depthwise_conv.cc | Explore in Territory

/* Copyright 2017 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/kernels/internal/optimized/integer_ops/depthwise_conv.h"

#include <stddef.h>
#include <stdint.h>

#include <memory>
#include <vector>

#include "tensorflow/lite/core/c/builtin_op_data.h"
#include "tensorflow/lite/core/c/c_api_types.h"
#include "tensorflow/lite/core/c/common.h"
#include "tensorflow/lite/kernels/cpu_backend_context.h"
#include "tensorflow/lite/kernels/internal/compatibility.h"
#include "tensorflow/lite/kernels/internal/optimized/cpu_check.h"
#include "tensorflow/lite/kernels/internal/optimized/depthwiseconv_multithread.h"
#include "tensorflow/lite/kernels/internal/optimized/integer_ops/depthwise_conv_hybrid.h"
#include "tensorflow/lite/kernels/internal/optimized/neon_check.h"
#include "tensorflow/lite/kernels/internal/quantization_util.h"
#include "tensorflow/lite/kernels/internal/reference/depthwiseconv_float.h"
#include "tensorflow/lite/kernels/internal/reference/depthwiseconv_uint8.h"
#include "tensorflow/lite/kernels/internal/reference/integer_ops/depthwise_conv.h"
#include "tensorflow/lite/kernels/internal/tensor.h"
#include "tensorflow/lite/kernels/internal/tensor_ctypes.h"
#include "tensorflow/lite/kernels/internal/tensor_utils.h"
#include "tensorflow/lite/kernels/internal/types.h"
#include "tensorflow/lite/kernels/kernel_util.h"
#include "tensorflow/lite/kernels/padding.h"

namespace tflite {
namespace ops {
namespace builtin {
namespace depthwise_conv {

constexpr int kInputTensor = …;
constexpr int kFilterTensor = …;
constexpr int kBiasTensor = …;
constexpr int kOutputTensor = …;

// This file has three implementation of DepthwiseConv.
enum KernelType { … };

const int kTensorNotAllocated = …;

struct OpData { … };

void* Init(TfLiteContext* context, const char* buffer, size_t length) { … }

void Free(TfLiteContext* context, void* buffer) { … }

TfLiteStatus Prepare(TfLiteContext* context, TfLiteNode* node) { … }

TfLiteStatus ComputeDepthMultiplier(TfLiteContext* context,
                                    const TfLiteTensor* input,
                                    const TfLiteTensor* filter,
                                    int16* depth_multiplier) { … }

template <KernelType kernel_type>
TfLiteStatus EvalFloat(TfLiteContext* context, TfLiteNode* node,
                       TfLiteDepthwiseConvParams* params, OpData* data,
                       const TfLiteTensor* input, const TfLiteTensor* filter,
                       const TfLiteTensor* bias, TfLiteTensor* output) { … }

template <KernelType kernel_type>
TfLiteStatus EvalQuantized(TfLiteContext* context, TfLiteNode* node,
                           TfLiteDepthwiseConvParams* params, OpData* data,
                           const TfLiteTensor* input,
                           const TfLiteTensor* filter, const TfLiteTensor* bias,
                           TfLiteTensor* output) { … }

template <KernelType kernel_type>
TfLiteStatus EvalQuantizedPerChannel(TfLiteContext* context, TfLiteNode* node,
                                     TfLiteDepthwiseConvParams* params,
                                     OpData* data, const TfLiteTensor* input,
                                     const TfLiteTensor* filter,
                                     const TfLiteTensor* bias,
                                     TfLiteTensor* output) { … }

TfLiteStatus EvalQuantizedPerChannel16x8(
    const TfLiteDepthwiseConvParams* params, const OpData* data,
    const TfLiteTensor* input, const TfLiteTensor* filter,
    const TfLiteTensor* bias, TfLiteTensor* output) { … }

template <KernelType kernel_type>
TfLiteStatus EvalHybridPerChannel(TfLiteContext* context, TfLiteNode* node,
                                  TfLiteDepthwiseConvParams* params,
                                  OpData* data, const TfLiteTensor* input,
                                  const TfLiteTensor* filter,
                                  const TfLiteTensor* bias,
                                  TfLiteTensor* output) { … }

template <KernelType kernel_type, TfLiteType input_type>
TfLiteStatus EvalImpl(TfLiteContext* context, TfLiteNode* node) { … }

template <KernelType kernel_type>
TfLiteStatus Eval(TfLiteContext* context, TfLiteNode* node) { … }

}  // namespace depthwise_conv

TfLiteRegistration* Register_DEPTHWISE_CONVOLUTION_REF() { … }

TfLiteRegistration* Register_DEPTHWISE_CONVOLUTION_GENERIC_OPT() { … }

TfLiteRegistration* Register_DEPTHWISE_CONVOLUTION_NEON_OPT() { … }

TfLiteRegistration* Register_DEPTHWISE_CONVOLUTION_NEON_OPT_UINT8() { … }

TfLiteRegistration* Register_DEPTHWISE_CONV_2D() { … }

// Warning: Clients using this variant are responsible for ensuring that their
// models only need the UINT8 type. TFLite's op registration mechanism doesn't
// yet allow for more nuanced registration mechanisms.
TfLiteRegistration* Register_DEPTHWISE_CONV_2D_UINT8() { … }

}  // namespace builtin
}  // namespace ops
}  // namespace tflite
chromium/third_party/tflite/src/tensorflow/lite/kernels/depthwise_conv.cc
