pool_op.cpp 2.4 KB
Newer Older
W
wangliu 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13
/* Copyright (c) 2018 PaddlePaddle 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. */
14

L
liuruilong 已提交
15 16
#ifdef POOL_OP

17
#include "pool_op.h"
18
#include "framework/op_proto_maker.h"
L
liuruilong 已提交
19
#include "framework/op_registry.h"
20 21 22 23 24 25

namespace paddle_mobile {
namespace operators {

int PoolOutputSize(int input_size, int filter_size, int padding, int stride,
                   bool ceil_mode) {
26 27 28 29 30 31 32 33
  int output_size;
  if (!ceil_mode) {
    output_size = (input_size - filter_size + 2 * padding) / stride + 1;
  } else {
    output_size =
        (input_size - filter_size + 2 * padding + stride - 1) / stride + 1;
  }
  return output_size;
34 35 36
}
template <typename DeviceType, typename T>
void PoolOp<DeviceType, T>::InferShape() const {
L
liuruilong 已提交
37 38 39 40 41
  auto in_x_dims = this->param_.Input()->dims();
  std::vector<int> ksize = this->param_.Ksize();
  std::vector<int> paddings = this->param_.Paddings();
  std::vector<int> strides = this->param_.Strides();
  bool ceil_mode = this->param_.isCeilMode();
42

L
liuruilong 已提交
43
  if (this->param_.isGlobalPooling()) {
44
    ksize.resize(static_cast<size_t>(in_x_dims.size()) - 2);
45
    for (size_t i = 0; i < ksize.size(); ++i) {
46 47
      paddings[i] = 0;
      ksize[i] = static_cast<int>(in_x_dims[i + 2]);
48
    }
49 50 51 52 53 54
  }
  std::vector<int64_t> output_shape({in_x_dims[0], in_x_dims[1]});
  for (size_t i = 0; i < ksize.size(); ++i) {
    output_shape.push_back(PoolOutputSize(in_x_dims[i + 2], ksize[i],
                                          paddings[i], strides[i], ceil_mode));
  }
L
liuruilong 已提交
55
  this->param_.Output()->Resize(framework::make_ddim(output_shape));
56 57
}
template class PoolOp<CPU, float>;
朔-望's avatar
朔-望 已提交
58 59
}  // namespace operators
}  // namespace paddle_mobile
E
eclipsess 已提交
60 61

namespace ops = paddle_mobile::operators;
L
for  
liuruilong 已提交
62 63 64 65 66
#ifdef PADDLE_MOBILE_CPU
USE_OP_CPU(pool2d);
REGISTER_OPERATOR_CPU(pool2d, ops::PoolOp);
#endif
#ifdef PADDLE_MOBILE_MALI_GPU
S
sharper 已提交
67 68
USE_OP_MALI_GPU(pool2d);
REGISTER_OPERATOR_MALI_GPU(pool2d, ops::PoolOp);
L
for  
liuruilong 已提交
69 70 71
#endif
#ifdef PADDLE_MOBILE_FPGA
#endif
L
liuruilong 已提交
72 73

#endif