferModel::BuildFromFile(GetParam().c_str());
+  ASSERT_NE(tflite_model, nullptr);
+
+  // Create the tflite interpreter.
+  tflite::ops::builtin::BuiltinOpResolver resolver;
+  std::unique_ptr<::tflite::Interpreter> interpreter;
+  ASSERT_EQ(::tflite::InterpreterBuilder(*tflite_model, resolver)(&interpreter),
+            kTfLiteOk);
+  ASSERT_NE(interpreter, nullptr);
+  interpreter->ModifyGraphWithDelegate(coral_delegate.get());
+
+  // Verifies that interpreter runs correctly.
+  // To open source the code under tensorflow/lite, the following code needs to
+  // be stript from the Task library dependency, meaning forking or rewriting
+  // `LoadImage` and `ImageData`.
+  // `ASSERT_OK_AND_ASSIGN` is not available externally.
+  auto rgb_image_or = DecodeImageFromFile(kImagePath);
+  ASSERT_TRUE(rgb_image_or.ok());
+
+  ImageData rgb_image = rgb_image_or.value();
+  const uint8_t* input_data = rgb_image.pixel_data;
+  size_t input_data_byte_size =
+      rgb_image.width * rgb_image.height * rgb_image.channels * sizeof(uint8_t);
+
+  ASSERT_EQ(interpreter->AllocateTensors(), kTfLiteOk);
+  uint8_t* input_tensor = interpreter->typed_input_tensor<uint8_t>(0);
+  memcpy(input_tensor, input_data, input_data_byte_size);
+  ASSERT_EQ(interpreter->Invoke(), kTfLiteOk);
+  uint8_t* output_tensor = interpreter->typed_output_tensor<uint8_t>(0);
+  // `cheeseburger` is the 935th item in the label file of
+  // "mobilenet_v1_1.0_224_quant_edgetpu.tflite". See labels.txt.
+  EXPECT_EQ(output_tensor[934], 255);
+  ImageDataFree(&rgb_image);
+}
+
+}  // namespace
+}  // namespace delegates
+}  // namespace tflite

third_party/tflite_support/src/tensorflow_lite_support/acceleration/configuration/testdata/BUILD

@@ -0,0 +1,12 @@
+package(
+    default_visibility = ["//tensorflow_lite_support:internal"],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+filegroup(
+    name = "test_files",
+    srcs = glob([
+        "*.tflite",
+        "*.jpg",
+    ]),
+)

third_party/tflite_support/src/tensorflow_lite_support/c/BUILD

@@ -0,0 +1,23 @@
+package(
+    default_visibility = ["//tensorflow_lite_support:internal"],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library(
+    name = "common",
+    srcs = ["common.cc"],
+    hdrs = ["common.h"],
+)
+
+cc_library(
+    name = "common_utils",
+    srcs = ["common_utils.cc"],
+    hdrs = ["common_utils.h"],
+    deps = [
+        ":common",
+        "//tensorflow_lite_support/cc:common",
+        "@com_google_absl//absl/status",
+        "@com_google_absl//absl/strings",
+        "@com_google_absl//absl/strings:cord",
+    ],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/common.cc

@@ -1,4 +1,4 @@
-/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
 
 Licensed under the Apache License, Version 2.0 (the "License");
 you may not use this file except in compliance with the License.
@@ -12,16 +12,14 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License.
 ==============================================================================*/
-#inmport "base/strings/sys_string_conversions.h"
-#import "third_party/tensorflow_lite_support/ios/utils/Sources/TFLStringUtil.h"
 
-std::string MakeString(NSString* str) {
-  return SysNSStringToUTF8(str);
-}
+#include "tensorflow_lite_support/c/common.h"
 
-NSString* MakeNSString(const std::string& str) {
-  return [[NSString alloc]
-      initWithBytes:const_cast<void*>(static_cast<const void*>(str.data()))
-             length:str.length()
-           encoding:NSUTF8StringEncoding];
+#include <cstdlib>
+
+void TfLiteSupportErrorDelete(TfLiteSupportError* error) {
+  // `strdup` obtains memory using `malloc` and the memory needs to be
+  // released using `free`.
+  free(error->message);
+  delete error;
 }

third_party/tflite_support/src/tensorflow_lite_support/c/common.h

@@ -0,0 +1,202 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_COMMON_H_
+#define TENSORFLOW_LITE_SUPPORT_C_COMMON_H_
+
+// Defines C struct and error codes for describing any error returned from the C
+// Task Library.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// Error codes for TensorFlow Lite Task Library C APIs.
+//
+// Holds one to one mapping with `TfLiteSupportStatus` code starting from kError
+// = 1. Omits `kOk` since `TfLiteErrorCode` is only to be used in the event of
+// an error and does not account for success unlike `TfLiteSupportStatus`. In
+// case of success, TensorFlow Lite Task Library C APIs return the appropriate
+// return value and a null error. One to one mapping makes it easier to convert
+// between `TfLiteSupportStatus` and `TfLiteSupportErrorCode` without long
+// switch statements.
+//
+// Also holds error codes mapping to absl::Status::code() starting from
+// kNotFound = 900 in cases where the absl::Status payload can't
+// be mapped to a `TfLiteSupportStatus` code. kErrorCodeFirst and kErrorCodeLast
+// are also provided for safety checks during conversion between
+// `TfLiteSupportStatus` and `TfLiteSupportErrorCode`. In case of modifications
+// in error codes, ensure that kErrorCodeFirst and kErrorCodeLast is
+// respectively, set to the least and greatest enum value amongst the error
+// codes mapping to TfLiteSupportStatus.
+enum TfLiteSupportErrorCode {
+  // Unspecified error.
+  kError = 1,
+  // Invalid argument specified.
+  kInvalidArgumentError = 2,
+  // Invalid FlatBuffer file or buffer specified.
+  kInvalidFlatBufferError = 3,
+  // Model contains a builtin op that isn't supported by the OpResolver or
+  // delegates.
+  kUnsupportedBuiltinOpError = 4,
+  // Model contains a custom op that isn't supported by the OpResolver or
+  // delegates.
+  kUnsupportedCustomOpError = 5,
+
+  // File I/O error codes.
+
+  // No such file.
+  kFileNotFoundError = 100,
+  // Permission issue.
+  kFilePermissionDeniedError,
+  // I/O error when reading file.
+  kFileReadError,
+  // I/O error when mmap-ing file.
+  kFileMmapError,
+
+  // TensorFlow Lite metadata error codes.
+
+  // Unexpected schema version (aka file_identifier) in the Metadata FlatBuffer.
+  kMetadataInvalidSchemaVersionError = 200,
+  // No such associated file within metadata, or file has not been packed.
+  kMetadataAssociatedFileNotFoundError,
+  // ZIP I/O error when unpacking an associated file.
+  kMetadataAssociatedFileZipError,
+  // Inconsistency error between the metadata and actual TF Lite model.
+  // E.g.: number of labels and output tensor values differ.
+  kMetadataInconsistencyError,
+  // Invalid process units specified.
+  // E.g.: multiple ProcessUnits with the same type for a given tensor.
+  kMetadataInvalidProcessUnitsError,
+  // Inconsistency error with the number of labels.
+  // E.g.: label files for different locales have a different number of labels.
+  kMetadataNumLabelsMismatchError,
+  // Score calibration parameters parsing error.
+  // E.g.: too many parameters provided in the corresponding associated file.
+  kMetadataMalformedScoreCalibrationError,
+  // Unexpected number of subgraphs for the current task.
+  // E.g.: image classification expects a single subgraph.
+  kMetadataInvalidNumSubgraphsError,
+  // A given tensor requires NormalizationOptions but none were found.
+  // E.g.: float input tensor requires normalization to preprocess input images.
+  kMetadataMissingNormalizationOptionsError,
+  // Invalid ContentProperties specified.
+  // E.g. expected ImageProperties, got BoundingBoxProperties.
+  kMetadataInvalidContentPropertiesError,
+  // Metadata is mandatory but was not found.
+  // E.g. current task requires TFLite Model Metadata but none was found.
+  kMetadataNotFoundError,
+  // Associated TENSOR_AXIS_LABELS or TENSOR_VALUE_LABELS file is mandatory but
+  // none was found or it was empty.
+  // E.g. current task requires labels but none were found.
+  kMetadataMissingLabelsError,
+  // The ProcessingUnit for tokenizer is not correctly configured.
+  // E.g BertTokenizer doesn't have a valid vocab file associated.
+  kMetadataInvalidTokenizerError,
+
+  // Input tensor(s) error codes.
+
+  // Unexpected number of input tensors for the current task.
+  // E.g. current task expects a single input tensor.
+  kInvalidNumInputTensorsError = 300,
+  // Unexpected input tensor dimensions for the current task.
+  // E.g.: only 4D input tensors supported.
+  kInvalidInputTensorDimensionsError,
+  // Unexpected input tensor type for the current task.
+  // E.g.: current task expects a uint8 pixel image as input.
+  kInvalidInputTensorTypeError,
+  // Unexpected input tensor bytes size.
+  // E.g.: size in bytes does not correspond to the expected number of pixels.
+  kInvalidInputTensorSizeError,
+  // No correct input tensor found for the model.
+  // E.g.: input tensor name is not part of the text model's input tensors.
+  kInputTensorNotFoundError,
+
+  // Output tensor(s) error codes.
+
+  // Unexpected output tensor dimensions for the current task.
+  // E.g.: only a batch size of 1 is supported.
+  kInvalidOutputTensorDimensionsError = 400,
+  // Unexpected input tensor type for the current task.
+  // E.g.: multi-head model with different output tensor types.
+  kInvalidOutputTensorTypeError,
+  // No correct output tensor found for the model.
+  // E.g.: output tensor name is not part of the text model's output tensors.
+  kOutputTensorNotFoundError,
+  // Unexpected number of output tensors for the current task.
+  // E.g.: current task expects a single output tensor.
+  kInvalidNumOutputTensorsError,
+
+  // Image processing error codes.
+
+  // Unspecified image processing failures.
+  kImageProcessingError = 500,
+  // Unexpected input or output buffer metadata.
+  // E.g.: rotate RGBA buffer to Grayscale buffer by 90 degrees.
+  kImageProcessingInvalidArgumentError,
+  // Image processing operation failures.
+  // E.g. libyuv rotation failed for an unknown reason.
+  kImageProcessingBackendError,
+
+  // Convenience error codes for condition checks during type casting.
+  //
+  // Codes mapping to absl status codes should not be considered for these
+  // ranges.
+  // They must be used exclsively for checking if error codes fall in valid
+  // ranges when converting between TfLiteSupportStatus and
+  // TfLiteSupportErrorCodee.
+
+  // Ensure it holds the least enum value amongst error codes mapping to
+  // TfLiteSupportStatus.
+  kErrorCodeFirst = kError,
+  // Ensure it holds the greatest enum value amongst error codes mapping to
+  // TfLiteSupportStatus.
+  kErrorCodeLast = kImageProcessingBackendError,
+
+  // Absl Status Codes Mapping
+  //
+  // Codes starting from 900 will be used to map absl::Status created by TfLite
+  // and are used as is by TfLite Support C++ layer. Such absl status objects
+  // don't have a TfLiteSupportStatus in the payload that can be mapped to other
+  // error codes in this struct. You must use the absl::Status::code() and map
+  // them to the following error codes in such cases.
+  // For more info on respective absl status codes, please see:
+  // https://github.com/abseil/abseil-cpp/blob/master/absl/status/status.h#L91
+
+  // kNotFound indicates some requested entity (such as a file or directory)
+  // was not found.
+  kNotFoundError = 900,
+  // kInternal indicates an internal error has occurred
+  // and some invariants expected by the underlying system have not been
+  // satisfied. This error code is reserved for serious errors.
+  kInternalError,
+};
+
+// A `TfLiteSupportError` encapsulates an error code and a descriptive message
+// to return in the event of an error being encountered in any TensorFlow Lite
+// Task Library C API.
+typedef struct TfLiteSupportError {
+  // Holds the error code.
+  enum TfLiteSupportErrorCode code;
+  // Detailed description of the error.
+  char* message;
+} TfLiteSupportError;
+
+void TfLiteSupportErrorDelete(TfLiteSupportError* error);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_COMMON_H_

third_party/tflite_support/src/tensorflow_lite_support/c/common_utils.cc

@@ -0,0 +1,111 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow_lite_support/c/common_utils.h"
+
+#include <string>
+
+#include "absl/status/status.h"  // from @com_google_absl
+#include "absl/strings/cord.h"   // from @com_google_absl
+#include "tensorflow_lite_support/cc/common.h"
+
+namespace tflite {
+namespace support {
+
+void CreateTfLiteSupportError(enum TfLiteSupportErrorCode code,
+                              const char* message,
+                              TfLiteSupportError** error) {
+  if (error == nullptr)
+    return;
+
+  *error = new TfLiteSupportError;
+  (*error)->code = code;
+  (*error)->message = strdup(message);
+}
+
+void CreateTfLiteSupportErrorWithStatus(const absl::Status& status,
+                                        TfLiteSupportError** error) {
+  if (status.ok() || error == nullptr)
+    return;
+
+  // Payload of absl::Status created by the tflite task library stores an
+  // appropriate value of the enum TfLiteSupportStatus. The integer value
+  // corresponding to the TfLiteSupportStatus enum stored in the payload is
+  // extracted here to later map to the appropriate error code to be returned.
+  // In cases where the enum is not stored in (payload is NULL or the payload
+  // string cannot be converted to an integer), we set the error code value to
+  // be 1 (kError of TfLiteErrorCode used in the C library to signify any errors
+  // not falling into other categories.) Since payload is of type absl::Cord
+  // that can be type cast into an absl::optional<std::string>, we use the
+  // std::stoi function to convert it into an integer code if possible.
+  int generic_error_code = static_cast<int>(kError);
+  int error_code;
+  try {
+    // Try converting payload to integer if payload is not empty. Otherwise
+    // convert a string signifying generic error code kError to integer.
+    error_code = std::stoi(static_cast<absl::optional<std::string>>(
+                               status.GetPayload(kTfLiteSupportPayload))
+                               .value_or(std::to_string(generic_error_code)));
+  } catch (std::invalid_argument& e) {
+    // If non empty payload string cannot be converted to an integer. Set error
+    // code to 1(kError).
+    error_code = generic_error_code;
+  }
+
+  // If error_code is outside the range of enum values possible or is kError, we
+  // try to map the absl::Status::code() to assign appropriate
+  // TfLiteSupportErrorCode or kError in default cases. Note: The mapping to
+  // absl::Status::code() is done to generate a more specific error code than
+  // kError in cases when the payload can't be mapped to TfLiteSupportStatus.
+  // This can happen when absl::Status returned by TfLite are in turn returned
+  // without moodification by TfLite Support Methods.
+  if (error_code > static_cast<int>(kErrorCodeLast) ||
+      error_code <= static_cast<int>(kErrorCodeFirst)) {
+    switch (status.code()) {
+      case absl::StatusCode::kInternal:
+        error_code = kInternalError;
+        break;
+      case absl::StatusCode::kInvalidArgument:
+        error_code = kInvalidArgumentError;
+        break;
+      case absl::StatusCode::kNotFound:
+        error_code = kNotFoundError;
+        break;
+      default:
+        error_code = kError;
+        break;
+    }
+  }
+
+  // Creates the TfLiteSupportError with the appropriate error
+  // TfLiteSupportErrorCode and message. TfLiteErrorCode has a one to one
+  // mapping with TfLiteSupportStatus starting from the value 1(kError) and
+  // hence will be correctly initialized if directly cast from the integer code
+  // derived from TfLiteSupportStatus stored in payload. TfLiteErrorCode omits
+  // kOk = 0 of TfLiteSupportStatus.
+  //
+  // Stores a string including absl status code and message(if non empty) as the
+  // error message See
+  // https://github.com/abseil/abseil-cpp/blob/master/absl/status/status.h#L514
+  // for explanation. absl::Status::message() can also be used but not always
+  // guaranteed to be non empty.
+  CreateTfLiteSupportError(
+      static_cast<TfLiteSupportErrorCode>(error_code),
+      status.ToString(absl::StatusToStringMode::kWithNoExtraData).c_str(),
+      error);
+}
+
+}  // namespace support
+}  // namespace tflite

third_party/tflite_support/src/tensorflow_lite_support/c/common_utils.h

@@ -0,0 +1,57 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_COMMON_UTILS_H_
+#define TENSORFLOW_LITE_SUPPORT_C_COMMON_UTILS_H_
+
+#include "absl/status/status.h"  // from @com_google_absl
+#include "tensorflow_lite_support/c/common.h"
+
+// Utils for Conversion of absl::Status to TfLiteError
+// -----------------------------------------------------------------
+// Meant to be used with task C apis.
+
+namespace tflite {
+namespace support {
+
+// Creates a TfLiteSupportError with a TfLiteSupportErrorCode and message.
+void CreateTfLiteSupportError(enum TfLiteSupportErrorCode code,
+                              const char* message,
+                              TfLiteSupportError** error);
+
+// Creates a TfLiteSupportError from absl::Status and passes it back as a
+// parameter which is a pointer to the error pointer.
+//
+// Example Usage With Image Classifier
+//
+// APIs: TfLiteImageClassifier* TfLiteImageClassifierFromOptions(
+//     const TfLiteImageClassifierOptions* options,
+//     TfLiteSupportError **error) {
+// // Necessary checks
+// tflite::support::StatusOr<std::unique_ptr<ImageClassifier>> classifier_status
+// = // Call to create Cpp Image Classifier.
+// if (classifier_status.ok()) {
+//     Code to return classifier
+// } else {
+//     ::tflite::support::CreateTfLiteSupportErrorWithStatus(classifier_status.status(),
+//     error);
+//     return nullptr;
+//  }
+//}
+void CreateTfLiteSupportErrorWithStatus(const absl::Status& status,
+                                        TfLiteSupportError** error);
+
+}  // namespace support
+}  // namespace tflite
+#endif  // TENSORFLOW_LITE_SUPPORT_C_COMMON_UTILS_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/core/BUILD

@@ -0,0 +1,9 @@
+package(
+    default_visibility = ["//tensorflow_lite_support:internal"],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library(
+    name = "base_options",
+    hdrs = ["base_options.h"],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/task/core/base_options.h

@@ -0,0 +1,73 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_CORE_BASE_OPTIONS_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_CORE_BASE_OPTIONS_H_
+
+#include <stdint.h>
+
+// Defines C Structs for Base Options Shared by all tasks.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// Holds cpu settings.
+typedef struct TfLiteCpuSettings {
+  // Specifies the number of threads to be used for TFLite
+  // ops that support multi-threading when running inference with CPU.
+  // num_threads should be greater than 0 or equal to -1. Setting num_threads to
+  // -1 has the effect to let TFLite runtime set the value.
+  int num_threads;
+} TfLiteCpuSettings;
+
+// Holds settings for one possible acceleration configuration.
+typedef struct TfLiteComputeSettings {
+  // Holds cpu settings
+  TfLiteCpuSettings cpu_settings;
+} TfLiteComputeSettings;
+
+// Represents external files used by the Task APIs (e.g. TF Lite Model File).
+// For now you can only specify the path of the file using file_path:
+// In future other sources may be supported.
+typedef struct TfLiteExternalFile {
+  // The path to the file to open.
+  const char* file_path;
+  // Additional option for byte data when it's supported.
+} TfLiteExternalFile;
+
+// Holds the base options that is used for creation of any type of task. It has
+// fields withh important information acceleration configuration, tflite model
+// source etc.
+// This struct must be zero initialized before setting any options as this
+// will result in seg faults.
+typedef struct TfLiteBaseOptions {
+  // The external model file, as a single standalone TFLite file. It could be
+  // packed with TFLite Model Metadata[1] and associated files if exist. Fail to
+  // provide the necessary metadata and associated files might result in errors.
+  // Check the documentation for each task about the specific requirement.
+  // [1]: https://www.tensorflow.org/lite/convert/metadata
+  TfLiteExternalFile model_file;
+
+  // Holds settings for one possible acceleration configuration
+  // including.cpu/gpu settings. Please see documentation of
+  // TfLiteComputeSettings and its members for more details.
+  TfLiteComputeSettings compute_settings;
+} TfLiteBaseOptions;
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_CORE_BASE_OPTIONS_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/processor/BUILD

@@ -0,0 +1,42 @@
+package(
+    default_visibility = ["//tensorflow_lite_support:internal"],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library(
+    name = "category",
+    hdrs = ["category.h"],
+)
+
+cc_library(
+    name = "classification_result",
+    srcs = [
+        "classification_result.cc",
+    ],
+    hdrs = ["classification_result.h"],
+    deps = [
+        ":category",
+    ],
+)
+
+cc_library(
+    name = "bounding_box",
+    hdrs = ["bounding_box.h"],
+)
+
+cc_library(
+    name = "classification_options",
+    hdrs = ["classification_options.h"],
+)
+
+cc_library(
+    name = "detection_result",
+    srcs = [
+        "detection_result.cc",
+    ],
+    hdrs = ["detection_result.h"],
+    deps = [
+        ":bounding_box",
+        ":category",
+    ],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/task/processor/bounding_box.h

@@ -0,0 +1,45 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_BOUNDING_BOX_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_BOUNDING_BOX_H_
+
+#include <stdint.h>
+
+// Defines C Struct for Bounding Box Shared by Vision Tasks.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// Holds the region of interest used for image classification.
+typedef struct TfLiteBoundingBox {
+  // The X coordinate of the top-left corner, in pixels.
+  int origin_x;
+
+  // The Y coordinate of the top-left corner, in pixels.
+  int origin_y;
+
+  // The width of the bounding box, in pixels.
+  int width;
+
+  // The height of the bounding box, in pixels.
+  int height;
+} TfLiteBoundingBox;
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_BOUNDING_BOX_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/processor/category.h

@@ -0,0 +1,49 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_CATEGORY_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_CATEGORY_H_
+
+// Defines C structure for a Category which encapsulates a single predicted
+// class.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// A single predicted class.
+typedef struct TfLiteCategory {
+  // The index of the class in the corresponding label map, usually packed in
+  // the TFLite Model Metadata [1].
+  //
+  // [1]: https://www.tensorflow.org/lite/convert/metadata
+  int index;
+
+  // The score for this class e.g. (but not necessarily) a probability in [0,1].
+  float score;
+
+  // A human readable name of the class filled from the label map.
+  char* display_name;
+  // An ID for the class, not necessarily human-readable (e.g. a Google
+  // Knowledge Graph ID [1]), filled from the label map.
+  //
+  // [1]: https://developers.google.com/knowledge-graph
+  char* label;
+} TfLiteCategory;
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_CATEGORY_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/processor/classification_options.h

@@ -0,0 +1,66 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_CLASSIFICATION_OPTIONS_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_CLASSIFICATION_OPTIONS_H_
+
+#include <stdint.h>
+
+// Defines C Struct for Classification Options Shared by All Classification
+// Tasks.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// Holds pointer to array of C strings and length for looping through the array.
+typedef struct TfLiteStringArrayOption {
+  // Length of list. length can be used to loop through list.
+  int length;
+
+  // Array of C strings.
+  char** list;
+} TfLiteStringArrayOption;
+
+// Holds settings for any single classification task.
+typedef struct TfLiteClassificationOptions {
+  // Optional denylist of class labels. If non NULL, classifications whose
+  // class label is in this set will be filtered out. Duplicate or unknown
+  // class labels are ignored. Mutually exclusive with label_allowlist.
+  TfLiteStringArrayOption label_denylist;
+
+  // Optional allowlist of class labels. If non-empty, classifications whose
+  // class label is not in this set will be filtered out. Duplicate or unknown
+  // class labels are ignored. Mutually exclusive with label_denylist.
+  TfLiteStringArrayOption label_allowlist;
+
+  // The locale to use for display names specified through the TFLite Model
+  // Metadata, if any. Defaults to English.
+  char* display_names_local;
+
+  // The maximum number of top-scored classification results to return. If < 0,
+  // all available results will be returned. If 0, an invalid argument error is
+  // returned. Defaults to -1.
+  int max_results;
+
+  // Score threshold, overrides the ones provided in the model metadata
+  // (if any). Results below this value are rejected.
+  float score_threshold;
+} TfLiteClassificationOptions;
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_CLASSIFICATION_OPTIONS_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/processor/classification_result.cc

@@ -0,0 +1,46 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow_lite_support/c/task/processor/classification_result.h"
+
+#include <cstdlib>
+#include <memory>
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+void TfLiteClassificationResultDelete(
+    TfLiteClassificationResult* classification_result) {
+  for (int head = 0; head < classification_result->size; ++head) {
+    TfLiteClassifications classifications =
+        classification_result->classifications[head];
+    for (int rank = 0; rank < classifications.size; ++rank) {
+      // `strdup` obtains memory using `malloc` and the memory needs to be
+      // released using `free`.
+      free(classifications.categories[rank].display_name);
+      free(classifications.categories[rank].label);
+    }
+
+    delete[] classifications.categories;
+  }
+
+  delete[] classification_result->classifications;
+  delete classification_result;
+}
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus

third_party/tflite_support/src/tensorflow_lite_support/c/task/processor/classification_result.h

@@ -0,0 +1,63 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_CLASSIFICATION_RESULT_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_CLASSIFICATION_RESULT_H_
+
+#include "tensorflow_lite_support/c/task/processor/category.h"
+
+// Defines C structure for Classification Results and associated helper methods.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// List of predicted classes (aka labels) for a given image classifier head.
+typedef struct TfLiteClassifications {
+  // The index of the image classifier head these classes refer to. This is
+  // useful for multi-head models.
+  int head_index;
+
+  // Number of predicted classes which can be used to traverse the array of
+  // predicted classes.
+  int size;
+
+  // The array of predicted classes, usually sorted by descending scores (e.g.
+  // from high to low probability). Since this array is dynamically allocated,
+  // use size to traverse through the array.
+  TfLiteCategory* categories;
+} TfLiteClassifications;
+
+// Holds Image Classification results.
+// Contains one set of results per image classifier head.
+typedef struct TfLiteClassificationResult {
+  // Number of predicted classes which can be used to traverse the array of
+  // predicted classes.
+  int size;
+
+  // Array of image classifier results per image classifier head. This array can
+  // have any number of results. size holds the size of this array. size should
+  // be used to traverse this array.
+  TfLiteClassifications* classifications;
+} TfLiteClassificationResult;
+
+// Frees up the ClassificationResult Structure.
+void TfLiteClassificationResultDelete(
+    TfLiteClassificationResult* classification_result);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_CLASSIFICATION_RESULT_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/processor/detection_result.h

@@ -0,0 +1,65 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_DETECTION_RESULT_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_PROCESSOR_DETECTION_RESULT_H_
+
+#include "tensorflow_lite_support/c/task/processor/bounding_box.h"
+#include "tensorflow_lite_support/c/task/processor/category.h"
+
+// Defines C structure for Object Detection Results and associated helper
+// methods.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// Bounding box and list of predicted classes (aka labels) for a detected
+// object.
+typedef struct TfLiteDetection {
+  // The bounding box of the detected object.
+  TfLiteBoundingBox bounding_box;
+
+  // The array of predicted classes for the object detection represented by an
+  // instance of TfLiteDetection, usually sorted by descending scores (e.g. from
+  // high to low probability). Since this array is dynamically allocated, use
+  // size to traverse through the array.
+  TfLiteCategory* categories;
+
+  // Number of detectd objects be used to traverse the array of the detected
+  // objects.
+  int size;
+} TfLiteDetection;
+
+// Holds Object Detection results.
+// Contains one set of results per detected object.
+typedef struct TfLiteDetectionResult {
+  // Number of detectd objects be used to traverse the array of the detected
+  // objects.
+  int size;
+
+  // Array of results per detected object. This array can
+  // have any number of results. size holds the size of this array. size should
+  // be used to traverse this array.
+  TfLiteDetection* detections;
+} TfLiteDetectionResult;
+
+// Frees up the DetectionResult Structure.
+void TfLiteDetectionResultDelete(TfLiteDetectionResult* detection_result);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_DETECTION_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/BUILD

@@ -0,0 +1,79 @@
+load(
+    "@org_tensorflow//tensorflow/lite/core/shims:cc_library_with_tflite.bzl",
+    "cc_library_with_tflite",
+)
+
+package(
+    default_visibility = ["//tensorflow_lite_support:internal"],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+exports_files([
+    "bert_nl_classifier.h",
+    "nl_classifier.h",
+    "nl_classifier_common.h",
+    "bert_question_answerer.h",
+])
+
+cc_library_with_tflite(
+    name = "nl_classifier",
+    srcs = [
+        "nl_classifier.cc",
+    ],
+    hdrs = [
+        "nl_classifier.h",
+        "nl_classifier_common.h",
+    ],
+    tflite_deps = [
+        "//tensorflow_lite_support/cc/task/text/nlclassifier:nl_classifier",
+    ],
+    deps = [
+        ":nl_classifier_common",
+        "//tensorflow_lite_support/cc/task/core:category",
+        "@com_google_absl//absl/strings",
+    ],
+)
+
+cc_library_with_tflite(
+    name = "bert_nl_classifier",
+    srcs = [
+        "bert_nl_classifier.cc",
+    ],
+    hdrs = [
+        "bert_nl_classifier.h",
+        "nl_classifier_common.h",
+    ],
+    tflite_deps = [
+        "//tensorflow_lite_support/cc/task/text:bert_nl_classifier",
+    ],
+    deps = [
+        ":nl_classifier_common",
+        "//tensorflow_lite_support/cc/task/core:category",
+        "//tensorflow_lite_support/cc/task/text/proto:bert_nl_classifier_options_proto_inc",
+        "@com_google_absl//absl/strings",
+    ],
+)
+
+cc_library(
+    name = "nl_classifier_common",
+    srcs = [
+        "nl_classifier_common.cc",
+    ],
+    hdrs = [
+        "nl_classifier_common.h",
+    ],
+)
+
+cc_library_with_tflite(
+    name = "bert_question_answerer",
+    srcs = [
+        "bert_question_answerer.cc",
+    ],
+    hdrs = [
+        "bert_question_answerer.h",
+    ],
+    tflite_deps = [
+        "//tensorflow_lite_support/cc/task/text:bert_question_answerer",
+        "//tensorflow_lite_support/cc/task/text:question_answerer",
+    ],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/bert_nl_classifier.cc

@@ -0,0 +1,93 @@
+/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow_lite_support/c/task/text/bert_nl_classifier.h"
+
+#include <memory>
+
+#include "absl/strings/string_view.h"  // from @com_google_absl
+#include "tensorflow_lite_support/cc/task/core/category.h"
+#include "tensorflow_lite_support/cc/task/text/bert_nl_classifier.h"
+#include "tensorflow_lite_support/cc/task/text/proto/bert_nl_classifier_options_proto_inc.h"
+
+namespace {
+using CategoryCpp = ::tflite::task::core::Category;
+using BertNLClassifierCpp = ::tflite::task::text::BertNLClassifier;
+using BertNLClassifierOptionsCpp =
+    ::tflite::task::text::BertNLClassifierOptions;
+
+const TfLiteBertNLClassifierOptions kBertNLClassifierOptionsDefault = {128};
+}  // namespace
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+struct TfLiteBertNLClassifier {
+  std::unique_ptr<BertNLClassifierCpp> impl;
+};
+
+TfLiteBertNLClassifier* TfLiteBertNLClassifierCreateFromOptions(
+    const char* model_path,
+    const TfLiteBertNLClassifierOptions* options) {
+  BertNLClassifierOptionsCpp cc_options;
+
+  cc_options.mutable_base_options()->mutable_model_file()->set_file_name(
+      model_path);
+  auto classifier_status = BertNLClassifierCpp::CreateFromOptions(cc_options);
+
+  if (classifier_status.ok()) {
+    return new TfLiteBertNLClassifier{
+        .impl = std::unique_ptr<BertNLClassifierCpp>(
+            dynamic_cast<BertNLClassifierCpp*>(
+                classifier_status.value().release()))};
+
+  } else {
+    return nullptr;
+  }
+}
+
+TfLiteBertNLClassifier* TfLiteBertNLClassifierCreate(const char* model_path) {
+  return TfLiteBertNLClassifierCreateFromOptions(
+      model_path, &kBertNLClassifierOptionsDefault);
+}
+
+Categories* TfLiteBertNLClassifierClassify(
+    const TfLiteBertNLClassifier* classifier,
+    const char* text) {
+  std::vector<CategoryCpp> results =
+
+      classifier->impl->Classify(absl::string_view(text).data());
+  size_t size = results.size();
+  auto* categories = new Category[size];
+
+  for (size_t i = 0; i < size; ++i) {
+    categories[i].text = strdup(results[i].class_name.c_str());
+    categories[i].score = results[i].score;
+  }
+
+  auto* c_categories = new Categories;
+  c_categories->size = size;
+  c_categories->categories = categories;
+  return c_categories;
+}
+
+void TfLiteBertNLClassifierDelete(TfLiteBertNLClassifier* classifier) {
+  delete classifier;
+}
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/bert_nl_classifier.h

@@ -0,0 +1,70 @@
+/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_BERT_NL_CLASSIFIER_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_BERT_NL_CLASSIFIER_H_
+
+#include "tensorflow_lite_support/c/task/text/nl_classifier_common.h"
+// --------------------------------------------------------------------------
+// C API for BertNLClassifier.
+//
+// Usage:
+// // Create the model and interpreter options.
+// TfLiteBertNLClassifier* classifier =
+//     TfLiteBertNLClassifierCreate("/path/to/model.tflite");
+//
+// // Classification.
+// Categories* categories = TfLiteBertNLClassifierClassify(classifier,
+//     question);
+//
+// // Dispose of the API object.
+// TfLiteBertNLClassifierDelete(classifier);
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+typedef struct TfLiteBertNLClassifier TfLiteBertNLClassifier;
+
+typedef struct TfLiteBertNLClassifierOptions {
+  // Max number of tokens to pass to the model.
+  //
+  // Deprecated: max_seq_len is now read from the model (i.e. input tensor size)
+  // automatically.
+  int max_seq_len;
+} TfLiteBertNLClassifierOptions;
+
+// Creates TfLiteBertNLClassifier from model path and options, returns nullptr
+// if the file doesn't exist or is not a well formatted TFLite model path.
+TfLiteBertNLClassifier* TfLiteBertNLClassifierCreateFromOptions(
+    const char* model_path,
+    const TfLiteBertNLClassifierOptions* options);
+
+// Creates TfLiteBertNLClassifier from model path and default options, returns
+// nullptr if the file doesn't exist or is not a well formatted TFLite model
+// path.
+TfLiteBertNLClassifier* TfLiteBertNLClassifierCreate(const char* model_path);
+
+// Invokes the encapsulated TFLite model and classifies the input text.
+Categories* TfLiteBertNLClassifierClassify(
+    const TfLiteBertNLClassifier* classifier,
+    const char* text);
+
+void TfLiteBertNLClassifierDelete(TfLiteBertNLClassifier* classifier);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_BERT_NL_CLASSIFIER_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/bert_question_answerer.cc

@@ -13,45 +13,48 @@ See the License for the specific language governing permissions and
 limitations under the License.
 ==============================================================================*/
 
-#include "tensorflow_lite_support/cc/task/text/qa/bert_qa_c_api.h"
+#include "tensorflow_lite_support/c/task/text/bert_question_answerer.h"
 
 #include <memory>
 
-#include "tensorflow_lite_support/cc/task/text/qa/bert_question_answerer.h"
-#include "tensorflow_lite_support/cc/task/text/qa/question_answerer.h"
+#include "tensorflow_lite_support/cc/task/text/bert_question_answerer.h"
+#include "tensorflow_lite_support/cc/task/text/question_answerer.h"
 
-using BertQuestionAnswererCPP = ::tflite::task::text::qa::BertQuestionAnswerer;
-using QaAnswerCPP = ::tflite::task::text::qa::QaAnswer;
+namespace {
+using BertQuestionAnswererCpp = ::tflite::task::text::BertQuestionAnswerer;
+using QaAnswerCpp = ::tflite::task::text::QaAnswer;
+}  // namespace
 
 #ifdef __cplusplus
 extern "C" {
 #endif  // __cplusplus
 
-struct BertQuestionAnswerer {
-  std::unique_ptr<BertQuestionAnswererCPP> impl;
+struct TfLiteBertQuestionAnswerer {
+  std::unique_ptr<BertQuestionAnswererCpp> impl;
 };
 
-BertQuestionAnswerer* BertQuestionAnswererFromFile(const char* model_path) {
+TfLiteBertQuestionAnswerer* TfLiteBertQuestionAnswererCreate(
+    const char* model_path) {
   auto bert_qa_status =
-      BertQuestionAnswererCPP::CreateFromFile(std::string(model_path));
+      BertQuestionAnswererCpp::CreateFromFile(std::string(model_path));
   if (bert_qa_status.ok()) {
-    return new BertQuestionAnswerer{
-        .impl = std::unique_ptr<BertQuestionAnswererCPP>(
-            dynamic_cast<BertQuestionAnswererCPP*>(
+    return new TfLiteBertQuestionAnswerer{
+        .impl = std::unique_ptr<BertQuestionAnswererCpp>(
+            dynamic_cast<BertQuestionAnswererCpp*>(
                 bert_qa_status.value().release()))};
   } else {
     return nullptr;
   }
 }
 
-QaAnswers* BertQuestionAnswererAnswer(
-    const BertQuestionAnswerer* question_answerer,
+TfLiteQaAnswers* TfLiteBertQuestionAnswererAnswer(
+    const TfLiteBertQuestionAnswerer* question_answerer,
     const char* context,
     const char* question) {
-  std::vector<QaAnswerCPP> answers = question_answerer->impl->Answer(
+  std::vector<QaAnswerCpp> answers = question_answerer->impl->Answer(
       absl::string_view(context).data(), absl::string_view(question).data());
   size_t size = answers.size();
-  auto* qa_answers = new QaAnswer[size];
+  auto* qa_answers = new TfLiteQaAnswer[size];
 
   for (size_t i = 0; i < size; ++i) {
     qa_answers[i].start = answers[i].pos.start;
@@ -60,17 +63,23 @@ QaAnswers* BertQuestionAnswererAnswer(
     qa_answers[i].text = strdup(answers[i].text.c_str());
   }
 
-  auto* c_answers = new QaAnswers;
+  auto* c_answers = new TfLiteQaAnswers;
   c_answers->size = size;
   c_answers->answers = qa_answers;
   return c_answers;
 }
 
-void BertQuestionAnswererDelete(BertQuestionAnswerer* bert_question_answerer) {
+void TfLiteBertQuestionAnswererDelete(
+    TfLiteBertQuestionAnswerer* bert_question_answerer) {
   delete bert_question_answerer;
 }
 
-void BertQuestionAnswererQaAnswersDelete(QaAnswers* qa_answers) {
+void TfLiteQaAnswersDelete(TfLiteQaAnswers* qa_answers) {
+  for (int i = 0; i < qa_answers->size; i++) {
+    // `strdup` obtains memory using `malloc` and the memory needs to be
+    // released using `free`.
+    free(qa_answers->answers[i].text);
+  }
   delete[] qa_answers->answers;
   delete qa_answers;
 }

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/bert_question_answerer.h

@@ -0,0 +1,74 @@
+/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_BERT_QUESTION_ANSWERER_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_BERT_QUESTION_ANSWERER_H_
+
+// --------------------------------------------------------------------------
+// C API for BertQuestionAnswerer.
+//
+// Usage:
+// <pre><code>
+// // Create the model and interpreter options.
+// TfLiteBertQuestionAnswerer* qa_answerer =
+//     TfLiteBertQuestionAnswererCreate("/path/to/model.tflite");
+//
+// // Answer a question.
+// TfLiteQaAnswers* answers = TfLiteBertQuestionAnswererAnswer(qa_answerer,
+//     question);
+//
+// // Dispose of the API and QaAnswers objects.
+// TfLiteBertQuestionAnswererDelete(qa_answerer);
+// TfLiteQaAnswersDelete(answers);
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+typedef struct TfLiteBertQuestionAnswerer TfLiteBertQuestionAnswerer;
+
+typedef struct TfLiteQaAnswer {
+  int start;
+  int end;
+  float logit;
+  char* text;
+} TfLiteQaAnswer;
+
+typedef struct TfLiteQaAnswers {
+  int size;
+  TfLiteQaAnswer* answers;
+} TfLiteQaAnswers;
+
+// Creates TfLiteBertQuestionAnswerer from model path, returns nullptr if the
+// file doesn't exist or is not a well formatted TFLite model path.
+TfLiteBertQuestionAnswerer* TfLiteBertQuestionAnswererCreate(
+    const char* model_path);
+
+// Invokes the encapsulated TFLite model and answers a question based on
+// context.
+TfLiteQaAnswers* TfLiteBertQuestionAnswererAnswer(
+    const TfLiteBertQuestionAnswerer* question_answerer,
+    const char* context,
+    const char* question);
+
+void TfLiteBertQuestionAnswererDelete(
+    TfLiteBertQuestionAnswerer* bert_question_answerer);
+
+void TfLiteQaAnswersDelete(TfLiteQaAnswers* qa_answers);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_BERT_QUESTION_ANSWERER_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/nl_classifier.cc

@@ -13,31 +13,33 @@ See the License for the specific language governing permissions and
 limitations under the License.
 ==============================================================================*/
 
-#include "tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier_c_api.h"
+#include "tensorflow_lite_support/c/task/text/nl_classifier.h"
 
 #include <memory>
 
-#include "absl/strings/string_view.h"
+#include "absl/strings/string_view.h"  // from @com_google_absl
 #include "tensorflow_lite_support/cc/task/core/category.h"
 #include "tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier.h"
 
-using CategoryCPP = ::tflite::task::core::Category;
-using NLClassifierCPP = ::tflite::task::text::nlclassifier::NLClassifier;
-using NLClassifierOptionsCPP =
+namespace {
+using CategoryCpp = ::tflite::task::core::Category;
+using NLClassifierCpp = ::tflite::task::text::nlclassifier::NLClassifier;
+using NLClassifierOptionsCpp =
     ::tflite::task::text::nlclassifier::NLClassifierOptions;
+}  // namespace
 
 #ifdef __cplusplus
 extern "C" {
 #endif  // __cplusplus
 
-struct NLClassifier {
-  std::unique_ptr<NLClassifierCPP> impl;
+struct TfLiteNLClassifier {
+  std::unique_ptr<NLClassifierCpp> impl;
 };
 
-NLClassifier* NLClassifierFromFileAndOptions(
+TfLiteNLClassifier* TfLiteNLClassifierCreateFromOptions(
     const char* model_path,
-    const NLClassifierOptions* options) {
-  auto classifier_status = NLClassifierCPP::CreateFromFileAndOptions(
+    const TfLiteNLClassifierOptions* options) {
+  auto classifier_status = NLClassifierCpp::CreateFromFileAndOptions(
       std::string(model_path),
       {
           .input_tensor_index = options->input_tensor_index,
@@ -57,17 +59,17 @@ NLClassifier* NLClassifierFromFileAndOptions(
       });
 
   if (classifier_status.ok()) {
-    return new NLClassifier{
-        .impl = std::unique_ptr<NLClassifierCPP>(dynamic_cast<NLClassifierCPP*>(
+    return new TfLiteNLClassifier{
+        .impl = std::unique_ptr<NLClassifierCpp>(dynamic_cast<NLClassifierCpp*>(
             classifier_status.value().release()))};
   } else {
     return nullptr;
   }
 }
 
-Categories* NLClassifierClassify(const NLClassifier* classifier,
-                                 const char* text) {
-  std::vector<CategoryCPP> results =
+Categories* TfLiteNLClassifierClassify(const TfLiteNLClassifier* classifier,
+                                       const char* text) {
+  std::vector<CategoryCpp> results =
       classifier->impl->Classify(absl::string_view(text).data());
   size_t size = results.size();
   auto* categories = new Category[size];
@@ -83,7 +85,7 @@ Categories* NLClassifierClassify(const NLClassifier* classifier,
   return c_categories;
 }
 
-void NLClassifierDelete(NLClassifier* classifier) {
+void TfLiteNLClassifierDelete(TfLiteNLClassifier* classifier) {
   delete classifier;
 }
 

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/nl_classifier.h

@@ -0,0 +1,64 @@
+/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_NL_CLASSIFIER_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_NL_CLASSIFIER_H_
+
+#include "tensorflow_lite_support/c/task/text/nl_classifier_common.h"
+// --------------------------------------------------------------------------
+// C API for NLClassifier.
+//
+// Usage:
+// // Create the model and interpreter options.
+// TfLiteNLClassifier* classifier = TfLiteNLClassifierCreate(
+//     "/path/to/model.tflite");
+//
+// // Classification.
+// Categories* categories = TfLiteNLClassifierClassify(classifier, question);
+//
+// // Dispose of the API object.
+// TfLiteNLClassifierDelete(classifier);
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+typedef struct TfLiteNLClassifier TfLiteNLClassifier;
+
+typedef struct TfLiteNLClassifierOptions {
+  int input_tensor_index;
+  int output_score_tensor_index;
+  int output_label_tensor_index;
+  const char* input_tensor_name;
+  const char* output_score_tensor_name;
+  const char* output_label_tensor_name;
+} TfLiteNLClassifierOptions;
+
+// Creates TfLiteNLClassifier from model path and options, returns nullptr if
+// the file doesn't exist or is not a well formatted TFLite model path.
+TfLiteNLClassifier* TfLiteNLClassifierCreateFromOptions(
+    const char* model_path,
+    const TfLiteNLClassifierOptions* options);
+
+// Invokes the encapsulated TFLite model and classifies the input text.
+Categories* TfLiteNLClassifierClassify(const TfLiteNLClassifier* classifier,
+                                       const char* text);
+
+void TfLiteNLClassifierDelete(TfLiteNLClassifier* classifier);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_NL_CLASSIFIER_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/nl_classifier_common.cc

@@ -13,13 +13,20 @@ See the License for the specific language governing permissions and
 limitations under the License.
 ==============================================================================*/
 
-#include "tensorflow_lite_support/cc/task/text/nlclassifier/nl_classifier_c_api_common.h"
+#include "tensorflow_lite_support/c/task/text/nl_classifier_common.h"
+
+#include <memory>
 
 #ifdef __cplusplus
 extern "C" {
 #endif  // __cplusplus
 
 void NLClassifierCategoriesDelete(Categories* categories) {
+  for (int i = 0; i < categories->size; i++) {
+    // `strdup` obtains memory using `malloc` and the memory needs to be
+    // released using `free`.
+    free(categories->categories[i].text);
+  }
   delete[] categories->categories;
   delete categories;
 }

third_party/tflite_support/src/tensorflow_lite_support/c/task/text/nl_classifier_common.h

@@ -12,32 +12,32 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
 limitations under the License.
 ==============================================================================*/
-#ifndef TENSORFLOW_LITE_SUPPORT_CC_TASK_TEXT_NLCLASSIFIER_NL_CLASSIFIER_C_API_COMMON_H_
-#define TENSORFLOW_LITE_SUPPORT_CC_TASK_TEXT_NLCLASSIFIER_NL_CLASSIFIER_C_API_COMMON_H_
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_NL_CLASSIFIER_COMMON_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_NL_CLASSIFIER_COMMON_H_
 
-// Common structs shared between NLClassifier APIs
-//
-/// // Dispose of the Categories object.
-/// NLClassifierCategoriesDelete(categories);
+// C API for the NLClassifier results, Catergory.
+
+// TODO(b/197355311): deprecate this class and use the unified one with image
+// and audio.
 
 #ifdef __cplusplus
 extern "C" {
 #endif  // __cplusplus
 
-struct Category {
+typedef struct Category {
   char* text;
   double score;
-};
+} Category;
 
-struct Categories {
+typedef struct Categories {
   int size;
-  struct Category* categories;
-};
+  Category* categories;
+} Categories;
 
-extern void NLClassifierCategoriesDelete(struct Categories* categories);
+void NLClassifierCategoriesDelete(Categories* categories);
 
 #ifdef __cplusplus
 }  // extern "C"
 #endif  // __cplusplus
 
-#endif  // TENSORFLOW_LITE_SUPPORT_CC_TASK_TEXT_NLCLASSIFIER_NL_CLASSIFIER_C_API_COMMON_H_
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_TEXT_NL_CLASSIFIER_COMMON_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/BUILD

@@ -0,0 +1,50 @@
+load("@org_tensorflow//tensorflow/lite/core/shims:cc_library_with_tflite.bzl", "cc_library_with_tflite")
+
+package(
+    default_visibility = [
+        "//tensorflow_lite_support:internal",
+    ],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library_with_tflite(
+    name = "image_classifier",
+    srcs = [
+        "image_classifier.cc",
+    ],
+    hdrs = [
+        "image_classifier.h",
+    ],
+    tflite_deps = [
+        "//tensorflow_lite_support/cc/task/vision:image_classifier",
+    ],
+    deps = [
+        "//tensorflow_lite_support/c:common",
+        "//tensorflow_lite_support/c:common_utils",
+        "//tensorflow_lite_support/c/task/core:base_options",
+        "//tensorflow_lite_support/c/task/processor:bounding_box",
+        "//tensorflow_lite_support/c/task/processor:classification_options",
+        "//tensorflow_lite_support/c/task/processor:classification_result",
+        "//tensorflow_lite_support/c/task/vision/core:frame_buffer",
+        "//tensorflow_lite_support/c/task/vision/utils:frame_buffer_cpp_c_utils",
+        "//tensorflow_lite_support/cc/task/vision/proto:classifications_proto_inc",
+        "//tensorflow_lite_support/cc/task/vision/proto:image_classifier_options_proto_inc",
+    ],
+)
+
+cc_library_with_tflite(
+    name = "object_detector",
+    hdrs = [
+        "object_detector.h",
+    ],
+    tflite_deps = [
+        "//tensorflow_lite_support/cc/task/vision:object_detector",
+    ],
+    deps = [
+        "//tensorflow_lite_support/c:common",
+        "//tensorflow_lite_support/c/task/core:base_options",
+        "//tensorflow_lite_support/c/task/processor:classification_options",
+        "//tensorflow_lite_support/c/task/processor:detection_result",
+        "//tensorflow_lite_support/c/task/vision/core:frame_buffer",
+    ],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/core/BUILD

@@ -0,0 +1,13 @@
+package(
+    default_visibility = [
+        "//tensorflow_lite_support:internal",
+    ],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library(
+    name = "frame_buffer",
+    hdrs = [
+        "frame_buffer.h",
+    ],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/core/frame_buffer.h

@@ -0,0 +1,81 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_FRAME_BUFFER_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_FRAME_BUFFER_H_
+
+#include <stdint.h>
+
+// Defines C structs for holding the frame buffer.
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+// Colorspace formats.
+enum TfLiteFrameBufferFormat {
+  kRGBA,
+  kRGB,
+  kNV12,
+  kNV21,
+  kYV12,
+  kYV21,
+  kGRAY,
+  kUNKNOWN
+};
+
+// FrameBuffer content orientation follows EXIF specification. The name of
+// each enum value defines the position of the 0th row and the 0th column of
+// the image content. See http://jpegclub.org/exif_orientation.html for
+// details.
+enum TfLiteFrameBufferOrientation {
+  kTopLeft = 1,
+  kTopRight = 2,
+  kBottomRight = 3,
+  kBottomLeft = 4,
+  kLeftTop = 5,
+  kRightTop = 6,
+  kRightBottom = 7,
+  kLeftBottom = 8
+};
+
+// Dimension information for the whole frame.
+struct TfLiteFrameBufferDimension {
+  // The width dimension in pixel unit.
+  int width;
+  // The height dimension in pixel unit.
+  int height;
+};
+
+// A `FrameBuffer` provides a view into the provided backing buffer (e.g. camera
+// frame or still image) with buffer format information. FrameBuffer doesn't
+// take ownership of the provided backing buffer. The caller is responsible to
+// manage the backing buffer lifecycle for the lifetime of the FrameBuffer.
+typedef struct TfLiteFrameBuffer {
+  // Colorspace format of the frame buffer.
+  enum TfLiteFrameBufferFormat format;
+  // Orientation of the frame buffer.
+  enum TfLiteFrameBufferOrientation orientation;
+  // Dimension information for the whole frame.
+  struct TfLiteFrameBufferDimension dimension;
+  // Holds the backing buffer for the frame buffer. Only single planar images
+  // are supported as of now.
+  uint8_t* buffer;
+} TfLiteFrameBuffer;
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_FRAME_BUFFER_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/image_classifier.cc

@@ -0,0 +1,236 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow_lite_support/c/task/vision/image_classifier.h"
+
+#include <memory>
+
+#include "tensorflow_lite_support/c/common_utils.h"
+#include "tensorflow_lite_support/c/task/vision/utils/frame_buffer_cpp_c_utils.h"
+#include "tensorflow_lite_support/cc/task/vision/image_classifier.h"
+#include "tensorflow_lite_support/cc/task/vision/proto/classifications_proto_inc.h"
+#include "tensorflow_lite_support/cc/task/vision/proto/image_classifier_options_proto_inc.h"
+
+namespace {
+using ::tflite::support::StatusOr;
+using ClassificationResultCpp = ::tflite::task::vision::ClassificationResult;
+using ClassificationsCpp = ::tflite::task::vision::Classifications;
+using ClassCpp = ::tflite::task::vision::Class;
+using BoundingBoxCpp = ::tflite::task::vision::BoundingBox;
+using ImageClassifierCpp = ::tflite::task::vision::ImageClassifier;
+using ImageClassifierOptionsCpp =
+    ::tflite::task::vision::ImageClassifierOptions;
+using FrameBufferCpp = ::tflite::task::vision::FrameBuffer;
+using ::tflite::support::TfLiteSupportStatus;
+
+StatusOr<ImageClassifierOptionsCpp> CreateImageClassifierCppOptionsFromCOptions(
+    const TfLiteImageClassifierOptions* c_options) {
+  if (c_options == nullptr) {
+    return CreateStatusWithPayload(
+        absl::StatusCode::kInvalidArgument,
+        absl::StrFormat("Expected non null options."),
+        TfLiteSupportStatus::kInvalidArgumentError);
+  }
+
+  ImageClassifierOptionsCpp cpp_options = {};
+
+  // More file sources can be added in else ifs
+  if (c_options->base_options.model_file.file_path)
+    cpp_options.mutable_base_options()->mutable_model_file()->set_file_name(
+        c_options->base_options.model_file.file_path);
+
+  // c_options->base_options.compute_settings.num_threads is expected to be
+  // set to value > 0 or -1. Otherwise invoking
+  // ImageClassifierCpp::CreateFromOptions() results in a not ok status.
+  cpp_options.mutable_base_options()
+      ->mutable_compute_settings()
+      ->mutable_tflite_settings()
+      ->mutable_cpu_settings()
+      ->set_num_threads(
+          c_options->base_options.compute_settings.cpu_settings.num_threads);
+
+  for (int i = 0; i < c_options->classification_options.label_denylist.length;
+       i++)
+    cpp_options.add_class_name_blacklist(
+        c_options->classification_options.label_denylist.list[i]);
+
+  for (int i = 0; i < c_options->classification_options.label_allowlist.length;
+       i++)
+    cpp_options.add_class_name_whitelist(
+        c_options->classification_options.label_allowlist.list[i]);
+
+  // Check needed since setting a nullptr for this field results in a segfault
+  // on invocation of ImageClassifierCpp::CreateFromOptions().
+  if (c_options->classification_options.display_names_local) {
+    cpp_options.set_display_names_locale(
+        c_options->classification_options.display_names_local);
+  }
+
+  // c_options->classification_options.max_results is expected to be set to -1
+  // or any value > 0. Otherwise invoking
+  // ImageClassifierCpp::CreateFromOptions() results in a not ok status.
+  cpp_options.set_max_results(c_options->classification_options.max_results);
+
+  cpp_options.set_score_threshold(
+      c_options->classification_options.score_threshold);
+
+  return cpp_options;
+}
+}  // namespace
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+struct TfLiteImageClassifier {
+  std::unique_ptr<ImageClassifierCpp> impl;
+};
+
+TfLiteImageClassifierOptions TfLiteImageClassifierOptionsCreate() {
+  // Use brace-enclosed initializer list will break the Kokoro test.
+  TfLiteImageClassifierOptions options = {{{0}}};
+  options.classification_options.max_results = -1;
+  options.classification_options.score_threshold = 0.0;
+  options.base_options.compute_settings.cpu_settings.num_threads = -1;
+  return options;
+}
+
+TfLiteImageClassifier* TfLiteImageClassifierFromOptions(
+    const TfLiteImageClassifierOptions* options,
+    TfLiteSupportError** error) {
+  StatusOr<ImageClassifierOptionsCpp> cpp_option_status =
+      CreateImageClassifierCppOptionsFromCOptions(options);
+
+  if (!cpp_option_status.ok()) {
+    ::tflite::support::CreateTfLiteSupportErrorWithStatus(
+        cpp_option_status.status(), error);
+    return nullptr;
+  }
+
+  StatusOr<std::unique_ptr<ImageClassifierCpp>> classifier_status =
+      ImageClassifierCpp::CreateFromOptions(cpp_option_status.value());
+
+  if (classifier_status.ok()) {
+    return new TfLiteImageClassifier{.impl =
+                                         std::move(classifier_status.value())};
+  } else {
+    ::tflite::support::CreateTfLiteSupportErrorWithStatus(
+        classifier_status.status(), error);
+    return nullptr;
+  }
+}
+
+TfLiteClassificationResult* GetClassificationResultCStruct(
+    const ClassificationResultCpp& classification_result_cpp) {
+  auto c_classifications =
+      new TfLiteClassifications[classification_result_cpp
+                                    .classifications_size()];
+
+  for (int head = 0; head < classification_result_cpp.classifications_size();
+       ++head) {
+    const ClassificationsCpp& classifications =
+        classification_result_cpp.classifications(head);
+    c_classifications[head].head_index = head;
+
+    auto c_categories = new TfLiteCategory[classifications.classes_size()];
+    c_classifications->size = classifications.classes_size();
+
+    for (int rank = 0; rank < classifications.classes_size(); ++rank) {
+      const ClassCpp& classification = classifications.classes(rank);
+      c_categories[rank].index = classification.index();
+      c_categories[rank].score = classification.score();
+
+      if (classification.has_class_name())
+        c_categories[rank].label = strdup(classification.class_name().c_str());
+      else
+        c_categories[rank].label = nullptr;
+
+      if (classification.has_display_name())
+        c_categories[rank].display_name =
+            strdup(classification.display_name().c_str());
+      else
+        c_categories[rank].display_name = nullptr;
+    }
+    c_classifications[head].categories = c_categories;
+  }
+
+  auto c_classification_result = new TfLiteClassificationResult;
+  c_classification_result->classifications = c_classifications;
+  c_classification_result->size =
+      classification_result_cpp.classifications_size();
+
+  return c_classification_result;
+}
+
+TfLiteClassificationResult* TfLiteImageClassifierClassifyWithRoi(
+    const TfLiteImageClassifier* classifier,
+    const TfLiteFrameBuffer* frame_buffer,
+    const TfLiteBoundingBox* roi,
+    TfLiteSupportError** error) {
+  if (classifier == nullptr) {
+    tflite::support::CreateTfLiteSupportError(
+        kInvalidArgumentError, "Expected non null image classifier.", error);
+    return nullptr;
+  }
+
+  StatusOr<std::unique_ptr<FrameBufferCpp>> cpp_frame_buffer_status =
+      ::tflite::task::vision::CreateCppFrameBuffer(frame_buffer);
+  if (!cpp_frame_buffer_status.ok()) {
+    tflite::support::CreateTfLiteSupportErrorWithStatus(
+        cpp_frame_buffer_status.status(), error);
+    return nullptr;
+  }
+
+  BoundingBoxCpp cc_roi;
+  if (roi == nullptr) {
+    cc_roi.set_width(frame_buffer->dimension.width);
+    cc_roi.set_height(frame_buffer->dimension.height);
+  } else {
+    cc_roi.set_origin_x(roi->origin_x);
+    cc_roi.set_origin_y(roi->origin_y);
+    cc_roi.set_width(roi->width);
+    cc_roi.set_height(roi->height);
+  }
+
+  // fnc_sample(cpp_frame_buffer_status);
+  StatusOr<ClassificationResultCpp> cpp_classification_result_status =
+      classifier->impl->Classify(*std::move(cpp_frame_buffer_status.value()),
+                                 cc_roi);
+
+  if (!cpp_classification_result_status.ok()) {
+    tflite::support::CreateTfLiteSupportErrorWithStatus(
+        cpp_classification_result_status.status(), error);
+    return nullptr;
+  }
+
+  return GetClassificationResultCStruct(
+      cpp_classification_result_status.value());
+}
+
+TfLiteClassificationResult* TfLiteImageClassifierClassify(
+    const TfLiteImageClassifier* classifier,
+    const TfLiteFrameBuffer* frame_buffer,
+    TfLiteSupportError** error) {
+  return TfLiteImageClassifierClassifyWithRoi(classifier, frame_buffer, nullptr,
+                                              error);
+}
+
+void TfLiteImageClassifierDelete(TfLiteImageClassifier* classifier) {
+  delete classifier;
+}
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/image_classifier.h

@@ -0,0 +1,214 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_IMAGE_CLASSIFIER_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_IMAGE_CLASSIFIER_H_
+
+#include <stdint.h>
+
+#include "tensorflow_lite_support/c/common.h"
+#include "tensorflow_lite_support/c/task/core/base_options.h"
+#include "tensorflow_lite_support/c/task/processor/bounding_box.h"
+#include "tensorflow_lite_support/c/task/processor/classification_options.h"
+#include "tensorflow_lite_support/c/task/processor/classification_result.h"
+#include "tensorflow_lite_support/c/task/vision/core/frame_buffer.h"
+
+// --------------------------------------------------------------------------
+/// C API for ImageClassifiier.
+///
+/// The API leans towards simplicity and uniformity instead of convenience, as
+/// most usage will be by language-specific wrappers. It provides largely the
+/// same set of functionality as that of the C++ TensorFlow Lite
+/// `ImageClassifier` API, but is useful for shared libraries where having
+/// a stable ABI boundary is important.
+///
+/// Usage:
+/// <pre><code>
+/// // Create the model
+/// Using the options initialized with default values returned by
+/// TfLiteImageClassifierOptionsCreate() makes sure that there will be no
+/// undefined behaviour due to garbage values in unitialized members.
+/// TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+///
+/// Set the model file path in options
+///   options.base_options.model_file.file_path = "/path/to/model.tflite";
+///
+/// If need be, set values for any options to customize behaviour.
+/// options.base_options.compute_settings.cpu_settings.num_threads = 3
+///
+/// Create TfLiteImageClassifier using the options:
+/// If error information is not nedded in case of failure:
+/// TfLiteImageClassifier* image_classifier =
+///       TfLiteImageClassifierFromOptions(&options, NULL);
+///
+/// If error information is nedded in case of failure:
+/// TfLiteSupportError* create_error = NULL;
+/// TfLiteImageClassifier* image_classifier =
+///       TfLiteImageClassifierFromOptions(&options, &create_error);
+///
+/// if (!image_classifier) {
+///   Handle failure.
+///   Do something with `create_error`, if requested as illustrated above.
+/// }
+///
+/// Dispose of the create_error object.
+/// TfLiteSupportErrorDelete(create_error);
+///
+/// Classify an image
+/// TfLiteFrameBuffer frame_buffer = { Initialize with image data }
+///
+/// If error information is not nedded in case of failure:
+/// TfLiteClassificationResult* classification_result =
+///       TfLiteImageClassifierClassify(image_classifier, &frame_buffer, NULL);
+///
+/// If error information is nedded in case of failure:
+/// TfLiteSupportError* classify_error = NULL;
+/// TfLiteClassificationResult* classification_result =
+///       TfLiteImageClassifierClassify(image_classifier, &frame_buffer,
+///       &classify_error);
+///
+/// if (!classification_result) {
+///   Handle failure.
+///   Do something with `classify_error`, if requested as illustrated above.
+/// }
+///
+/// Dispose of the classify_error object.
+/// TfLiteSupportErrorDelete(classify_error);
+///
+/// Dispose of the API object.
+/// TfLiteImageClassifierDelete(image_classifier);
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+typedef struct TfLiteImageClassifier TfLiteImageClassifier;
+
+typedef struct TfLiteImageClassifierOptions {
+  TfLiteClassificationOptions classification_options;
+  TfLiteBaseOptions base_options;
+} TfLiteImageClassifierOptions;
+
+// Creates and returns TfLiteImageClassifierOptions initialized with default
+// values. Default values are as follows:
+// 1. .classification_options.max_results = -1, which returns all classification
+// categories by default.
+// 2. .base_options.compute_settings.tflite_settings.cpu_settings.num_threads =
+// -1, which makes the TFLite runtime choose the value.
+// 3. .classification_options.score_threshold = 0
+// 4. All pointers like .base_options.model_file.file_path,
+// .base_options.classification_options.display_names_local,
+// .classification_options.label_allowlist.list,
+// options.classification_options.label_denylist.list are NULL.
+// 5. All other integer values are initialized to 0.
+TfLiteImageClassifierOptions TfLiteImageClassifierOptionsCreate();
+
+// Creates TfLiteImageClassifier from options.
+// .base_options.model_file.file_path in TfLiteImageClassifierOptions should be
+// set to the path of the tflite model you wish to create the
+// TfLiteImageClassifier with.
+// Create TfLiteImageClassifierOptions using
+// TfLiteImageClassifierOptionsCreate(). If need be, you can change the default
+// values of options for customizing classification, If options are not created
+// in the aforementioned way, you have to make sure that all members are
+// initialized to respective default values and all pointer members are zero
+// initialized to avoid any undefined behaviour.
+//
+// Returns the created image classifier in case of success.
+// Returns nullptr on failure which happens commonly due to one of the following
+// issues:
+// 1. file doesn't exist or is not a well formatted.
+// 2. options is nullptr.
+// 3. Both options.classification_options.label_denylist and
+// options.classification_options.label_allowlist are non empty. These
+// fields are mutually exclusive.
+//
+// The caller can check if an error was encountered by testing if the returned
+// value of the function is null. If the caller doesn't want the reason for
+// failure, they can simply pass a NULL for the address of the error pointer as
+// shown below:
+//
+// TfLiteImageClassifier* classifier = TfLiteImageClassifierFromOptions(options,
+// NULL);
+//
+// If the caller wants to be informed of the reason for failure, they must pass
+// the adress of a pointer of type TfLiteSupportError to the `error` param as
+// shown below:
+//
+// TfLiteSupport *error = NULL:
+// TfLiteImageClassifier* classifier = TfLiteImageClassifierFromOptions(options,
+// &error);
+//
+// In case of unsuccessful execution, Once the function returns, the error
+// pointer will point to a struct containing the error information. If error
+// info is passed back to the caller, it is the responsibility of the caller to
+// free the error struct by calling the following method defined in common.h:
+//
+// TfLiteSupportErrorDelete(error)
+//
+TfLiteImageClassifier* TfLiteImageClassifierFromOptions(
+    const TfLiteImageClassifierOptions* options,
+    TfLiteSupportError** error);
+
+// Invokes the encapsulated TFLite model and classifies the frame_buffer.
+// Returns a pointer to the created classification result in case of success or
+// NULL in case of failure. The caller must test the return value to identify
+// success or failure. If the caller doesn't want the reason for failure, they
+// can simply pass a NULL for the address of the error pointer as shown below:
+//
+// TfLiteClassificationResult* classification_result =
+// TfLiteImageClassifierClassify(&options, NULL);
+//
+// If the caller wants to be informed of the reason for failure, they must pass
+// the adress of a pointer of type TfLiteSupportError to the `error` param as
+// shown below:
+//
+// TfLiteSupport *error = NULL:
+// TfLiteImageClassifier* classifier = TfLiteImageClassifierFromOptions(options,
+// &error);
+//
+// In case of unsuccessful execution, Once the function returns, the error
+// pointer will point to a struct containing the error information. If error
+// info is passed back to the caller, it is the responsibility of the caller to
+// free the error struct by calling the following method defined in common.h:
+//
+// TfLiteSupportErrorDelete(error)
+//
+TfLiteClassificationResult* TfLiteImageClassifierClassify(
+    const TfLiteImageClassifier* classifier,
+    const TfLiteFrameBuffer* frame_buffer,
+    TfLiteSupportError** error);
+
+// Invokes the encapsulated TFLite model and classifies the region of the
+// frame_buffer specified by the bounding box. Same as TfLiteImageClassifier*
+// TfLiteImageClassifierFromOptions(
+//    const TfLiteImageClassifierOptions* options, TfLiteSupportError** error),
+//    except that the
+// classification is performed based on the input region of interest. Cropping
+// according to this region of interest is prepended to the pre-processing
+// operations.
+TfLiteClassificationResult* TfLiteImageClassifierClassifyWithRoi(
+    const TfLiteImageClassifier* classifier,
+    const TfLiteFrameBuffer* frame_buffer,
+    const TfLiteBoundingBox* roi,
+    TfLiteSupportError** error);
+
+// Disposes off the image classifier.
+void TfLiteImageClassifierDelete(TfLiteImageClassifier* classifier);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_IMAGE_CLASSIFIER_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/object_detector.h

@@ -0,0 +1,200 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_IMAGE_CLASSIFIER_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_IMAGE_CLASSIFIER_H_
+
+#include <stdint.h>
+
+#include "tensorflow_lite_support/c/common.h"
+#include "tensorflow_lite_support/c/task/core/base_options.h"
+#include "tensorflow_lite_support/c/task/processor/classification_options.h"
+#include "tensorflow_lite_support/c/task/processor/detection_result.h"
+#include "tensorflow_lite_support/c/task/vision/core/frame_buffer.h"
+
+// --------------------------------------------------------------------------
+/// C API for Object Detector.
+///
+/// The API leans towards simplicity and uniformity instead of convenience, as
+/// most usage will be by language-specific wrappers. It provides largely the
+/// same set of functionality as that of the C++ TensorFlow Lite
+/// `ObjectDetector` API, but is useful for shared libraries where having
+/// a stable ABI boundary is important.
+///
+/// Usage:
+/// <pre><code>
+/// // Create the model
+/// Using the options initialized with default values returned by
+/// TfLiteObjectDetectorOptionsCreate() makes sure that there will be no
+/// undefined behaviour due to garbage values in unitialized members.
+/// TfLiteObjectDetectorOptions options = TfLiteObjectDetectorOptionsCreate();
+///
+/// Set the model file path in options
+///   options.base_options.model_file.file_path = "/path/to/model.tflite";
+///
+/// If need be, set values for any options to customize behaviour.
+/// options.base_options.compute_settings.cpu_settings.num_threads = 3
+///
+/// Create TfLiteObjectDetector using the options:
+/// If error information is not nedded in case of failure:
+/// TfLiteObjectDetector* object_detector =
+///       TfLiteObjectDetectorFromOptions(&options, NULL);
+///
+/// If error information is nedded in case of failure:
+/// TfLiteSupportError* create_error = NULL;
+/// TfLiteObjectDetector* object_detector =
+///       TfLiteObjectDetectorFromOptions(&options, &create_error);
+///
+/// if (!object_detector) {
+///   Handle failure.
+///   Do something with `create_error`, if requested as illustrated above.
+/// }
+///
+/// Dispose of the create_error object.
+/// TfLiteSupportErrorDelete(create_error);
+///
+/// Classify an image
+/// TfLiteFrameBuffer frame_buffer = { Initialize with image data }
+///
+/// If error information is not nedded in case of failure:
+/// TfLiteDetectionResult* detection_result =
+///       TfLiteObjectDetectorClassify(object_detector, &frame_buffer, NULL);
+///
+/// If error information is needed in case of failure:
+/// TfLiteSupportError* detect_error = NULL;
+/// TfLiteDetectionResult* detection_result =
+///       TfLiteObjectDetectorDetect(object_detector, &frame_buffer,
+///       &detect_error);
+///
+/// if (!detection_result) {
+///   Handle failure.
+///   Do something with `detection_error`, if requested as illustrated above.
+/// }
+///
+/// Dispose of the detection_error object.
+/// TfLiteSupportErrorDelete(detection_error);
+///
+/// Dispose of the API object.
+/// TfLiteObjectDetectorOptionsDelete(object_detector);
+
+#ifdef __cplusplus
+extern "C" {
+#endif  // __cplusplus
+
+typedef struct TfLiteObjectDetector TfLiteObjectDetector;
+
+typedef struct TfLiteObjectDetectorOptions {
+  TfLiteClassificationOptions classification_options;
+  TfLiteBaseOptions base_options;
+} TfLiteObjectDetectorOptions;
+
+// Creates and returns TfLiteObjectDetectorOptions initialized with default
+// values. Default values are as follows:
+// 1. .classification_options.max_results = -1, which returns all classification
+// categories by default.
+// 2. .base_options.compute_settings.tflite_settings.cpu_settings.num_threads =
+// -1, which makes the TFLite runtime choose the value.
+// 3. .classification_options.score_threshold = 0
+// 4. All pointers like .base_options.model_file.file_path,
+// .base_options.classification_options.display_names_local,
+// .classification_options.label_allowlist.list,
+// options.classification_options.label_denylist.list are NULL.
+// 5. All other integer values are initialized to 0.
+TfLiteObjectDetectorOptions TfLiteObjectDetectorOptionsCreate();
+
+// Creates TfLiteObjectDetector from options.
+// .base_options.model_file.file_path in TfLiteObjectDetectorOptions should be
+// set to the path of the tflite model you wish to create the
+// TfLiteObjectDetector with.
+// Create TfLiteObjectDetectorOptions using
+// TfLiteObjectDetectorOptionsCreate(). If need be, you can change the default
+// values of options for customizing classification, If options are not created
+// in the aforementioned way, you have to make sure that all members are
+// initialized to respective default values and all pointer members are zero
+// initialized to avoid any undefined behaviour.
+//
+// Returns the created object detector in case of success.
+// Returns nullptr on failure which happens commonly due to one of the following
+// issues:
+// 1. file doesn't exist or is not a well formatted.
+// 2. options is nullptr.
+// 3. Both options.classification_options.label_denylist and
+// options.classification_options.label_allowlist are non empty. These
+// fields are mutually exclusive.
+//
+// The caller can check if an error was encountered by testing if the returned
+// value of the function is null. If the caller doesn't want the reason for
+// failure, they can simply pass a NULL for the address of the error pointer as
+// shown below:
+//
+// TfLiteObjectDetector* detector = TfLiteObjectDetectorFromOptions(options,
+// NULL);
+//
+// If the caller wants to be informed of the reason for failure, they must pass
+// the adress of a pointer of type TfLiteSupportError to the `error` param as
+// shown below:
+//
+// TfLiteSupport *error = NULL:
+// TfLiteObjectDetector* classifier = TfLiteObjectDetectorFromOptions(options,
+// &error);
+//
+// In case of unsuccessful execution, Once the function returns, the error
+// pointer will point to a struct containing the error information. If error
+// info is passed back to the caller, it is the responsibility of the caller to
+// free the error struct by calling the following method defined in common.h:
+//
+// TfLiteSupportErrorDelete(error)
+//
+TfLiteObjectDetector* TfLiteObjectDetectorFromOptions(
+    const TfLiteObjectDetectorOptions* options,
+    TfLiteSupportError** error);
+
+// Invokes the encapsulated TFLite model and performs object detection on the
+// frame_buffer. Returns a pointer to the created object detection result result
+// in case of success or NULL in case of failure. The caller must test the
+// return value to identify success or failure. If the caller doesn't want the
+// reason for failure, they can simply pass a NULL for the address of the error
+// pointer as shown below:
+//
+// TfLiteDetectionResult* detection_result =
+// TfLiteObjectDetectorDetect(&options, NULL);
+//
+// If the caller wants to be informed of the reason for failure, they must pass
+// the adress of a pointer of type TfLiteSupportError to the `error` param as
+// shown below:
+//
+// TfLiteSupport *error = NULL:
+// TfLiteObjectDetector* detector = TfLiteObjectDetectorFromOptions(options,
+// &error);
+//
+// In case of unsuccessful execution, Once the function returns, the error
+// pointer will point to a struct containing the error information. If error
+// info is passed back to the caller, it is the responsibility of the caller to
+// free the error struct by calling the following method defined in common.h:
+//
+// TfLiteSupportErrorDelete(error)
+//
+TfLiteDetectionResult* TfLiteObjectDetectorDetect(
+    const TfLiteObjectDetector* detector,
+    const TfLiteFrameBuffer* frame_buffer,
+    TfLiteSupportError** error);
+
+// Disposes off the object detector.
+void TfLiteObjectDetectorDelete(TfLiteObjectDetector* detector);
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif  // __cplusplus
+
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_IMAGE_CLASSIFIER_H_

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/utils/BUILD

@@ -0,0 +1,22 @@
+package(
+    default_visibility = [
+        "//tensorflow_lite_support:internal",
+    ],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library(
+    name = "frame_buffer_cpp_c_utils",
+    srcs = [
+        "frame_buffer_cpp_c_utils.cc",
+    ],
+    hdrs = [
+        "frame_buffer_cpp_c_utils.h",
+    ],
+    deps = [
+        "//tensorflow_lite_support/c/task/vision/core:frame_buffer",
+        "//tensorflow_lite_support/cc:common",
+        "//tensorflow_lite_support/cc/task/vision/utils:frame_buffer_common_utils",
+        "@com_google_absl//absl/strings:str_format",
+    ],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/utils/frame_buffer_cpp_c_utils.cc

@@ -0,0 +1,50 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow_lite_support/c/task/vision/utils/frame_buffer_cpp_c_utils.h"
+
+#include "absl/strings/str_format.h"  // from @com_google_absl
+#include "tensorflow_lite_support/cc/common.h"
+
+namespace tflite {
+namespace task {
+namespace vision {
+
+namespace {
+using FrameBufferCpp = ::tflite::task::vision::FrameBuffer;
+using ::tflite::support::StatusOr;
+using ::tflite::support::TfLiteSupportStatus;
+}  // namespace
+
+StatusOr<std::unique_ptr<FrameBufferCpp>> CreateCppFrameBuffer(
+    const TfLiteFrameBuffer* frame_buffer) {
+  if (frame_buffer == nullptr)
+    return CreateStatusWithPayload(
+        absl::StatusCode::kInvalidArgument,
+        absl::StrFormat("Expected non null frame buffer."),
+        TfLiteSupportStatus::kInvalidArgumentError);
+
+  FrameBufferCpp::Format frame_buffer_format =
+      FrameBufferCpp::Format(frame_buffer->format);
+
+  return CreateFromRawBuffer(
+      frame_buffer->buffer,
+      {frame_buffer->dimension.width, frame_buffer->dimension.height},
+      frame_buffer_format);
+}
+
+}  // namespace vision
+}  // namespace task
+}  // namespace tflite

third_party/tflite_support/src/tensorflow_lite_support/c/task/vision/utils/frame_buffer_cpp_c_utils.h

@@ -0,0 +1,36 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+#ifndef TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_FRAME_BUFFER_CPP_C_UTILS_H_
+#define TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_FRAME_BUFFER_CPP_C_UTILS_H_
+
+#include "tensorflow_lite_support/c/task/vision/core/frame_buffer.h"
+#include "tensorflow_lite_support/cc/task/vision/utils/frame_buffer_common_utils.h"
+
+// Utils for Conversions between C and C++ FrameBuffer
+// -----------------------------------------------------------------
+// Meant to be used with vision C apis.
+
+// Creates the C++ FrameBuffer from the C FrameBuffer
+namespace tflite {
+namespace task {
+namespace vision {
+
+tflite::support::StatusOr<std::unique_ptr<tflite::task::vision::FrameBuffer>>
+CreateCppFrameBuffer(const TfLiteFrameBuffer* frame_buffer);
+
+}  // namespace vision
+}  // namespace task
+}  // namespace tflite
+#endif  // TENSORFLOW_LITE_SUPPORT_C_TASK_VISION_FRAME_BUFFER_CPP_C_UTILS_H_

third_party/tflite_support/src/tensorflow_lite_support/c/test/task/vision/BUILD

@@ -0,0 +1,34 @@
+load(
+    "@org_tensorflow//tensorflow/lite/core/shims:cc_library_with_tflite.bzl",
+    "cc_test_with_tflite",
+)
+
+package(
+    default_visibility = [
+        "//visibility:private",
+    ],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+# To test it with Bazel, run the following command from the terminal of your desktop:
+# bazel test tensorflow_lite_support/c/test/task/vision:image_classifier_test
+cc_test_with_tflite(
+    name = "image_classifier_test",
+    srcs = ["image_classifier_test.cc"],
+    data = [
+        "//tensorflow_lite_support/cc/test/testdata/task/vision:test_images",
+        "//tensorflow_lite_support/cc/test/testdata/task/vision:test_models",
+    ],
+    tflite_deps = [
+        "//tensorflow_lite_support/c/task/vision:image_classifier",
+        "@org_tensorflow//tensorflow/lite/core/shims:cc_shims_test_util",
+    ],
+    deps = [
+        "//tensorflow_lite_support/c:common",
+        "//tensorflow_lite_support/c/task/processor:classification_result",
+        "//tensorflow_lite_support/c/task/vision/core:frame_buffer",
+        "//tensorflow_lite_support/cc/port:gtest_main",
+        "//tensorflow_lite_support/cc/test:test_utils",
+        "//tensorflow_lite_support/examples/task/vision/desktop/utils:image_utils",
+    ],
+)

third_party/tflite_support/src/tensorflow_lite_support/c/test/task/vision/image_classifier_test.cc

@@ -0,0 +1,432 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "tensorflow_lite_support/c/task/vision/image_classifier.h"
+
+#include <string.h>
+
+#include "tensorflow/lite/core/shims/cc/shims_test_util.h"
+#include "tensorflow_lite_support/c/common.h"
+#include "tensorflow_lite_support/c/task/processor/classification_result.h"
+#include "tensorflow_lite_support/c/task/vision/core/frame_buffer.h"
+#include "tensorflow_lite_support/cc/port/gmock.h"
+#include "tensorflow_lite_support/cc/port/gtest.h"
+#include "tensorflow_lite_support/cc/port/status_matchers.h"
+#include "tensorflow_lite_support/cc/test/test_utils.h"
+#include "tensorflow_lite_support/examples/task/vision/desktop/utils/image_utils.h"
+
+namespace tflite {
+namespace task {
+namespace vision {
+namespace {
+
+using ::testing::HasSubstr;
+using ::tflite::support::StatusOr;
+using ::tflite::task::JoinPath;
+
+constexpr char kTestDataDirectory[] =
+    "/tensorflow_lite_support/cc/test/testdata/task/"
+    "vision/";
+// Quantized model.
+constexpr char kMobileNetQuantizedWithMetadata[] =
+    "mobilenet_v1_0.25_224_quant.tflite";
+
+StatusOr<ImageData> LoadImage(const char* image_name) {
+  return DecodeImageFromFile(
+      JoinPath("./" /*test src dir*/, kTestDataDirectory, image_name));
+}
+
+class ImageClassifierFromOptionsTest : public tflite_shims::testing::Test {};
+
+TEST_F(ImageClassifierFromOptionsTest, FailsWithNullOptionsAndError) {
+  TfLiteSupportError* error = nullptr;
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(nullptr, &error);
+
+  EXPECT_EQ(image_classifier, nullptr);
+  if (image_classifier)
+    TfLiteImageClassifierDelete(image_classifier);
+
+  ASSERT_NE(error, nullptr);
+  EXPECT_EQ(error->code, kInvalidArgumentError);
+  EXPECT_NE(error->message, nullptr);
+  EXPECT_THAT(error->message, HasSubstr("Expected non null options"));
+
+  TfLiteSupportErrorDelete(error);
+}
+
+TEST_F(ImageClassifierFromOptionsTest, FailsWithMissingModelPath) {
+  TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(&options, nullptr);
+  EXPECT_EQ(image_classifier, nullptr);
+  if (image_classifier)
+    TfLiteImageClassifierDelete(image_classifier);
+}
+
+TEST_F(ImageClassifierFromOptionsTest, FailsWithMissingModelPathAndError) {
+  TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+
+  TfLiteSupportError* error = nullptr;
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(&options, &error);
+
+  EXPECT_EQ(image_classifier, nullptr);
+  if (image_classifier)
+    TfLiteImageClassifierDelete(image_classifier);
+
+  ASSERT_NE(error, nullptr);
+  EXPECT_EQ(error->code, kInvalidArgumentError);
+  EXPECT_NE(error->message, nullptr);
+  EXPECT_THAT(error->message, HasSubstr("`base_options.model_file`"));
+
+  TfLiteSupportErrorDelete(error);
+}
+
+TEST_F(ImageClassifierFromOptionsTest, SucceedsWithModelPath) {
+  std::string model_path = JoinPath("./" /*test src dir*/, kTestDataDirectory,
+                                    kMobileNetQuantizedWithMetadata);
+  TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+  options.base_options.model_file.file_path = model_path.data();
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(&options, nullptr);
+
+  EXPECT_NE(image_classifier, nullptr);
+  TfLiteImageClassifierDelete(image_classifier);
+}
+
+TEST_F(ImageClassifierFromOptionsTest, SucceedsWithNumberOfThreadsAndError) {
+  std::string model_path = JoinPath("./" /*test src dir*/, kTestDataDirectory,
+                                    kMobileNetQuantizedWithMetadata);
+  TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+  options.base_options.model_file.file_path = model_path.data();
+  options.base_options.compute_settings.cpu_settings.num_threads = 3;
+
+  TfLiteSupportError* error = nullptr;
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(&options, &error);
+
+  EXPECT_NE(image_classifier, nullptr);
+  EXPECT_EQ(error, nullptr);
+
+  if (image_classifier)
+    TfLiteImageClassifierDelete(image_classifier);
+  if (error)
+    TfLiteSupportErrorDelete(error);
+}
+
+TEST_F(ImageClassifierFromOptionsTest,
+       FailsWithClassNameDenyListAndClassNameAllowListAndError) {
+  std::string model_path = JoinPath("./" /*test src dir*/, kTestDataDirectory,
+                                    kMobileNetQuantizedWithMetadata);
+
+  TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+  options.base_options.model_file.file_path = model_path.data();
+
+  char* label_denylist[9] = {(char*)"brambling"};
+  options.classification_options.label_denylist.list = label_denylist;
+  options.classification_options.label_denylist.length = 1;
+
+  char* label_allowlist[12] = {(char*)"cheeseburger"};
+  options.classification_options.label_allowlist.list = label_allowlist;
+  options.classification_options.label_allowlist.length = 1;
+
+  TfLiteSupportError* error = nullptr;
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(&options, &error);
+
+  EXPECT_EQ(image_classifier, nullptr);
+  if (image_classifier)
+    TfLiteImageClassifierDelete(image_classifier);
+
+  ASSERT_NE(error, nullptr);
+  EXPECT_EQ(error->code, kInvalidArgumentError);
+  EXPECT_NE(error->message, nullptr);
+  EXPECT_THAT(error->message, HasSubstr("mutually exclusive options"));
+
+  TfLiteSupportErrorDelete(error);
+}
+
+TEST(ImageClassifierNullClassifierClassifyTest,
+     FailsWithNullImageClassifierAndError) {
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteSupportError* error = nullptr;
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassify(nullptr, nullptr, &error);
+
+  ImageDataFree(&image_data);
+
+  EXPECT_EQ(classification_result, nullptr);
+  if (classification_result)
+    TfLiteClassificationResultDelete(classification_result);
+
+  ASSERT_NE(error, nullptr);
+  EXPECT_EQ(error->code, kInvalidArgumentError);
+  EXPECT_NE(error->message, nullptr);
+  EXPECT_THAT(error->message, HasSubstr("Expected non null image classifier"));
+
+  TfLiteSupportErrorDelete(error);
+}
+
+class ImageClassifierClassifyTest : public tflite_shims::testing::Test {
+ protected:
+  void SetUp() override {
+    std::string model_path = JoinPath("./" /*test src dir*/, kTestDataDirectory,
+                                      kMobileNetQuantizedWithMetadata);
+
+    TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+    options.base_options.model_file.file_path = model_path.data();
+    image_classifier = TfLiteImageClassifierFromOptions(&options, nullptr);
+    ASSERT_NE(image_classifier, nullptr);
+  }
+
+  void TearDown() override { TfLiteImageClassifierDelete(image_classifier); }
+  TfLiteImageClassifier* image_classifier;
+};
+
+TEST_F(ImageClassifierClassifyTest, SucceedsWithImageData) {
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteFrameBuffer frame_buffer = {
+      .format = kRGB,
+      .orientation = kTopLeft,
+      .dimension = {.width = image_data.width, .height = image_data.height},
+      .buffer = image_data.pixel_data};
+
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassify(image_classifier, &frame_buffer, nullptr);
+
+  ImageDataFree(&image_data);
+
+  ASSERT_NE(classification_result, nullptr);
+  EXPECT_GE(classification_result->size, 1);
+  EXPECT_NE(classification_result->classifications, nullptr);
+  EXPECT_GE(classification_result->classifications->size, 1);
+  EXPECT_NE(classification_result->classifications->categories, nullptr);
+  EXPECT_EQ(strcmp(classification_result->classifications->categories[0].label,
+                   "cheeseburger"),
+            0);
+  EXPECT_GE(classification_result->classifications->categories[0].score, 0.90);
+
+  TfLiteClassificationResultDelete(classification_result);
+}
+
+TEST_F(ImageClassifierClassifyTest, FailsWithNullFrameBufferAndError) {
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteSupportError* error = nullptr;
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassify(image_classifier, nullptr, &error);
+
+  ImageDataFree(&image_data);
+
+  EXPECT_EQ(classification_result, nullptr);
+  if (classification_result)
+    TfLiteClassificationResultDelete(classification_result);
+
+  ASSERT_NE(error, nullptr);
+  EXPECT_EQ(error->code, kInvalidArgumentError);
+  EXPECT_NE(error->message, nullptr);
+  EXPECT_THAT(error->message, HasSubstr("Expected non null frame buffer"));
+
+  TfLiteSupportErrorDelete(error);
+}
+
+TEST_F(ImageClassifierClassifyTest, FailsWithNullImageDataAndError) {
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteFrameBuffer frame_buffer = {.format = kRGB, .orientation = kTopLeft};
+
+  TfLiteSupportError* error = nullptr;
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassify(image_classifier, &frame_buffer, &error);
+
+  ImageDataFree(&image_data);
+
+  EXPECT_EQ(classification_result, nullptr);
+  if (classification_result)
+    TfLiteClassificationResultDelete(classification_result);
+
+  ASSERT_NE(error, nullptr);
+  EXPECT_EQ(error->code, kInvalidArgumentError);
+  EXPECT_NE(error->message, nullptr);
+  EXPECT_THAT(error->message, HasSubstr("Invalid stride information"));
+
+  TfLiteSupportErrorDelete(error);
+}
+
+TEST_F(ImageClassifierClassifyTest, SucceedsWithRoiWithinImageBounds) {
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteFrameBuffer frame_buffer = {
+      .format = kRGB,
+      .orientation = kTopLeft,
+      .dimension = {.width = image_data.width, .height = image_data.height},
+      .buffer = image_data.pixel_data};
+
+  TfLiteBoundingBox bounding_box = {
+      .origin_x = 0, .origin_y = 0, .width = 100, .height = 100};
+  TfLiteSupportError* error = nullptr;
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassifyWithRoi(image_classifier, &frame_buffer,
+                                           &bounding_box, &error);
+
+  ImageDataFree(&image_data);
+
+  ASSERT_NE(classification_result, nullptr);
+  EXPECT_GE(classification_result->size, 1);
+  EXPECT_NE(classification_result->classifications, nullptr);
+  EXPECT_GE(classification_result->classifications->size, 1);
+  EXPECT_NE(classification_result->classifications->categories, nullptr);
+  EXPECT_EQ(strcmp(classification_result->classifications->categories[0].label,
+                   "bagel"),
+            0);
+  EXPECT_GE(classification_result->classifications->categories[0].score, 0.30);
+
+  TfLiteClassificationResultDelete(classification_result);
+}
+
+TEST_F(ImageClassifierClassifyTest, FailsWithRoiOutsideImageBoundsAndError) {
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteFrameBuffer frame_buffer = {
+      .format = kRGB,
+      .orientation = kTopLeft,
+      .dimension = {.width = image_data.width, .height = image_data.height},
+      .buffer = image_data.pixel_data};
+
+  TfLiteBoundingBox bounding_box = {
+      .origin_x = 0, .origin_y = 0, .width = 250, .height = 250};
+  TfLiteSupportError* error = nullptr;
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassifyWithRoi(image_classifier, &frame_buffer,
+                                           &bounding_box, &error);
+
+  ImageDataFree(&image_data);
+
+  EXPECT_EQ(classification_result, nullptr);
+  if (classification_result)
+    TfLiteClassificationResultDelete(classification_result);
+
+  ASSERT_NE(error, nullptr);
+  EXPECT_EQ(error->code, kInvalidArgumentError);
+  EXPECT_NE(error->message, nullptr);
+  EXPECT_THAT(error->message, HasSubstr("Invalid crop coordinates"));
+
+  TfLiteSupportErrorDelete(error);
+}
+
+TEST(ImageClassifierWithUserDefinedOptionsClassifyTest,
+     SucceedsWithClassNameDenyList) {
+  char* denylisted_label_name = (char*)"cheeseburger";
+  std::string model_path = JoinPath("./" /*test src dir*/, kTestDataDirectory,
+                                    kMobileNetQuantizedWithMetadata);
+
+  TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+  options.base_options.model_file.file_path = model_path.data();
+
+  char* label_denylist[12] = {denylisted_label_name};
+  options.classification_options.label_denylist.list = label_denylist;
+  options.classification_options.label_denylist.length = 1;
+
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(&options, nullptr);
+  ASSERT_NE(image_classifier, nullptr);
+
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteFrameBuffer frame_buffer = {
+      .format = kRGB,
+      .orientation = kTopLeft,
+      .dimension = {.width = image_data.width, .height = image_data.height},
+      .buffer = image_data.pixel_data};
+
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassify(image_classifier, &frame_buffer, nullptr);
+
+  ImageDataFree(&image_data);
+  if (image_classifier)
+    TfLiteImageClassifierDelete(image_classifier);
+
+  ASSERT_NE(classification_result, nullptr);
+  EXPECT_GE(classification_result->size, 1);
+  EXPECT_NE(classification_result->classifications, nullptr);
+  EXPECT_GE(classification_result->classifications->size, 1);
+  EXPECT_NE(classification_result->classifications->categories, nullptr);
+  EXPECT_NE(strcmp(classification_result->classifications->categories[0].label,
+                   denylisted_label_name),
+            0);
+
+  TfLiteClassificationResultDelete(classification_result);
+}
+
+TEST(ImageClassifierWithUserDefinedOptionsClassifyTest,
+     SucceedsWithClassNameAllowList) {
+  char* allowlisted_label_name = (char*)"cheeseburger";
+  std::string model_path = JoinPath("./" /*test src dir*/, kTestDataDirectory,
+                                    kMobileNetQuantizedWithMetadata)
+                               .data();
+
+  TfLiteImageClassifierOptions options = TfLiteImageClassifierOptionsCreate();
+  options.base_options.model_file.file_path = model_path.data();
+
+  char* label_allowlist[12] = {allowlisted_label_name};
+  options.classification_options.label_allowlist.list = label_allowlist;
+  options.classification_options.label_allowlist.length = 1;
+
+  TfLiteImageClassifier* image_classifier =
+      TfLiteImageClassifierFromOptions(&options, nullptr);
+  ASSERT_NE(image_classifier, nullptr);
+
+  SUPPORT_ASSERT_OK_AND_ASSIGN(ImageData image_data,
+                               LoadImage("burger-224.png"));
+
+  TfLiteFrameBuffer frame_buffer = {
+      .format = kRGB,
+      .orientation = kTopLeft,
+      .dimension = {.width = image_data.width, .height = image_data.height},
+      .buffer = image_data.pixel_data};
+
+  TfLiteClassificationResult* classification_result =
+      TfLiteImageClassifierClassify(image_classifier, &frame_buffer, nullptr);
+
+  ImageDataFree(&image_data);
+  if (image_classifier)
+    TfLiteImageClassifierDelete(image_classifier);
+
+  ASSERT_NE(classification_result, nullptr);
+  EXPECT_GE(classification_result->size, 1);
+  EXPECT_NE(classification_result->classifications, nullptr);
+  EXPECT_GE(classification_result->classifications->size, 1);
+  EXPECT_NE(classification_result->classifications->categories, nullptr);
+  EXPECT_EQ(strcmp(classification_result->classifications->categories[0].label,
+                   allowlisted_label_name),
+            0);
+
+  TfLiteClassificationResultDelete(classification_result);
+}
+
+}  // namespace
+}  // namespace vision
+}  // namespace task
+}  // namespace tflite

third_party/tflite_support/src/tensorflow_lite_support/cc/BUILD

@@ -1,5 +1,5 @@
 package(
-    default_visibility = ["//tensorflow_lite_support:users"],
+    default_visibility = ["//tensorflow_lite_support:internal"],
     licenses = ["notice"],  # Apache 2.0
 )
 
@@ -9,17 +9,12 @@ cc_library(
         "common.cc",
     ],
     hdrs = ["common.h"],
+    visibility = [
+        "//tensorflow_lite_support:internal",
+    ],
     deps = [
         "@com_google_absl//absl/status",
         "@com_google_absl//absl/strings",
         "@com_google_absl//absl/strings:cord",
     ],
 )
-
-config_setting(
-    name = "tflite_use_c_api",
-    values = {
-        "copt": "-DTFLITE_USE_C_API",
-    },
-    visibility = ["//tensorflow_lite_support:__subpackages__"],
-)

third_party/tflite_support/src/tensorflow_lite_support/cc/common.cc

@@ -15,7 +15,8 @@ limitations under the License.
 
 #include "tensorflow_lite_support/cc/common.h"
 
-#include "absl/strings/str_cat.h"
+#include "absl/strings/cord.h"     // from @com_google_absl
+#include "absl/strings/str_cat.h"  // from @com_google_absl
 
 namespace tflite {
 namespace support {
@@ -25,6 +26,8 @@ absl::Status CreateStatusWithPayload(absl::StatusCode canonical_code,
                                      TfLiteSupportStatus tfls_code) {
   // NOTE: Ignores `message` if the canonical code is ok.
   absl::Status status = absl::Status(canonical_code, message);
+  // NOTE: Does nothing if the canonical code is ok.
+  status.SetPayload(kTfLiteSupportPayload, absl::Cord(absl::StrCat(tfls_code)));
   return status;
 }
 

third_party/tflite_support/src/tensorflow_lite_support/cc/common.h

@@ -16,8 +16,8 @@ limitations under the License.
 #ifndef TENSORFLOW_LITE_SUPPORT_CC_COMMON_H_
 #define TENSORFLOW_LITE_SUPPORT_CC_COMMON_H_
 
-#include "absl/status/status.h"
-#include "absl/strings/string_view.h"
+#include "absl/status/status.h"        // from @com_google_absl
+#include "absl/strings/string_view.h"  // from @com_google_absl
 
 namespace tflite {
 namespace support {
@@ -53,6 +53,12 @@ enum class TfLiteSupportStatus {
   kInvalidArgumentError = 2,
   // Invalid FlatBuffer file or buffer specified.
   kInvalidFlatBufferError = 3,
+  // Model contains a builtin op that isn't supported by the OpResolver or
+  // delegates.
+  kUnsupportedBuiltinOp = 4,
+  // Model contains a custom op that isn't supported by the OpResolver or
+  // delegates.
+  kUnsupportedCustomOp = 5,
 
   // File I/O error codes.
 

third_party/tflite_support/src/tensorflow_lite_support/cc/port/BUILD

@@ -1,5 +1,7 @@
+load("@org_tensorflow//tensorflow/lite/core/shims:cc_library_with_tflite.bzl", "cc_library_with_tflite")
+
 package(
-    default_visibility = ["//tensorflow_lite_support:users"],
+    default_visibility = ["//tensorflow_lite_support:internal"],
     licenses = ["notice"],  # Apache 2.0
 )
 
@@ -8,8 +10,9 @@ cc_library(
     hdrs = [
         "statusor.h",
     ],
+    visibility = ["//visibility:public"],
     deps = [
-        "//tensorflow_lite_support/cc/port/default:statusor",
+        "@com_google_absl//absl/status:statusor",
     ],
 )
 
@@ -18,31 +21,54 @@ cc_library(
     hdrs = [
         "status_macros.h",
     ],
+    visibility = ["//visibility:public"],
     deps = [
         "//tensorflow_lite_support/cc/port/default:status_macros",
     ],
 )
 
 cc_library(
+    name = "configuration_proto_inc",
+    hdrs = ["configuration_proto_inc.h"],
+    deps = ["@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:configuration_cc_proto"],
+)
+
+cc_library_with_tflite(
     name = "tflite_wrapper",
     hdrs = ["tflite_wrapper.h"],
     deps = ["//tensorflow_lite_support/cc/port/default:tflite_wrapper"],
 )
 
-# This is identical to the rule above, except that it gets built with
-# '-DTFLITE_USE_C_API'. This rule is used for unit tests that verify things
-# work correctly when built with TFLITE_USE_C_API defined.
 cc_library(
-    name = "tflite_wrapper_with_c_api_for_test",
+    name = "integral_types",
+    hdrs = ["integral_types.h"],
+    visibility = ["//tensorflow_lite_support:users"],
+)
+
+cc_library(
+    name = "gtest_main",
     testonly = 1,
-    hdrs = ["tflite_wrapper.h"],
+    hdrs = [
+        "benchmark.h",
+        "gmock.h",
+        "gtest.h",
+        "status_matchers.h",
+    ],
+    visibility = [
+        "//tensorflow_lite_support:internal",
+    ],
     deps = [
-        "//intelligence/mobile_acceleration/proto:allowlist_portable_proto",
-        "//intelligence/mobile_acceleration/support_library:tflite_wrapper_with_c_api_for_test",
+        "//tensorflow_lite_support/cc/port/default:status_matchers",
+        "@com_google_googletest//:gtest_main",
     ],
 )
 
 cc_library(
-    name = "integral_types",
-    hdrs = ["integral_types.h"],
+    name = "proto2",
+    hdrs = [
+        "proto2.h",
+    ],
+    deps = [
+        "@com_google_protobuf//:protobuf",
+    ],
 )

third_party/tflite_support/src/tensorflow_lite_support/cc/port/build_defs.bzl

@@ -1,30 +1,22 @@
 """.bzl file for TFLite Support open source build configs."""
 
-load("@com_google_protobuf//:protobuf.bzl", "cc_proto_library")
-
 def provided_args(**kwargs):
     """Returns the keyword arguments omitting None arguments."""
     return {k: v for k, v in kwargs.items() if v != None}
 
-def support_cc_proto_library(name, srcs, visibility = None, deps = [], cc_deps = [], testonly = 0):
-    """Generate cc_proto_library for TFLite Support open source version.
+def support_cc_proto_library(name, deps = [], visibility = None):
+    """Generates cc_proto_library for TFLite Support open source version.
 
       Args:
         name: the name of the cc_proto_library.
-        srcs: the .proto files of the cc_proto_library for Bazel use.
+        deps: a list of dependency labels for Bazel use; must be proto_library.
         visibility: visibility of this target.
-        deps: a list of dependency labels for Bazel use; must be cc_proto_library.
-        testonly: test only proto or not.
     """
-    _ignore = [deps]
-    cc_proto_library(**provided_args(
+
+    # Verified in the external path.
+    # buildifier: disable=native-cc-proto
+    native.cc_proto_library(**provided_args(
         name = name,
-        srcs = srcs,
         visibility = visibility,
-        deps = cc_deps,
-        testonly = testonly,
-        cc_libs = ["@com_google_protobuf//:protobuf"],
-        protoc = "@com_google_protobuf//:protoc",
-        default_runtime = "@com_google_protobuf//:protobuf",
-        alwayslink = 1,
+        deps = deps,
     ))

third_party/tflite_support/src/tensorflow_lite_support/cc/port/configuration_proto_inc.h

@@ -0,0 +1,21 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#ifndef THIRD_PARTY_TENSORFLOW_LITE_SUPPORT_CC_PORT_CONFIGURATION_PROTO_INC_H_
+#define THIRD_PARTY_TENSORFLOW_LITE_SUPPORT_CC_PORT_CONFIGURATION_PROTO_INC_H_
+
+#include "tensorflow/lite/experimental/acceleration/configuration/configuration.pb.h"
+
+#endif  // THIRD_PARTY_TENSORFLOW_LITE_SUPPORT_CC_PORT_CONFIGURATION_PROTO_INC_H_

third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/BUILD

@@ -1,29 +1,11 @@
 package(
     default_visibility = [
-        "//tensorflow_lite_support/cc/port:__pkg__",
-        "//tensorflow_lite_support/cc/test:__pkg__",
+        "//tensorflow_lite_support/cc/port:__subpackages__",
+        "//tensorflow_lite_support/cc/test:__subpackages__",
     ],
     licenses = ["notice"],  # Apache 2.0
 )
 
-cc_library(
-    name = "statusor",
-    srcs = ["statusor.cc"],
-    hdrs = [
-        "statusor.h",
-        "statusor_internals.h",
-    ],
-    deps = [
-        "@com_google_absl//absl/base:core_headers",
-        "@com_google_absl//absl/meta:type_traits",
-        "@com_google_absl//absl/status",
-        "@com_google_absl//absl/strings",
-        "@com_google_absl//absl/types:variant",
-        "@com_google_absl//absl/utility",
-        "@com_google_glog//:glog",
-    ],
-)
-
 cc_library(
     name = "status_macros",
     hdrs = [
@@ -35,6 +17,15 @@ cc_library(
     ],
 )
 
+cc_library(
+    name = "status_matchers",
+    testonly = 1,
+    hdrs = ["status_matchers.h"],
+    deps = [
+        "@com_google_googletest//:gtest_main",
+    ],
+)
+
 cc_library(
     name = "tflite_wrapper",
     srcs = ["tflite_wrapper.cc"],
@@ -42,9 +33,32 @@ cc_library(
         "tflite_wrapper.h",
     ],
     deps = [
-        "//tensorflow_lite_support/cc/port:status_macros",
         "@com_google_absl//absl/status",
+        "@com_google_absl//absl/strings:str_format",
+        "@com_google_absl//absl/synchronization",
+        "@flatbuffers",
         "@org_tensorflow//tensorflow/lite:framework",
+        "@org_tensorflow//tensorflow/lite:minimal_logging",
+        "@org_tensorflow//tensorflow/lite/c:common",
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:nnapi_plugin",
+        "@org_tensorflow//tensorflow/lite/delegates:interpreter_utils",
         "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:configuration_cc_proto",
-    ],
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:delegate_registry",
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:flatbuffer_to_proto",
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/configuration:proto_to_flatbuffer",
+        "@org_tensorflow//tensorflow/lite/experimental/acceleration/mini_benchmark",
+        "//tensorflow_lite_support/cc/port:status_macros",
+    ] + select({
+        # We only intend to use TFLite mini-benchmark on arm-based Andorid and x86_64 Linux.
+        "@org_tensorflow//tensorflow:android_arm": [
+            "@org_tensorflow//tensorflow/lite/experimental/acceleration/mini_benchmark:mini_benchmark_implementation",
+        ],
+        "@org_tensorflow//tensorflow:android_arm64": [
+            "@org_tensorflow//tensorflow/lite/experimental/acceleration/mini_benchmark:mini_benchmark_implementation",
+        ],
+        "@org_tensorflow//tensorflow:linux_x86_64": [
+            "@org_tensorflow//tensorflow/lite/experimental/acceleration/mini_benchmark:mini_benchmark_implementation",
+        ],
+        "//conditions:default": [],
+    }),
 )

third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/status_macros.h

@@ -17,8 +17,8 @@ limitations under the License.
 #ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUS_MACROS_H_
 #define TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUS_MACROS_H_
 
-#include "absl/base/optimization.h"
-#include "absl/status/status.h"
+#include "absl/base/optimization.h"  // from @com_google_absl
+#include "absl/status/status.h"      // from @com_google_absl
 
 // Evaluates an expression that produces a `absl::Status`. If the status is not
 // ok, returns it from the current function.

third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/statusor.cc

@@ -1,67 +0,0 @@
-/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-// This file is forked from absl.
-
-#include "tensorflow_lite_support/cc/port/default/statusor.h"
-
-#include <utility>
-
-#include "absl/strings/str_cat.h"
-#include "base/logging.h"
-
-namespace tflite {
-namespace support {
-
-BadStatusOrAccess::BadStatusOrAccess(absl::Status status)
-    : status_(std::move(status)) {}
-
-BadStatusOrAccess::~BadStatusOrAccess() = default;
-
-const char* BadStatusOrAccess::what() const noexcept {
-  return "Bad StatusOr access";
-}
-
-const absl::Status& BadStatusOrAccess::status() const {
-  return status_;
-}
-
-namespace internal_statusor {
-
-void Helper::HandleInvalidStatusCtorArg(absl::Status* status) {
-  const char* kMessage =
-      "An OK status is not a valid constructor argument to StatusOr<T>";
-  LOG(DFATAL) << kMessage;
-  // In optimized builds, we will fall back to ::util::error::INTERNAL.
-  *status = absl::InternalError(kMessage);
-}
-
-void Helper::Crash(const absl::Status& status) {
-  LOG(FATAL) << "Attempting to fetch value instead of handling error "
-             << status;
-  _Exit(1);
-}
-
-void ThrowBadStatusOrAccess(absl::Status status) {
-#ifdef ABSL_HAVE_EXCEPTIONS
-  throw BadStatusOrAccess(std::move(status));
-#else
-  LOG(FATAL) << "Attempting to fetch value instead of handling error "
-             << status;
-#endif
-}
-
-}  // namespace internal_statusor
-}  // namespace support
-}  // namespace tflite

third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/statusor.h

@@ -1,584 +0,0 @@
-/* Copyright 2020 The TensorFlow Authors. All Rights Reserved.
-
-Licensed under the Apache License, Version 2.0 (the "License");
-you may not use this file except in compliance with the License.
-You may obtain a copy of the License at
-
-    http://www.apache.org/licenses/LICENSE-2.0
-
-Unless required by applicable law or agreed to in writing, software
-distributed under the License is distributed on an "AS IS" BASIS,
-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License.
-==============================================================================*/
-// This file is forked from absl.
-
-#ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUSOR_H_
-#define TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUSOR_H_
-
-#include <exception>
-#include <initializer_list>
-#include <new>
-#include <string>
-#include <type_traits>
-#include <utility>
-
-#include "absl/base/optimization.h"
-#include "absl/meta/type_traits.h"
-#include "absl/status/status.h"
-#include "absl/types/variant.h"
-#include "absl/utility/utility.h"
-#include "tensorflow_lite_support/cc/port/default/statusor_internals.h"
-
-namespace tflite {
-namespace support {
-
-#ifndef SWIG
-class BadStatusOrAccess : public std::exception {
- public:
-  explicit BadStatusOrAccess(absl::Status status);
-  ~BadStatusOrAccess() override;
-  const char* what() const noexcept override;
-  const absl::Status& status() const;
-
- private:
-  absl::Status status_;
-};
-#endif  // !SWIG
-
-// Returned StatusOr objects may not be ignored.
-// Note: Disabled for SWIG as it doesn't parse attributes correctly.  Codesearch
-// doesn't handle ifdefs as part of a class definitions (b/6995610), so we use a
-// forward declaration.
-#ifndef SWIG
-template <typename T>
-class ABSL_MUST_USE_RESULT StatusOr;
-#endif
-
-template <typename T>
-class StatusOr : private internal_statusor::StatusOrData<T>,
-                 private internal_statusor::CopyCtorBase<T>,
-                 private internal_statusor::MoveCtorBase<T>,
-                 private internal_statusor::CopyAssignBase<T>,
-                 private internal_statusor::MoveAssignBase<T> {
-  template <typename U>
-  friend class StatusOr;
-
-  typedef internal_statusor::StatusOrData<T> Base;
-
- public:
-  typedef T value_type;
-
-  // Constructs a new StatusOr with Status::UNKNOWN status.  This is marked
-  // 'explicit' to try to catch cases like 'return {};', where people think
-  // tflite::support::StatusOr<std::vector<int>> will be initialized with an
-  // empty vector, instead of a Status::UNKNOWN status.
-  explicit StatusOr();
-
-  // StatusOr<T> is copy constructible if T is copy constructible.
-  StatusOr(const StatusOr&) = default;
-  // StatusOr<T> is copy assignable if T is copy constructible and copy
-  // assignable.
-  StatusOr& operator=(const StatusOr&) = default;
-
-#ifndef SWIG
-
-  // StatusOr<T> is move constructible if T is move constructible.
-  StatusOr(StatusOr&&) = default;
-  // StatusOr<T> is moveAssignable if T is move constructible and move
-  // assignable.
-  StatusOr& operator=(StatusOr&&) = default;
-
-  // Converting constructors from StatusOr<U>, when T is constructible from U.
-  // To avoid ambiguity, they are disabled if T is also constructible from
-  // StatusOr<U>. Explicit iff the corresponding construction of T from U is
-  // explicit.
-  template <
-      typename U,
-      absl::enable_if_t<
-          absl::conjunction<
-              absl::negation<std::is_same<T, U>>,
-              std::is_constructible<T, const U&>,
-              std::is_convertible<const U&, T>,
-              absl::negation<
-                  internal_statusor::
-                      IsConstructibleOrConvertibleFromStatusOr<T, U>>>::value,
-          int> = 0>
-  StatusOr(const StatusOr<U>& other)  // NOLINT
-      : Base(static_cast<const typename StatusOr<U>::Base&>(other)) {}
-  template <
-      typename U,
-      absl::enable_if_t<
-          absl::conjunction<
-              absl::negation<std::is_same<T, U>>,
-              std::is_constructible<T, const U&>,
-              absl::negation<std::is_convertible<const U&, T>>,
-              absl::negation<
-                  internal_statusor::
-                      IsConstructibleOrConvertibleFromStatusOr<T, U>>>::value,
-          int> = 0>
-  explicit StatusOr(const StatusOr<U>& other)
-      : Base(static_cast<const typename StatusOr<U>::Base&>(other)) {}
-
-  template <
-      typename U,
-      absl::enable_if_t<
-          absl::conjunction<
-              absl::negation<std::is_same<T, U>>,
-              std::is_constructible<T, U&&>,
-              std::is_convertible<U&&, T>,
-              absl::negation<
-                  internal_statusor::
-                      IsConstructibleOrConvertibleFromStatusOr<T, U>>>::value,
-          int> = 0>
-  StatusOr(StatusOr<U>&& other)  // NOLINT
-      : Base(static_cast<typename StatusOr<U>::Base&&>(other)) {}
-  template <
-      typename U,
-      absl::enable_if_t<
-          absl::conjunction<
-              absl::negation<std::is_same<T, U>>,
-              std::is_constructible<T, U&&>,
-              absl::negation<std::is_convertible<U&&, T>>,
-              absl::negation<
-                  internal_statusor::
-                      IsConstructibleOrConvertibleFromStatusOr<T, U>>>::value,
-          int> = 0>
-  explicit StatusOr(StatusOr<U>&& other)
-      : Base(static_cast<typename StatusOr<U>::Base&&>(other)) {}
-
-  // Conversion copy/move assignment operator, T must be constructible and
-  // assignable from U. Only enable if T cannot be directly assigned from
-  // StatusOr<U>.
-  template <
-      typename U,
-      absl::enable_if_t<
-          absl::conjunction<
-              absl::negation<std::is_same<T, U>>,
-              std::is_constructible<T, const U&>,
-              std::is_assignable<T, const U&>,
-              absl::negation<
-                  internal_statusor::
-                      IsConstructibleOrConvertibleOrAssignableFromStatusOr<
-                          T,
-                          U>>>::value,
-          int> = 0>
-  StatusOr& operator=(const StatusOr<U>& other) {
-    this->Assign(other);
-    return *this;
-  }
-  template <
-      typename U,
-      absl::enable_if_t<
-          absl::conjunction<
-              absl::negation<std::is_same<T, U>>,
-              std::is_constructible<T, U&&>,
-              std::is_assignable<T, U&&>,
-              absl::negation<
-                  internal_statusor::
-                      IsConstructibleOrConvertibleOrAssignableFromStatusOr<
-                          T,
-                          U>>>::value,
-          int> = 0>
-  StatusOr& operator=(StatusOr<U>&& other) {
-    this->Assign(std::move(other));
-    return *this;
-  }
-
-#endif  // SWIG
-
-  // Constructs a new StatusOr with the given value. After calling this
-  // constructor, this->ok() will be true and the contained value may be
-  // retrieved with value(), operator*(), or operator->().
-  //
-  // NOTE: Not explicit - we want to use StatusOr<T> as a return type
-  // so it is convenient and sensible to be able to do 'return T()'
-  // when the return type is StatusOr<T>.
-  //
-  // REQUIRES: T is copy constructible.
-  // TODO(b/113125838): Replace this constructor with a direct-initialization
-  // constructor.
-  StatusOr(const T& value);
-
-  // Constructs a new StatusOr with the given non-ok status. After calling this
-  // constructor, this->ok() will be false and calls to value() will CHECK-fail.
-  //
-  // NOTE: Not explicit - we want to use StatusOr<T> as a return
-  // value, so it is convenient and sensible to be able to do 'return
-  // Status()' when the return type is StatusOr<T>.
-  //
-  // REQUIRES: !status.ok(). This requirement is DCHECKed.
-  // In optimized builds, passing util::OkStatus() here will have the effect
-  // of passing util::error::INTERNAL as a fallback.
-  StatusOr(const absl::Status& status);
-  StatusOr& operator=(const absl::Status& status);
-
-#ifndef SWIG
-  // Perfect-forwarding value assignment operator.
-  // If `*this` contains a `T` value before the call, the contained value is
-  // assigned from `std::forward<U>(v)`; Otherwise, it is directly-initialized
-  // from `std::forward<U>(v)`.
-  // This function does not participate in overload unless:
-  // 1. `std::is_constructible_v<T, U>` is true,
-  // 2. `std::is_assignable_v<T&, U>` is true.
-  // 3. `std::is_same_v<StatusOr<T>, std::remove_cvref_t<U>>` is false.
-  // 4. Assigning `U` to `T` is not ambiguous:
-  //  If `U` is `StatusOr<V>` and `T` is constructible and assignable from
-  //  both `StatusOr<V>` and `V`, the assignment is considered bug-prone and
-  //  ambiguous thus will fail to compile. For example:
-  //    StatusOr<bool> s1 = true;  // s1.ok() && *s1 == true
-  //    StatusOr<bool> s2 = false;  // s2.ok() && *s2 == false
-  //    s1 = s2;  // ambiguous, `s1 = *s2` or `s1 = bool(s2)`?
-  template <
-      typename U = T,
-      typename = typename std::enable_if<absl::conjunction<
-          std::is_constructible<T, U&&>,
-          std::is_assignable<T&, U&&>,
-          internal_statusor::IsForwardingAssignmentValid<T, U&&>>::value>::type>
-  StatusOr& operator=(U&& v) {
-    this->Assign(std::forward<U>(v));
-    return *this;
-  }
-
-  // Similar to the `const T&` overload.
-  //
-  // REQUIRES: T is move constructible.
-  StatusOr(T&& value);
-
-  // RValue versions of the operations declared above.
-  StatusOr(absl::Status&& status);
-  StatusOr& operator=(absl::Status&& status);
-
-  // Constructs the inner value T in-place using the provided args, using the
-  // T(args...) constructor.
-  template <typename... Args>
-  explicit StatusOr(absl::in_place_t, Args&&... args);
-  template <typename U, typename... Args>
-  explicit StatusOr(absl::in_place_t,
-                    std::initializer_list<U> ilist,
-                    Args&&... args);
-
-  // Constructs the inner value T in-place using the provided args, using the
-  // T(U) (direct-initialization) constructor. Only valid if T can be
-  // constructed from a U. Can accept move or copy constructors. Explicit if
-  // U is not convertible to T. To avoid ambiguity, this is disabled if U is
-  // a StatusOr<J>, where J is convertible to T.
-  // Style waiver for implicit conversion granted in cl/209187539.
-  template <typename U = T,
-            absl::enable_if_t<
-                absl::conjunction<
-                    internal_statusor::IsDirectInitializationValid<T, U&&>,
-                    std::is_constructible<T, U&&>,
-                    std::is_convertible<U&&, T>>::value,
-                int> = 0>
-  StatusOr(U&& u)  // NOLINT
-      : StatusOr(absl::in_place, std::forward<U>(u)) {}
-
-  template <typename U = T,
-            absl::enable_if_t<
-                absl::conjunction<
-                    internal_statusor::IsDirectInitializationValid<T, U&&>,
-                    std::is_constructible<T, U&&>,
-                    absl::negation<std::is_convertible<U&&, T>>>::value,
-                int> = 0>
-  explicit StatusOr(U&& u)  // NOLINT
-      : StatusOr(absl::in_place, std::forward<U>(u)) {}
-
-#endif  // SWIG
-
-  // Returns this->status().ok()
-  ABSL_MUST_USE_RESULT bool ok() const { return this->status_.ok(); }
-
-  // Returns a reference to our status. If this contains a T, then
-  // returns util::OkStatus().
-#ifdef SWIG
-  const ::util::Status& status() const;
-#else   // SWIG
-  const absl::Status& status() const&;
-  absl::Status status() &&;
-#endif  // SWIG
-
-  // Returns a reference to the held value if `this->ok()`. Otherwise, throws
-  // `absl::BadStatusOrAccess` if exception is enabled, or `LOG(FATAL)` if
-  // exception is disabled.
-  // If you have already checked the status using `this->ok()` or
-  // `operator bool()`, you probably want to use `operator*()` or `operator->()`
-  // to access the value instead of `value`.
-  // Note: for value types that are cheap to copy, prefer simple code:
-  //
-  //   T value = statusor.value();
-  //
-  // Otherwise, if the value type is expensive to copy, but can be left
-  // in the StatusOr, simply assign to a reference:
-  //
-  //   T& value = statusor.value();  // or `const T&`
-  //
-  // Otherwise, if the value type supports an efficient move, it can be
-  // used as follows:
-  //
-  //   T value = std::move(statusor).value();
-  //
-  // The `std::move` on statusor instead of on the whole expression enables
-  // warnings about possible uses of the statusor object after the move.
-#ifdef SWIG
-  const T& value() const;
-#else   // SWIG
-  const T& value() const&;
-  T& value() &;
-  const T&& value() const&&;
-  T&& value() &&;
-#endif  // SWIG
-
-#ifndef SWIG
-  // Returns a reference to the current value.
-  //
-  // REQUIRES: this->ok() == true, otherwise the behavior is undefined.
-  //
-  // Use this->ok() or `operator bool()` to verify that there is a current
-  // value. Alternatively, see value() for a similar API that guarantees
-  // CHECK-failing if there is no current value.
-  const T& operator*() const&;
-  T& operator*() &;
-  const T&& operator*() const&&;
-  T&& operator*() &&;
-#endif  // SWIG
-
-#ifndef SWIG
-  // Returns a pointer to the current value.
-  //
-  // REQUIRES: this->ok() == true, otherwise the behavior is undefined.
-  //
-  // Use this->ok() or `operator bool()` to verify that there is a current
-  // value.
-  const T* operator->() const;
-  T* operator->();
-#endif  // SWIG
-
-#ifndef SWIG
-  // Returns a copy of the current value if this->ok() == true. Otherwise
-  // returns a default value.
-  template <typename U>
-  T value_or(U&& default_value) const&;
-  template <typename U>
-  T value_or(U&& default_value) &&;
-#endif  // SWIG
-
-  // Ignores any errors. This method does nothing except potentially suppress
-  // complaints from any tools that are checking that errors are not dropped on
-  // the floor.
-  void IgnoreError() const;
-
-#ifndef SWIG
-  // Reconstructs the inner value T in-place using the provided args, using the
-  // T(args...) constructor. Returns reference to the reconstructed `T`.
-  template <typename... Args>
-  T& emplace(Args&&... args) {
-    if (ok()) {
-      this->Clear();
-      this->MakeValue(std::forward<Args>(args)...);
-    } else {
-      this->MakeValue(std::forward<Args>(args)...);
-      this->status_ = absl::OkStatus();
-    }
-    return this->data_;
-  }
-
-  template <
-      typename U,
-      typename... Args,
-      absl::enable_if_t<
-          std::is_constructible<T, std::initializer_list<U>&, Args&&...>::value,
-          int> = 0>
-  T& emplace(std::initializer_list<U> ilist, Args&&... args) {
-    if (ok()) {
-      this->Clear();
-      this->MakeValue(ilist, std::forward<Args>(args)...);
-    } else {
-      this->MakeValue(ilist, std::forward<Args>(args)...);
-      this->status_ = absl::OkStatus();
-    }
-    return this->data_;
-  }
-#endif  // SWIG
-
- private:
-#ifndef SWIG
-  using internal_statusor::StatusOrData<T>::Assign;
-  template <typename U>
-  void Assign(const StatusOr<U>& other);
-  template <typename U>
-  void Assign(StatusOr<U>&& other);
-#endif  // SWIG
-};
-
-#ifndef SWIG
-////////////////////////////////////////////////////////////////////////////////
-// Implementation details for StatusOr<T>
-
-template <typename T>
-tflite::support::StatusOr<T>::StatusOr()
-    : Base(absl::Status(absl::StatusCode::kUnknown, "")) {}
-
-template <typename T>
-tflite::support::StatusOr<T>::StatusOr(const T& value) : Base(value) {}
-
-template <typename T>
-tflite::support::StatusOr<T>::StatusOr(const absl::Status& status)
-    : Base(status) {}
-
-template <typename T>
-tflite::support::StatusOr<T>& StatusOr<T>::operator=(
-    const absl::Status& status) {
-  this->Assign(status);
-  return *this;
-}
-
-template <typename T>
-tflite::support::StatusOr<T>::StatusOr(T&& value) : Base(std::move(value)) {}
-
-template <typename T>
-tflite::support::StatusOr<T>::StatusOr(absl::Status&& status)
-    : Base(std::move(status)) {}
-
-template <typename T>
-tflite::support::StatusOr<T>& StatusOr<T>::operator=(absl::Status&& status) {
-  this->Assign(std::move(status));
-  return *this;
-}
-
-template <typename T>
-template <typename U>
-inline void StatusOr<T>::Assign(const StatusOr<U>& other) {
-  if (other.ok()) {
-    this->Assign(other.value());
-  } else {
-    this->Assign(other.status());
-  }
-}
-
-template <typename T>
-template <typename U>
-inline void StatusOr<T>::Assign(StatusOr<U>&& other) {
-  if (other.ok()) {
-    this->Assign(std::move(other).value());
-  } else {
-    this->Assign(std::move(other).status());
-  }
-}
-template <typename T>
-template <typename... Args>
-tflite::support::StatusOr<T>::StatusOr(absl::in_place_t, Args&&... args)
-    : Base(absl::in_place, std::forward<Args>(args)...) {}
-
-template <typename T>
-template <typename U, typename... Args>
-tflite::support::StatusOr<T>::StatusOr(absl::in_place_t,
-                                       std::initializer_list<U> ilist,
-                                       Args&&... args)
-    : Base(absl::in_place, ilist, std::forward<Args>(args)...) {}
-
-template <typename T>
-const absl::Status& StatusOr<T>::status() const& {
-  return this->status_;
-}
-template <typename T>
-absl::Status StatusOr<T>::status() && {
-  return ok() ? absl::OkStatus() : std::move(this->status_);
-}
-
-template <typename T>
-const T& StatusOr<T>::value() const& {
-  if (!this->ok())
-    internal_statusor::ThrowBadStatusOrAccess(this->status_);
-  return this->data_;
-}
-
-template <typename T>
-T& StatusOr<T>::value() & {
-  if (!this->ok())
-    internal_statusor::ThrowBadStatusOrAccess(this->status_);
-  return this->data_;
-}
-
-template <typename T>
-const T&& StatusOr<T>::value() const&& {
-  if (!this->ok()) {
-    internal_statusor::ThrowBadStatusOrAccess(std::move(this->status_));
-  }
-  return std::move(this->data_);
-}
-
-template <typename T>
-T&& StatusOr<T>::value() && {
-  if (!this->ok()) {
-    internal_statusor::ThrowBadStatusOrAccess(std::move(this->status_));
-  }
-  return std::move(this->data_);
-}
-
-template <typename T>
-const T& StatusOr<T>::operator*() const& {
-  this->EnsureOk();
-  return this->data_;
-}
-
-template <typename T>
-T& StatusOr<T>::operator*() & {
-  this->EnsureOk();
-  return this->data_;
-}
-
-template <typename T>
-const T&& StatusOr<T>::operator*() const&& {
-  this->EnsureOk();
-  return std::move(this->data_);
-}
-
-template <typename T>
-T&& StatusOr<T>::operator*() && {
-  this->EnsureOk();
-  return std::move(this->data_);
-}
-
-template <typename T>
-const T* StatusOr<T>::operator->() const {
-  this->EnsureOk();
-  return &this->data_;
-}
-
-template <typename T>
-T* StatusOr<T>::operator->() {
-  this->EnsureOk();
-  return &this->data_;
-}
-
-template <typename T>
-template <typename U>
-T StatusOr<T>::value_or(U&& default_value) const& {
-  if (ok()) {
-    return this->data_;
-  }
-  return std::forward<U>(default_value);
-}
-
-template <typename T>
-template <typename U>
-T StatusOr<T>::value_or(U&& default_value) && {
-  if (ok()) {
-    return std::move(this->data_);
-  }
-  return std::forward<U>(default_value);
-}
-
-template <typename T>
-void StatusOr<T>::IgnoreError() const {
-  // no-op
-}
-
-#endif  // SWIG
-
-}  // namespace support
-}  // namespace tflite
-
-#endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_STATUSOR_H_

third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/statusor_internals.h

@@ -20,9 +20,9 @@ limitations under the License.
 #include <type_traits>
 #include <utility>
 
-#include "absl/meta/type_traits.h"
-#include "absl/status/status.h"
-#include "absl/utility/utility.h"
+#include "absl/meta/type_traits.h"  // from @com_google_absl
+#include "absl/status/status.h"     // from @com_google_absl
+#include "absl/utility/utility.h"   // from @com_google_absl
 
 namespace tflite {
 namespace support {

third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/tflite_wrapper.cc

@@ -15,45 +15,332 @@ limitations under the License.
 
 #include "tensorflow_lite_support/cc/port/default/tflite_wrapper.h"
 
-#include "absl/status/status.h"
+#include "absl/status/status.h"       // from @com_google_absl
+#include "absl/strings/str_format.h"  // from @com_google_absl
+#include "tensorflow/lite/c/common.h"
+#include "tensorflow/lite/delegates/interpreter_utils.h"
+#include "tensorflow/lite/experimental/acceleration/configuration/flatbuffer_to_proto.h"
+#include "tensorflow/lite/experimental/acceleration/configuration/proto_to_flatbuffer.h"
+#include "tensorflow/lite/minimal_logging.h"
 #include "tensorflow_lite_support/cc/port/status_macros.h"
 
 namespace tflite {
 namespace support {
 
+namespace {
+using tflite::delegates::DelegatePluginRegistry;
+using tflite::delegates::InterpreterUtils;
+using tflite::proto::ComputeSettings;
+using tflite::proto::Delegate;
+}  // namespace
+
+/* static */
+absl::Status TfLiteInterpreterWrapper::SanityCheckComputeSettings(
+    const ComputeSettings& compute_settings) {
+  Delegate delegate = compute_settings.tflite_settings().delegate();
+  if (delegate != Delegate::NONE && delegate != Delegate::GPU &&
+      delegate != Delegate::HEXAGON && delegate != Delegate::NNAPI &&
+      delegate != Delegate::XNNPACK && delegate != Delegate::EDGETPU_CORAL) {
+    return absl::UnimplementedError(absl::StrFormat(
+        "Using delegate '%s' is not supported.", Delegate_Name(delegate)));
+  }
+  return absl::OkStatus();
+}
+
+TfLiteInterpreterWrapper::TfLiteInterpreterWrapper(
+    const std::string& default_model_namespace,
+    const std::string& default_model_id)
+    : delegate_(nullptr, nullptr),
+      got_error_do_not_delegate_anymore_(false),
+      default_model_namespace_(default_model_namespace),
+      default_model_id_(default_model_id),
+      mini_benchmark_(nullptr) {}
+
+std::string TfLiteInterpreterWrapper::ModelNamespace() {
+  const auto& ns_from_acceleration =
+      compute_settings_.model_namespace_for_statistics();
+  return ns_from_acceleration.empty() ? default_model_namespace_
+                                      : ns_from_acceleration;
+}
+
+std::string TfLiteInterpreterWrapper::ModelID() {
+  const auto& id_from_acceleration =
+      compute_settings_.model_identifier_for_statistics();
+  return id_from_acceleration.empty() ? default_model_id_
+                                      : id_from_acceleration;
+}
+
+// This is the deprecated overload that doesn't take an
+// InterpreterCreationResources parameter.
 absl::Status TfLiteInterpreterWrapper::InitializeWithFallback(
     std::function<absl::Status(std::unique_ptr<tflite::Interpreter>*)>
         interpreter_initializer,
-    const tflite::proto::ComputeSettings& compute_settings) {
-  if (compute_settings.has_preference() ||
-      compute_settings.has_tflite_settings()) {
-    return absl::UnimplementedError(
-        "Acceleration via ComputeSettings is not supported yet.");
+    const ComputeSettings& compute_settings) {
+  return InitializeWithFallback(
+      [interpreter_initializer](
+          const InterpreterCreationResources& resources,
+          std::unique_ptr<tflite::Interpreter>* interpreter_out)
+          -> absl::Status {
+        RETURN_IF_ERROR(interpreter_initializer(interpreter_out));
+        if (*interpreter_out != nullptr &&
+            resources.optional_delegate != nullptr) {
+          TfLiteStatus status =
+              (*interpreter_out)
+                  ->ModifyGraphWithDelegate(resources.optional_delegate);
+          if (status != kTfLiteOk) {
+            *interpreter_out = nullptr;
+            RETURN_IF_ERROR(
+                absl::InvalidArgumentError("Applying delegate failed"));
+          }
+        }
+        return absl::OkStatus();
+      },
+      compute_settings);
+}
+
+absl::Status TfLiteInterpreterWrapper::InitializeWithFallback(
+    std::function<absl::Status(const InterpreterCreationResources&,
+                               std::unique_ptr<tflite::Interpreter>*)>
+        interpreter_initializer,
+    const ComputeSettings& compute_settings) {
+  // Store interpreter initializer if not already here.
+  if (interpreter_initializer_) {
+    return absl::FailedPreconditionError(
+        "InitializeWithFallback already called.");
   }
-  RETURN_IF_ERROR(interpreter_initializer(&interpreter_));
-  return interpreter_->AllocateTensors() != kTfLiteOk
-             ? absl::InternalError(
-                   "TFLite interpreter: AllocateTensors() failed.")
-             : absl::OkStatus();
+  interpreter_initializer_ = std::move(interpreter_initializer);
+
+  // Sanity check and copy ComputeSettings.
+  RETURN_IF_ERROR(SanityCheckComputeSettings(compute_settings));
+  compute_settings_ = compute_settings;
+  if (compute_settings_.has_settings_to_test_locally()) {
+    flatbuffers::FlatBufferBuilder mini_benchmark_settings_fbb;
+    const auto* mini_benchmark_settings =
+        tflite::ConvertFromProto(compute_settings_.settings_to_test_locally(),
+                                 &mini_benchmark_settings_fbb);
+    mini_benchmark_ = tflite::acceleration::CreateMiniBenchmark(
+        *mini_benchmark_settings, ModelNamespace(), ModelID());
+    const tflite::ComputeSettingsT from_minibenchmark =
+        mini_benchmark_->GetBestAcceleration();
+    if (from_minibenchmark.tflite_settings != nullptr) {
+      TFLITE_LOG_PROD_ONCE(TFLITE_LOG_INFO, "Using mini benchmark results\n");
+      compute_settings_ = tflite::ConvertFromFlatbuffer(
+          from_minibenchmark, /*skip_mini_benchmark_settings=*/true);
+    }
+    // Trigger mini benchmark if it hasn't already run. Vast majority of cases
+    // should not actually do anything, since first runs are rare.
+    mini_benchmark_->TriggerMiniBenchmark();
+    mini_benchmark_->MarkAndGetEventsToLog();
+  }
+
+  // Initialize fallback behavior.
+  fallback_on_compilation_error_ =
+      compute_settings_.tflite_settings()
+          .fallback_settings()
+          .allow_automatic_fallback_on_compilation_error() ||
+      // Deprecated, keep supporting for backward compatibility.
+      compute_settings_.tflite_settings()
+          .nnapi_settings()
+          .fallback_settings()
+          .allow_automatic_fallback_on_compilation_error();
+  fallback_on_execution_error_ =
+      compute_settings_.tflite_settings()
+          .fallback_settings()
+          .allow_automatic_fallback_on_execution_error() ||
+      // Deprecated, keep supporting for backward compatibility.
+      compute_settings_.tflite_settings()
+          .nnapi_settings()
+          .fallback_settings()
+          .allow_automatic_fallback_on_execution_error();
+
+  return InitializeWithFallbackAndResize();
+}
+
+absl::Status TfLiteInterpreterWrapper::AllocateTensors() {
+  if (interpreter_->AllocateTensors() != kTfLiteOk) {
+    return absl::InternalError("AllocateTensors() failed.");
+  }
+  return absl::OkStatus();
+}
+
+// TODO(b/173406463): the `resize` parameter is going to be used by
+// ResizeAndAllocateTensors functions, coming soon.
+absl::Status TfLiteInterpreterWrapper::InitializeWithFallbackAndResize(
+    std::function<absl::Status(Interpreter*)> resize) {
+  InterpreterCreationResources resources{};
+  if (got_error_do_not_delegate_anymore_ ||
+      compute_settings_.tflite_settings().delegate() == Delegate::NONE) {
+    delegate_.reset(nullptr);
+  } else {
+    // Initialize delegate and add it to 'resources'.
+    RETURN_IF_ERROR(InitializeDelegate());
+    resources.optional_delegate = delegate_.get();
+  }
+
+  absl::Status status = interpreter_initializer_(resources, &interpreter_);
+  if (resources.optional_delegate == nullptr) {
+    RETURN_IF_ERROR(status);
+  }
+  if (resources.optional_delegate != nullptr && !status.ok()) {
+    // Any error when constructing the interpreter is assumed to be a delegate
+    // compilation error.  If a delegate compilation error occurs, stop
+    // delegation from happening in the future.
+    got_error_do_not_delegate_anymore_ = true;
+    delegate_.reset(nullptr);
+    if (fallback_on_compilation_error_) {
+      InterpreterCreationResources fallback_resources{};
+      fallback_resources.optional_delegate = nullptr;
+      RETURN_IF_ERROR(
+          interpreter_initializer_(fallback_resources, &interpreter_));
+    } else {
+      // If instructed not to fallback, return error.
+      return absl::InternalError(absl::StrFormat(
+          "ModifyGraphWithDelegate() failed for delegate '%s'.",
+          Delegate_Name(compute_settings_.tflite_settings().delegate())));
+    }
+  }
+
+  RETURN_IF_ERROR(resize(interpreter_.get()));
+  if (compute_settings_.tflite_settings().cpu_settings().num_threads() != -1) {
+    if (interpreter_->SetNumThreads(
+            compute_settings_.tflite_settings().cpu_settings().num_threads()) !=
+        kTfLiteOk) {
+      return absl::InternalError("Failed setting number of CPU threads");
+    }
+  }
+  SetTfLiteCancellation();
+
+  if (!delegate_) {
+    // Just allocate tensors and return.
+    return AllocateTensors();
+  }
+
+  // The call to ModifyGraphWithDelegate() leaves the interpreter in a usable
+  // state in case of failure: calling AllocateTensors() will silently fallback
+  // on CPU in such a situation.
+  return AllocateTensors();
+}
+
+absl::Status TfLiteInterpreterWrapper::InitializeDelegate() {
+  if (delegate_ == nullptr) {
+    Delegate which_delegate = compute_settings_.tflite_settings().delegate();
+    const tflite::ComputeSettings* compute_settings =
+        tflite::ConvertFromProto(compute_settings_, &flatbuffers_builder_);
+
+    if (which_delegate == Delegate::NNAPI) {
+      RETURN_IF_ERROR(
+          LoadDelegatePlugin("Nnapi", *compute_settings->tflite_settings()));
+    } else if (which_delegate == Delegate::HEXAGON) {
+      RETURN_IF_ERROR(
+          LoadDelegatePlugin("Hexagon", *compute_settings->tflite_settings()));
+    } else if (which_delegate == Delegate::GPU) {
+      RETURN_IF_ERROR(
+          LoadDelegatePlugin("Gpu", *compute_settings->tflite_settings()));
+    } else if (which_delegate == Delegate::EDGETPU) {
+      RETURN_IF_ERROR(
+          LoadDelegatePlugin("EdgeTpu", *compute_settings->tflite_settings()));
+    } else if (which_delegate == Delegate::EDGETPU_CORAL) {
+      RETURN_IF_ERROR(LoadDelegatePlugin("EdgeTpuCoral",
+                                         *compute_settings->tflite_settings()));
+    } else if (which_delegate == Delegate::XNNPACK) {
+      RETURN_IF_ERROR(
+          LoadDelegatePlugin("XNNPack", *compute_settings->tflite_settings()));
+    }
+  }
+  return absl::OkStatus();
 }
 
 absl::Status TfLiteInterpreterWrapper::InvokeWithFallback(
     const std::function<absl::Status(tflite::Interpreter* interpreter)>&
         set_inputs) {
   RETURN_IF_ERROR(set_inputs(interpreter_.get()));
-  return interpreter_->Invoke() != kTfLiteOk
-             ? absl::InternalError("TFLite interpreter: Invoke() failed.")
-             : absl::OkStatus();
+  // Reset cancel flag before calling `Invoke()`.
+  cancel_flag_.Set(false);
+  TfLiteStatus status = kTfLiteError;
+  if (fallback_on_execution_error_) {
+    status = InterpreterUtils::InvokeWithCPUFallback(interpreter_.get());
+  } else {
+    status = interpreter_->Invoke();
+  }
+  if (status == kTfLiteOk) {
+    return absl::OkStatus();
+  }
+  // Assume InvokeWithoutFallback() is guarded under caller's synchronization.
+  // Assume the inference is cancelled successfully if Invoke() returns
+  // kTfLiteError and the cancel flag is `true`.
+  if (status == kTfLiteError && cancel_flag_.Get()) {
+    return absl::CancelledError("Invoke() cancelled.");
+  }
+  if (delegate_) {
+    // Mark that an error occurred so that later invocations immediately
+    // fallback to CPU.
+    got_error_do_not_delegate_anymore_ = true;
+    // InvokeWithCPUFallback returns `kTfLiteDelegateError` in case of
+    // *successful* fallback: convert it to an OK status.
+    if (status == kTfLiteDelegateError) {
+      return absl::OkStatus();
+    }
+  }
+  return absl::InternalError("Invoke() failed.");
 }
 
 absl::Status TfLiteInterpreterWrapper::InvokeWithoutFallback() {
-  return interpreter_->Invoke() != kTfLiteOk
-             ? absl::InternalError("TFLite interpreter: Invoke() failed.")
-             : absl::OkStatus();
+  // Reset cancel flag before calling `Invoke()`.
+  cancel_flag_.Set(false);
+  TfLiteStatus status = interpreter_->Invoke();
+  if (status != kTfLiteOk) {
+    // Assume InvokeWithoutFallback() is guarded under caller's synchronization.
+    // Assume the inference is cancelled successfully if Invoke() returns
+    // kTfLiteError and the cancel flag is `true`.
+    if (status == kTfLiteError && cancel_flag_.Get()) {
+      return absl::CancelledError("Invoke() cancelled.");
+    }
+    return absl::InternalError("Invoke() failed.");
+  }
+  return absl::OkStatus();
 }
 
 void TfLiteInterpreterWrapper::Cancel() {
-  // NOP
+  cancel_flag_.Set(true);
+}
+
+void TfLiteInterpreterWrapper::SetTfLiteCancellation() {
+  // Create a cancellation check function and set to the TFLite interpreter.
+  auto check_cancel_flag = [](void* data) {
+    auto* cancel_flag = reinterpret_cast<CancelFlag*>(data);
+    return cancel_flag->Get();
+  };
+  interpreter_->SetCancellationFunction(reinterpret_cast<void*>(&cancel_flag_),
+                                        check_cancel_flag);
+}
+
+absl::Status TfLiteInterpreterWrapper::LoadDelegatePlugin(
+    const std::string& name,
+    const tflite::TFLiteSettings& tflite_settings) {
+  delegate_plugin_ = DelegatePluginRegistry::CreateByName(
+      absl::StrFormat("%sPlugin", name), tflite_settings);
+
+  if (delegate_plugin_ == nullptr) {
+    return absl::InternalError(absl::StrFormat(
+        "Could not create %s plugin. Have you linked in the %s_plugin target?",
+        name, name));
+  }
+
+  delegate_ = delegate_plugin_->Create();
+  if (delegate_ == nullptr) {
+    return absl::InternalError(
+        absl::StrFormat("Plugin did not create %s delegate.", name));
+  }
+
+  return absl::OkStatus();
+}
+
+bool TfLiteInterpreterWrapper::HasMiniBenchmarkCompleted() {
+  if (mini_benchmark_ != nullptr &&
+      mini_benchmark_->NumRemainingAccelerationTests() == 0) {
+    return true;
+  }
+  return false;
 }
 
 }  // namespace support

third_party/tflite_support/src/tensorflow_lite_support/cc/port/default/tflite_wrapper.h

@@ -16,40 +16,123 @@ limitations under the License.
 #define TENSORFLOW_LITE_SUPPORT_CC_PORT_DEFAULT_TFLITE_WRAPPER_H_
 
 #include <memory>
+#include <string>
 #include <utility>
 
-#include "absl/status/status.h"
+#include "absl/status/status.h"       // from @com_google_absl
+#include "flatbuffers/flatbuffers.h"  // from @flatbuffers
+#include "tensorflow/lite/c/common.h"
 #include "tensorflow/lite/experimental/acceleration/configuration/configuration.pb.h"
+#include "tensorflow/lite/experimental/acceleration/configuration/delegate_registry.h"
+#include "tensorflow/lite/experimental/acceleration/mini_benchmark/mini_benchmark.h"
 #include "tensorflow/lite/interpreter.h"
+#include "tensorflow/lite/interpreter_builder.h"
 
 namespace tflite {
 namespace support {
 
-// Wrapper for a TfLiteInterpreter that may be accelerated[1]. This is NOT yet
-// implemented: this class only provides a first, minimal interface in the
-// meanwhile.
+// Options that are created by `TFLiteInterpreterWrapper` and will help to
+// initialize Interpreter in the callback function. `TFLiteInterpreterWrapper`
+// retains ownership of the included options, and will ensure that they remain
+// valid for the duration of the created interpreter's lifetime.
+struct InterpreterCreationResources {
+  // The delegate created, based on the parameters in `ComputeSettings`.
+  // `TfLiteInterpreterWrapper` exclusively owns the `TfLiteDelegate` object,
+  // and maintains it through out the lifetime of `TfLiteInterpreterWrapper`.
+  TfLiteDelegate* optional_delegate;
+
+  // Number of threads to use, or -1 to use the default number of threads.
+  int num_threads = -1;
+
+  // Apply the InterpreterCreationResources to the InterpreterBuilder.
+  // Note: caller is responsible for ensuring that arguments are valid,
+  // e.g. that num_threads >= -1.
+  void ApplyTo(tflite::InterpreterBuilder* interpreter_builder) const {
+    if (optional_delegate != nullptr) {
+      interpreter_builder->AddDelegate(optional_delegate);
+    }
+    if (num_threads != -1) {
+      // We ignore the TfLiteStatus return value here; caller is responsible
+      // for checking that num_threads is valid.
+      (void)interpreter_builder->SetNumThreads(num_threads);
+    }
+  }
+};
+
+// Wrapper for a TfLiteInterpreter that may be accelerated [1]. Meant to be
+// substituted for `unique_ptr<tflite::Interpreter>` class members.
 //
-// [1] See tensorflow/lite/experimental/acceleration for more details.
+// This class is in charge of:
+// * Picking, instantiating and configuring the right delegate for the provided
+//   ComputeSettings [2],
+// * Providing methods to initialize and invoke the Interpreter with optional
+//   (controlled through the ComputeSettings) automatic fallback to CPU if any
+//   acceleration-related error occurs at compilation or runtime.
+// * TODO(b/169474250) Cache interpreters for multiple input sizes to enable
+//   performant acceleration for the case where input size changes frequently.
+//
+// IMPORTANT: The only supported delegates are (as defined in [1]) NONE, GPU,
+// HEXAGON and NNAPI. Trying to use this class with EDGETPU or XNNPACK delegates
+// will cause an UnimplementedError to be thrown at initialization time.
+//
+// Like TfLiteInterpreter, this class is thread-compatible. Use from multiple
+// threads must be guarded by synchronization outside this class.
+//
+// [1]:
+// https://github.com/tensorflow/tensorflow/blob/master/tensorflow/lite/experimental/acceleration/configuration/configuration.proto
 class TfLiteInterpreterWrapper {
  public:
-  TfLiteInterpreterWrapper() = default;
+  // Creates an instance to be associated with a TfLite model that could be
+  // identified by (`default_model_namespace`, `default_model_id`). Note the
+  // model identifier is generally used for the sake of logging.
+  TfLiteInterpreterWrapper(const std::string& default_model_namespace,
+                           const std::string& default_model_id);
+  TfLiteInterpreterWrapper()
+      : TfLiteInterpreterWrapper("org.tensorflow.lite.support",
+                                 "unknown_model_id") {}
 
   virtual ~TfLiteInterpreterWrapper() = default;
 
-  // Calls `interpreter_initializer` and then `AllocateTensors`. Future
-  // implementation of this method will attempt to apply the provided
-  // `compute_settings` with a graceful fallback in case a failure occurs.
-  // Note: before this gets implemented, do NOT call this method with non-empty
-  // `compute_settings` otherwise an unimplemented error occurs.
+  // Calls `interpreter_initializer` to construct the Interpreter, then
+  // initializes it with the appropriate delegate (if any) specified through
+  // `compute_settings` and finally calls AllocateTensors() on it.
+  //
+  // Whether or not this function automatically falls back to using CPU in case
+  // initialization with a delegate fails depends on the FallbackSettings
+  // specified in the TFLiteSettings of the provided ComputeSettings: if the
+  // `allow_automatic_fallback_on_compilation_error` field is set to true,
+  // fallback will automatically happen; otherwise an InternalError will be
+  // thrown.
+  // This flag allows callers to rely on this function whether or not they
+  // actually want fallback to happen; if they don't, it will ensure that the
+  // configuration doesn't accidentally trigger fallback.
+  //
+  // IMPORTANT: Supported delegate type includes: NONE, NNAPI, GPU, HEXAGON,
+  // XNNPACK, EDGETPU (Google internal), and EDGETPU_CORAL. Specifying another
+  // delegate type may cause an UnimplementedError to be thrown.
+  absl::Status InitializeWithFallback(
+      std::function<absl::Status(const InterpreterCreationResources&,
+                                 std::unique_ptr<tflite::Interpreter>*)>
+          interpreter_initializer,
+      const tflite::proto::ComputeSettings& compute_settings);
+
+  // Deprecated: Use the one above with `InterpreterCreationResources` instead.
   absl::Status InitializeWithFallback(
       std::function<absl::Status(std::unique_ptr<tflite::Interpreter>*)>
           interpreter_initializer,
       const tflite::proto::ComputeSettings& compute_settings);
 
-  // Calls `set_inputs` and then Invoke() on the interpreter. Future
-  // implementation of this method will perform a graceful fallback in case a
-  // failure occur due to the `compute_settings` provided at initialization
-  // time.
+  // Calls `set_inputs` and then Invoke() on the interpreter.
+  //
+  // Whether or not this function automatically falls back to using CPU in case
+  // invocation with a delegate fails depends on the FallbackSettings
+  // specified in the TFLiteSettings of the ComputeSettings provided at
+  // initialization: if the `allow_automatic_fallback_on_execution_error`
+  // field is set to true, fallback will automatically happen; otherwise an
+  // InternalError will be thrown.
+  // This flag allows callers to rely on this function whether or not they
+  // actually want fallback to happen; if they don't, it will ensure that the
+  // configuration doesn't accidentally trigger fallback.
   absl::Status InvokeWithFallback(
       const std::function<absl::Status(tflite::Interpreter* interpreter)>&
           set_inputs);
@@ -58,8 +141,23 @@ class TfLiteInterpreterWrapper {
   // before-hand.
   absl::Status InvokeWithoutFallback();
 
-  // Cancels the current running TFLite invocation on CPU. This method is not
-  // yet implemented though it is safe to use it as it acts as a NOP.
+  // Cancels the current TFLite **CPU** inference.
+  //
+  // IMPORTANT: If inference is entirely running on a delegate, this has no
+  // effect; if inference is partially delegated, only the CPU part is
+  // cancelled.
+  //
+  // Usually called on a different thread than the one Invoke() is running
+  // on. Calling Cancel() while InvokeWithFallback() or InvokeWithoutFallback()
+  // is running may cause these methods to return a `CancelledError` with empty
+  // results. Calling Cancel() at any other time doesn't have any effect.
+  //
+  // InvokeWithFallback() and InvokeWithoutFallback() reset the cancel flag
+  // right before the underlying Invoke() is called, so these two methods can be
+  // called again on the same instance after a call to Cancel().
+  //
+  // Note that this is the only method that can be called from another thread
+  // without locking.
   void Cancel();
 
   // Accesses the underlying interpreter for other methods.
@@ -72,8 +170,109 @@ class TfLiteInterpreterWrapper {
   TfLiteInterpreterWrapper(const TfLiteInterpreterWrapper&) = delete;
   TfLiteInterpreterWrapper& operator=(const TfLiteInterpreterWrapper&) = delete;
 
+  // Whether an error has occurred with the delegate.
+  bool HasDelegateError() { return got_error_do_not_delegate_anymore_; }
+
+  // Whether the on-device mini-benchmark has completed for those TfLite
+  // acceleration configurations that are specified in passed-in
+  // ComputeSettings. If it finishes, the next time this same InterpreterWrapper
+  // object is created (i.e. w/ the same model and the same
+  // mini-benchmark-related configurations), the best acceleration configuration
+  // will be picked up and used.
+  bool HasMiniBenchmarkCompleted();
+
+  const tflite::proto::ComputeSettings& compute_settings() const {
+    return compute_settings_;
+  }
+
+ protected:
+  // The delegate used to accelerate inference.
+  Interpreter::TfLiteDelegatePtr delegate_;
+  // The corresponding delegate plugin.
+  std::unique_ptr<tflite::delegates::DelegatePluginInterface> delegate_plugin_;
+
  private:
+  // Performs sanity checks on the provided ComputeSettings.
+  static absl::Status SanityCheckComputeSettings(
+      const tflite::proto::ComputeSettings& compute_settings);
+
+  // Inner function for initializing an interpreter with fallback, optionally
+  // resizing input tensors by calling `resize` on the newly initialized
+  // interpreter.
+  absl::Status InitializeWithFallbackAndResize(
+      std::function<absl::Status(Interpreter* interpreter)> resize =
+          [](Interpreter* interpreter) { return absl::OkStatus(); });
+
+  // Initializes the delegate plugin and creates the delegate.
+  absl::Status InitializeDelegate();
+
+  // Wrapper around the interpreter's `AllocateTensors()` method converting the
+  // returned `TfLiteStatus` to an `absl::Status`.
+  absl::Status AllocateTensors();
+
+  absl::Status LoadDelegatePlugin(const std::string&,
+                                  const tflite::TFLiteSettings&);
+
+  std::string ModelNamespace();
+  std::string ModelID();
+
+  // The interpreter instance being used.
   std::unique_ptr<tflite::Interpreter> interpreter_;
+  // The function used to initialize the interpreter and store it into the
+  // provided `std::unique_ptr`.
+  // This is typically a wrapper function around `tflite::InterpreterBuilder`,
+  // giving the caller the opportunity to hook-up a custom `tflite::OpResolver`
+  // and / or `tflite::ErrorReporter`.
+  std::function<absl::Status(const InterpreterCreationResources&,
+                             std::unique_ptr<Interpreter>*)>
+      interpreter_initializer_;
+
+  // The ComputeSettings provided at initialization time.
+  // Note when TfLite mini-benchmark is enabled, it could be changed to the
+  // best TfLite acceleration setting selected.
+  tflite::proto::ComputeSettings compute_settings_;
+
+  // Set to true if an occurs with the specified delegate (if any), causing
+  // future calls to fallback on CPU.
+  bool got_error_do_not_delegate_anymore_;
+
+  // Fallback behavior as specified through the ComputeSettings.
+  bool fallback_on_compilation_error_;
+  bool fallback_on_execution_error_;
+
+  std::string default_model_namespace_;
+  std::string default_model_id_;
+
+  // Used to convert the ComputeSettings proto to FlatBuffer format.
+  flatbuffers::FlatBufferBuilder flatbuffers_builder_;
+
+  // Cancellation flag definition.
+  struct CancelFlag {
+    // Mutex to guard the `cancel_flag`.
+    mutable absl::Mutex cancel_mutex;
+
+    // A flag indicates if the caller cancels the TFLite interpreter invocation.
+    bool cancel_flag ABSL_GUARDED_BY(cancel_mutex) = false;
+
+    // Returns `cancel_flag`.
+    bool Get() const ABSL_LOCKS_EXCLUDED(cancel_mutex) {
+      absl::MutexLock cancel_lock(&cancel_mutex);
+      return cancel_flag;
+    }
+
+    // Sets `cancel_flag` to `value`.
+    void Set(bool value) ABSL_LOCKS_EXCLUDED(cancel_mutex) {
+      absl::MutexLock cancel_lock(&cancel_mutex);
+      cancel_flag = value;
+    }
+  };
+  CancelFlag cancel_flag_;
+
+  std::unique_ptr<tflite::acceleration::MiniBenchmark> mini_benchmark_;
+
+  // Sets up the TFLite invocation cancellation by
+  // tflite::Interpreter::SetCancellationFunction().
+  void SetTfLiteCancellation();
 };
 
 }  // namespace support

third_party/tflite_support/src/tensorflow_lite_support/cc/port/integral_types.h

@@ -37,6 +37,19 @@ typedef unsigned long uword_t;
 #define GG_LL_FORMAT "ll"  // As in "%lld". Note that "q" is poor form also.
 #define GG_LL_FORMAT_W L"ll"
 
+const uint8 kuint8max{0xFF};
+const uint16 kuint16max{0xFFFF};
+const uint32 kuint32max{0xFFFFFFFF};
+const uint64 kuint64max{GG_ULONGLONG(0xFFFFFFFFFFFFFFFF)};
+const int8 kint8min{~0x7F};
+const int8 kint8max{0x7F};
+const int16 kint16min{~0x7FFF};
+const int16 kint16max{0x7FFF};
+const int32 kint32min{~0x7FFFFFFF};
+const int32 kint32max{0x7FFFFFFF};
+const int64 kint64min{GG_LONGLONG(~0x7FFFFFFFFFFFFFFF)};
+const int64 kint64max{GG_LONGLONG(0x7FFFFFFFFFFFFFFF)};
+
 typedef uint64 Fprint;
 static const Fprint kIllegalFprint = 0;
 static const Fprint kMaxFprint = GG_ULONGLONG(0xFFFFFFFFFFFFFFFF);

third_party/tflite_support/src/tensorflow_lite_support/cc/port/status_matchers.h

@@ -0,0 +1,21 @@
+/* Copyright 2021 The TensorFlow Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_STATUS_MATCHERS_H_
+#define TENSORFLOW_LITE_SUPPORT_CC_PORT_STATUS_MATCHERS_H_
+
+#include "tensorflow_lite_support/cc/port/default/status_matchers.h"
+
+#endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_STATUS_MATCHERS_H_

third_party/tflite_support/src/tensorflow_lite_support/cc/port/statusor.h

@@ -16,5 +16,17 @@ limitations under the License.
 #ifndef TENSORFLOW_LITE_SUPPORT_CC_PORT_STATUSOR_H_
 #define TENSORFLOW_LITE_SUPPORT_CC_PORT_STATUSOR_H_
 
-#include "tensorflow_lite_support/cc/port/default/statusor.h"
+// This header file is used to manage the depended StatusOr library. It creates
+// an extra layer that makes it easier to switch between the desired version of
+// StatusOr.
+#include "absl/status/statusor.h"  // from @com_google_absl
+
+namespace tflite {
+namespace support {
+
+template <typename T>
+using StatusOr = absl::StatusOr<T>;
+
+}  // namespace support
+}  // namespace tflite
 #endif  // TENSORFLOW_LITE_SUPPORT_CC_PORT_STATUSOR_H_

third_party/tflite_support/src/tensorflow_lite_support/cc/task/README.md

@@ -15,8 +15,8 @@ answer question based on context.
 Use the C++ API to answer questions as follows:
 
 ```cc
-using tflite::task::text::qa::BertQuestionAnswerer;
-using tflite::task::text::qa::QaAnswer;
+using tflite::task::text::BertQuestionAnswerer;
+using tflite::task::text::QaAnswer;
 // Create API handler with Mobile Bert model.
 auto qa_client = BertQuestionAnswerer::CreateBertQuestionAnswererFromFile("/path/to/mobileBertModel", "/path/to/vocab");
 // Or create API handler with Albert model.

third_party/tflite_support/src/tensorflow_lite_support/cc/task/audio/BUILD

@@ -0,0 +1,65 @@
+load(
+    "@org_tensorflow//tensorflow/lite/core/shims:cc_library_with_tflite.bzl",
+    "cc_library_with_tflite",
+)
+
+package(
+    default_visibility = [
+        "//visibility:public",
+    ],
+    licenses = ["notice"],  # Apache 2.0
+)
+
+cc_library_with_tflite(
+    name = "audio_classifier",
+    srcs = ["audio_classifier.cc"],
+    hdrs = ["audio_classifier.h"],
+    tflite_deps = [
+        "@org_tensorflow//tensorflow/lite/core/shims:builtin_ops",
+        "//tensorflow_lite_support/cc/task/processor:classification_postprocessor",
+        "//tensorflow_lite_support/cc/task/processor:audio_preprocessor",
+        "//tensorflow_lite_support/cc/task/core:base_task_api",
+        "//tensorflow_lite_support/cc/task/core:task_api_factory",
+    ],
+    deps = [
+        "//tensorflow_lite_support/cc:common",
+        "//tensorflow_lite_support/cc/port:integral_types",
+        "//tensorflow_lite_support/cc/port:statusor",
+        "//tensorflow_lite_support/cc/task/audio/core:audio_buffer",
+        "//tensorflow_lite_support/cc/task/audio/proto:audio_classifier_options_cc_proto",
+        "//tensorflow_lite_support/cc/task/audio/proto:class_proto_inc",
+        "//tensorflow_lite_support/cc/task/audio/proto:classifications_proto_inc",
+        "//tensorflow_lite_support/cc/task/core:classification_head",
+        "//tensorflow_lite_support/cc/task/core:label_map_item",
+        "//tensorflow_lite_support/cc/task/core:task_utils",
+        "//tensorflow_lite_support/cc/task/processor/proto:classification_options_cc_proto",
+        "//tensorflow_lite_support/metadata:metadata_schema_cc",
+        "@com_google_absl//absl/status",
+        "@com_google_absl//absl/strings:str_format",
+        "@org_tensorflow//tensorflow/lite/c:c_api_types",
+        "@org_tensorflow//tensorflow/lite/core/api:op_resolver",
+    ],
+    alwayslink = 1,
+)
+
+cc_library_with_tflite(
+    name = "audio_embedder",
+    srcs = ["audio_embedder.cc"],
+    hdrs = ["audio_embedder.h"],
+    tflite_deps = [
+        "//tensorflow_lite_support/cc/task/processor:embedding_postprocessor",
+        "//tensorflow_lite_support/cc/task/processor:audio_preprocessor",
+        "//tensorflow_lite_support/cc/task/core:task_api_factory",
+        "//tensorflow_lite_support/cc/task/core:base_task_api",
+    ],
+    deps = [
+        "//tensorflow_lite_support/cc:common",
+        "//tensorflow_lite_support/cc/port:status_macros",
+        "//tensorflow_lite_support/cc/port:statusor",
+        "//tensorflow_lite_support/cc/task/audio/proto:audio_embedder_options_cc_proto",
+        "//tensorflow_lite_support/cc/task/processor/proto:embedding_cc_proto",
+        "//tensorflow_lite_support/cc/task/processor/proto:embedding_options_cc_proto",
+        "@com_google_absl//absl/status",
+        "@org_tensorflow//tensorflow/lite/c:common",
+    ],
+)