-
Notifications
You must be signed in to change notification settings - Fork 5.2k
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Make InferenceCalculatorDarwinn support float and int32 as input data…
… type. Before, input data type is default to float for Darwinn. Now some tflite models require int32 as input, thus this support is added. PiperOrigin-RevId: 602450883
- Loading branch information
1 parent
850da4e
commit 82060a1
Showing
5 changed files
with
293 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
173 changes: 173 additions & 0 deletions
173
mediapipe/calculators/tensor/inference_calculator_utils_test.cc
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,173 @@ | ||
| // Copyright 2024 The MediaPipe Authors. | ||
| // | ||
| // 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 "mediapipe/calculators/tensor/inference_calculator_utils.h" | ||
|
|
||
| #include <cstdint> | ||
| #include <cstring> | ||
| #include <vector> | ||
|
|
||
| #include "absl/log/absl_check.h" | ||
| #include "absl/status/status.h" | ||
| #include "mediapipe/framework/formats/tensor.h" | ||
| #include "mediapipe/framework/port/gmock.h" | ||
| #include "mediapipe/framework/port/gtest.h" | ||
| #include "mediapipe/framework/port/status_matchers.h" | ||
| #include "tensorflow/lite/c/common.h" | ||
| #include "tensorflow/lite/interpreter.h" | ||
|
|
||
| namespace mediapipe { | ||
| namespace { | ||
|
|
||
| using ElementType = ::mediapipe::Tensor::ElementType; | ||
| using ::testing::ElementsAreArray; | ||
| using ::testing::HasSubstr; | ||
| using ::tflite::Interpreter; | ||
|
|
||
| // Adds a tensor of certain type and size inside the interpreter, and update | ||
| // the tensor index. | ||
| void AddInterpreterInput(TfLiteType type, int size, int& tensor_index, | ||
| bool allocate_tensor, Interpreter& interpreter) { | ||
| ABSL_CHECK_EQ(interpreter.AddTensors(1, &tensor_index), kTfLiteOk); | ||
| TfLiteQuantizationParams quant; | ||
| interpreter.SetTensorParametersReadWrite(tensor_index, type, "", {size}, | ||
| quant); | ||
| interpreter.SetInputs({tensor_index}); | ||
| ABSL_CHECK_EQ(interpreter.tensor(interpreter.inputs()[tensor_index])->type, | ||
| type); | ||
| if (allocate_tensor) { | ||
| ABSL_CHECK_EQ(interpreter.AllocateTensors(), kTfLiteOk); | ||
| } | ||
| } | ||
|
|
||
| template <typename T> | ||
| std::vector<T> TfLiteTensorData(const Interpreter& interpreter, | ||
| int tensor_index) { | ||
| const TfLiteTensor* tensor = | ||
| interpreter.tensor(interpreter.inputs()[tensor_index]); | ||
| const T* tensor_ptr = reinterpret_cast<T*>(tensor->data.data); | ||
| ABSL_CHECK_NE(tensor_ptr, nullptr); | ||
| size_t tensor_size = tensor->bytes / sizeof(T); | ||
| return std::vector<T>(tensor_ptr, tensor_ptr + tensor_size); | ||
| } | ||
|
|
||
| TEST(InferenceCalculatorUtilsTest, | ||
| CopyCpuInputIntoInterpreterTensorWorksCorrectlyForInt32) { | ||
| tflite::Interpreter interpreter; | ||
| int tensor_index, tensor_len = 4; | ||
| AddInterpreterInput(kTfLiteInt32, tensor_len, tensor_index, | ||
| /*allocate_tensor=*/true, interpreter); | ||
| std::vector<int32_t> values{1, 2, 3, 4}; | ||
| int values_len = values.size(); | ||
| Tensor tensor(ElementType::kInt32, Tensor::Shape({values_len})); | ||
| std::memcpy(tensor.GetCpuWriteView().buffer<int32_t>(), values.data(), | ||
| values_len * sizeof(int32_t)); | ||
| MP_EXPECT_OK( | ||
| CopyCpuInputIntoInterpreterTensor(tensor, interpreter, tensor_index)); | ||
| EXPECT_THAT(TfLiteTensorData<int32_t>(interpreter, tensor_index), | ||
| ElementsAreArray(values)); | ||
| } | ||
|
|
||
| TEST(InferenceCalculatorUtilsTest, | ||
| CopyCpuInputIntoInterpreterTensorWorksCorrectlyForFloat32) { | ||
| tflite::Interpreter interpreter; | ||
| int tensor_index, tensor_len = 4; | ||
| AddInterpreterInput(kTfLiteFloat32, tensor_len, tensor_index, | ||
| /*allocate_tensor=*/true, interpreter); | ||
| std::vector<float> values{1.0f, 2.0f, 3.0f, 4.0f}; | ||
| int values_len = values.size(); | ||
| Tensor tensor(ElementType::kFloat32, Tensor::Shape({values_len})); | ||
| std::memcpy(tensor.GetCpuWriteView().buffer<float>(), values.data(), | ||
| values_len * sizeof(float)); | ||
| MP_EXPECT_OK( | ||
| CopyCpuInputIntoInterpreterTensor(tensor, interpreter, tensor_index)); | ||
| EXPECT_THAT(TfLiteTensorData<float>(interpreter, tensor_index), | ||
| ElementsAreArray(values)); | ||
| } | ||
|
|
||
| TEST(InferenceCalculatorUtilsTest, | ||
| CopyCpuInputIntoInterpreterTensorTypeMismatch) { | ||
| tflite::Interpreter interpreter; | ||
| int tensor_index, tensor_len = 4; | ||
| AddInterpreterInput(kTfLiteInt32, tensor_len, tensor_index, | ||
| /*allocate_tensor=*/true, interpreter); | ||
| std::vector<float> values{1.0f, 2.0f, 3.0f, 4.0f}; | ||
| int values_len = values.size(); | ||
| Tensor tensor(ElementType::kFloat32, Tensor::Shape({values_len})); | ||
| std::memcpy(tensor.GetCpuWriteView().buffer<float>(), values.data(), | ||
| values_len * sizeof(float)); | ||
| absl::Status status = | ||
| CopyCpuInputIntoInterpreterTensor(tensor, interpreter, tensor_index); | ||
| EXPECT_FALSE(status.ok()); | ||
| EXPECT_THAT(status.message(), | ||
| HasSubstr("Input and interpreter tensor type mismatch:")); | ||
| } | ||
|
|
||
| TEST(InferenceCalculatorUtilsTest, | ||
| CopyCpuInputIntoInterpreterTensorSizeMismatch) { | ||
| tflite::Interpreter interpreter; | ||
| int tensor_index, tensor_len = 5; | ||
| AddInterpreterInput(kTfLiteFloat32, tensor_len, tensor_index, | ||
| /*allocate_tensor=*/true, interpreter); | ||
| std::vector<float> values{1.0f, 2.0f, 3.0f, 4.0f}; | ||
| int values_len = values.size(); | ||
| Tensor tensor(ElementType::kFloat32, Tensor::Shape({values_len})); | ||
| std::memcpy(tensor.GetCpuWriteView().buffer<float>(), values.data(), | ||
| values_len * sizeof(float)); | ||
| absl::Status status = | ||
| CopyCpuInputIntoInterpreterTensor(tensor, interpreter, tensor_index); | ||
| EXPECT_FALSE(status.ok()); | ||
| EXPECT_THAT(status.message(), | ||
| HasSubstr("Interpreter's input size do not match the input " | ||
| "tensor's size for index")); | ||
| } | ||
|
|
||
| TEST(InferenceCalculatorUtilsTest, | ||
| CopyCpuInputIntoInterpreterTensorNullBuffer) { | ||
| tflite::Interpreter interpreter; | ||
| int tensor_index, tensor_len = 4; | ||
| // Make TFLite interpreter's buffer null. | ||
| AddInterpreterInput(kTfLiteFloat32, tensor_len, tensor_index, | ||
| /*allocate_tensor=*/false, interpreter); | ||
| std::vector<float> values{1.0f, 2.0f, 3.0f, 4.0f}; | ||
| int values_len = values.size(); | ||
| Tensor tensor(ElementType::kFloat32, Tensor::Shape({values_len})); | ||
| std::memcpy(tensor.GetCpuWriteView().buffer<float>(), values.data(), | ||
| values_len * sizeof(float)); | ||
| absl::Status status = | ||
| CopyCpuInputIntoInterpreterTensor(tensor, interpreter, tensor_index); | ||
| EXPECT_FALSE(status.ok()); | ||
| EXPECT_THAT(status.message(), | ||
| HasSubstr("Interpreter's input tensor buffer is null")); | ||
| } | ||
|
|
||
| TEST(InferenceCalculatorUtilsTest, | ||
| CopyCpuInputIntoInterpreterTensorUnsupportedType) { | ||
| tflite::Interpreter interpreter; | ||
| int tensor_index, tensor_len = 4; | ||
| AddInterpreterInput(kTfLiteUInt8, tensor_len, tensor_index, | ||
| /*allocate_tensor=*/true, interpreter); | ||
| std::vector<uint8_t> values{1, 2, 3, 4}; | ||
| int values_len = values.size(); | ||
| Tensor tensor(ElementType::kUInt8, Tensor::Shape({values_len})); | ||
| std::memcpy(tensor.GetCpuWriteView().buffer<uint8_t>(), values.data(), | ||
| values_len * sizeof(uint8_t)); | ||
| absl::Status status = | ||
| CopyCpuInputIntoInterpreterTensor(tensor, interpreter, tensor_index); | ||
| EXPECT_FALSE(status.ok()); | ||
| EXPECT_THAT(status.message(), HasSubstr("Unsupported input data type:")); | ||
| } | ||
|
|
||
| } // namespace | ||
| } // namespace mediapipe |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters