conv_transpose_op.cc 9.2 KB
Newer Older
C
chengduoZH 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserve.

   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. */

C
chengduoZH 已提交
15
#include "paddle/operators/conv_transpose_op.h"
C
chengduoZH 已提交
16 17 18 19

namespace paddle {
namespace operators {

C
chengduoZH 已提交
20
void ConvTransposeOp::InferShape(framework::InferShapeContext* ctx) const {
C
chengduoZH 已提交
21
  PADDLE_ENFORCE(ctx->HasInput("Input"),
C
chengduoZH 已提交
22
                 "Input(Input) of ConvTransposeOp should not be null.");
C
chengduoZH 已提交
23
  PADDLE_ENFORCE(ctx->HasInput("Filter"),
C
chengduoZH 已提交
24
                 "Input(Filter) of ConvTransposeOp should not be null.");
C
chengduoZH 已提交
25
  PADDLE_ENFORCE(ctx->HasOutput("Output"),
C
chengduoZH 已提交
26
                 "Output(Output) of ConvTransposeOp should not be null.");
C
chengduoZH 已提交
27 28 29 30 31 32

  auto in_dims = ctx->GetInputDim("Input");
  auto filter_dims = ctx->GetInputDim("Filter");
  std::vector<int> strides = ctx->Attrs().Get<std::vector<int>>("strides");
  std::vector<int> paddings = ctx->Attrs().Get<std::vector<int>>("paddings");

C
chengduoZH 已提交
33 34 35 36 37 38 39 40 41 42 43
  PADDLE_ENFORCE(in_dims.size() == 4 || in_dims.size() == 5,
                 "ConvTransposeOp intput should be 4-D or 5-D tensor.");
  PADDLE_ENFORCE_EQ(in_dims.size(), filter_dims.size(),
                    "ConvTransposeOp input dimension and filter dimension "
                    "should be the same.");
  PADDLE_ENFORCE(in_dims.size() - strides.size() == 2U,
                 "ConvTransposeOp input dimension and strides dimension should "
                 "be consistent.");
  PADDLE_ENFORCE_EQ(paddings.size(), strides.size(),
                    "ConvTransposeOp paddings dimension and Conv strides "
                    "dimension should be the same.");
C
chengduoZH 已提交
44 45 46
  PADDLE_ENFORCE_EQ(in_dims[1], filter_dims[0],
                    "In ConvTransposeOp, The input channel should be the same "
                    "as the number of filters.");
C
chengduoZH 已提交
47

C
chengduoZH 已提交
48
  std::vector<int64_t> output_shape({in_dims[0], filter_dims[1]});
C
chengduoZH 已提交
49
  for (size_t i = 0; i < strides.size(); ++i) {
C
chengduoZH 已提交
50
    output_shape.push_back((in_dims[i + 2] - 1) * strides[i] - 2 * paddings[i] +
C
chengduoZH 已提交
51 52
                           filter_dims[i + 2]);
  }
C
chengduoZH 已提交
53
  ctx->SetOutputDim("Output", framework::make_ddim(output_shape));
C
chengduoZH 已提交
54 55
}

C
chengduoZH 已提交
56 57 58 59 60 61 62
Conv2DTransposeOpMaker::Conv2DTransposeOpMaker(
    framework::OpProto* proto, framework::OpAttrChecker* op_checker)
    : OpProtoAndCheckerMaker(proto, op_checker) {
  AddInput(
      "Input",
      "(Tensor) The input tensor of convolution transpose operator. "
      "The format of input tensor is NCHW. Where N is batch size, C is the "
C
chengduoZH 已提交
63 64
      "number of input channels, H is the height of the feature, and "
      "W is the width of the feature.");
C
chengduoZH 已提交
65
  AddInput("Filter",
C
chengduoZH 已提交
66
           "(Tensor) The filter tensor of convolution transpose operator. "
C
chengduoZH 已提交
67 68
           "The format of the filter tensor is CMHW, where C is the number of "
           "output image channels, M is the number of input image channels, "
C
chengduoZH 已提交
69
           "H is the height of the filter, and W is the width of the filter. "
C
chengduoZH 已提交
70
           "We enforce groups number == 1 and padding == 0 in "
C
chengduoZH 已提交
71
           "the convolution transpose scenario.");
C
chengduoZH 已提交
72
  AddOutput("Output",
C
chengduoZH 已提交
73
            "(Tensor) The output tensor of convolution transpose operator. "
C
chengduoZH 已提交
74
            "The format of output tensor is also NCHW.");
C
chengduoZH 已提交
75 76
  AddAttr<std::vector<int>>(
      "strides",
77
      "(vector<int> default:{1, 1}), the strides(h_stride, w_stride) of "
78
      "convolution transpose operator.")
C
chengduoZH 已提交
79
      .SetDefault({1, 1});
C
chengduoZH 已提交
80 81
  AddAttr<std::vector<int>>(
      "paddings",
82
      "(vector<int> default:{0, 0}), the paddings(h_pad, w_pad) of convolution "
C
chengduoZH 已提交
83
      "transpose operator.")
C
chengduoZH 已提交
84 85
      .SetDefault({0, 0});
  AddComment(R"DOC(
C
chengduoZH 已提交
86 87
Convolution2D Transpose Operator.

C
chengduoZH 已提交
88 89 90
The convolution transpose operation calculates the output based on the input, filter
and strides, paddings, groups parameters. The size of each dimension of the
parameters is checked in the infer-shape.
C
chengduoZH 已提交
91 92

Input(Input, Filter) and output(Output) are in NCHW format. Where N is batch
C
chengduoZH 已提交
93 94
size, C is the number of channels, H is the height of the feature, and 
W is the width of the feature. Parameters(ksize, strides, paddings) are two elements.
C
chengduoZH 已提交
95 96 97 98 99 100 101 102 103
These two elements represent height and width, respectively.
The input(X) size and output(Out) size may be different.
Example:
  Input:
       Input shape: (N, C_in, H_in, W_in)
       Filter shape: (C_in, C_out, H_f, W_f)
  Output:
       Output shape: (N, C_out, H_out, W_out)
  where
104 105
       H_out = (H_in - 1) * strides[0] - 2 * paddings[0] + H_f;
       W_out = (W_in - 1) * strides[1] - 2 * paddings[1] + W_f;
C
chengduoZH 已提交
106 107 108
)DOC");
}

C
chengduoZH 已提交
109 110 111
Conv3DTransposeOpMaker::Conv3DTransposeOpMaker(
    framework::OpProto* proto, framework::OpAttrChecker* op_checker)
    : OpProtoAndCheckerMaker(proto, op_checker) {
C
chengduoZH 已提交
112 113 114 115 116 117
  AddInput("Input",
           "(Tensor) The input tensor of convolution transpose operator."
           "The format of input tensor is NCDHW. Where N is batch size, C is "
           "the number of channels, D is the depth of the feature, H is the "
           "height of the feature, and "
           "W is the width of the feature.");
C
chengduoZH 已提交
118 119 120
  AddInput("Filter",
           "(Tensor) The filter tensor of convolution transpose operator."
           "The format of the filter tensor is CMDHW, where C is the number of "
C
chengduoZH 已提交
121 122 123
           "output image channels, M is the number of input image channels, D "
           "is the depth of the filter, H is the height of the filter, and "
           "W is the width of the filter."
C
chengduoZH 已提交
124
           "We enforce groups number == 1 and padding == 0 in "
C
chengduoZH 已提交
125
           "the convolution3d transpose scenario.");
C
chengduoZH 已提交
126 127 128 129
  AddOutput("Output",
            "(Tensor) The output tensor of convolution transpose operator."
            "The format of output tensor is also NCDHW."
            "Where N is batch size, C is "
C
chengduoZH 已提交
130 131
            "the number of channels, D is the depth of the feature, H is the "
            "height of the feature, and W is the width of the feature.");
C
chengduoZH 已提交
132
  AddAttr<std::vector<int>>("strides",
133
                            "(vector<int> default:{1, 1, 1}), the "
134
                            "strides{d_stride, h_stride, w_stride} of "
C
chengduoZH 已提交
135
                            "convolution transpose operator.")
C
chengduoZH 已提交
136
      .SetDefault({1, 1, 1});
C
chengduoZH 已提交
137
  AddAttr<std::vector<int>>("paddings",
138
                            "(vector<int> default:{0, 0, 0}), paddings(d_pad, "
C
chengduoZH 已提交
139
                            "h_pad, w_pad) of convolution transpose operator.")
C
chengduoZH 已提交
140 141
      .SetDefault({0, 0, 0});
  AddComment(R"DOC(
C
chengduoZH 已提交
142 143
Convolution3D Transpose Operator.

C
chengduoZH 已提交
144 145 146
The convolution transpose operation calculates the output based on the input, filter
and strides, paddings, groups parameters. The size of each dimension of the
parameters is checked in the infer-shape.
C
chengduoZH 已提交
147 148

Input(Input, Filter) and output(Output) are in NCDHW format. Where N is batch
C
chengduoZH 已提交
149 150 151
size, C is the number of channels, D is the depth of the feature, 
H is the height of the feature, and W is the width of the feature. 
Parameters(ksize, strides, paddings) are three elements.
C
chengduoZH 已提交
152 153 154 155 156 157 158 159 160
These three elements represent depth, height and width, respectively.
The input(X) size and output(Out) size may be different.
Example:
  Input:
       Input shape: (N, C_in, D_in, H_in, W_in)
       Filter shape: (C_in, C_out, D_f, H_f, W_f)
  Output:
       Output shape: (N, C_out, D_out, H_out, W_out)
  where
161 162 163
       D_out = (D_in - 1) * strides[0] - 2 * paddings[0] + D_f;
       H_out = (H_in - 1) * strides[1] - 2 * paddings[1] + H_f;
       W_out = (W_in - 1) * strides[2] - 2 * paddings[2] + W_f;
C
chengduoZH 已提交
164 165 166
)DOC");
}

C
chengduoZH 已提交
167
void ConvTransposeOpGrad::InferShape(framework::InferShapeContext* ctx) const {
C
chengduoZH 已提交
168 169 170 171 172 173 174 175 176 177 178 179 180 181
  auto in_dims = ctx->GetInputDim("Input");
  auto filter_dims = ctx->GetInputDim("Filter");
  if (ctx->HasOutput(framework::GradVarName("Input"))) {
    ctx->SetOutputDim(framework::GradVarName("Input"), in_dims);
  }
  if (ctx->HasOutput(framework::GradVarName("Filter"))) {
    ctx->SetOutputDim(framework::GradVarName("Filter"), filter_dims);
  }
}

}  // namespace operators
}  // namespace paddle

namespace ops = paddle::operators;
C
chengduoZH 已提交
182

C
chengduoZH 已提交
183 184
REGISTER_OP(conv2d_transpose, ops::ConvTransposeOp, ops::Conv2DTransposeOpMaker,
            conv2d_transpose_grad, ops::ConvTransposeOpGrad);
C
chengduoZH 已提交
185 186

REGISTER_OP_CPU_KERNEL(
C
chengduoZH 已提交
187
    conv2d_transpose,
C
chengduoZH 已提交
188 189
    ops::GemmConvTransposeKernel<paddle::platform::CPUPlace, float>,
    ops::GemmConvTransposeKernel<paddle::platform::CPUPlace, double>);
C
chengduoZH 已提交
190
REGISTER_OP_CPU_KERNEL(
C
chengduoZH 已提交
191
    conv2d_transpose_grad,
C
chengduoZH 已提交
192 193
    ops::GemmConvTransposeGradKernel<paddle::platform::CPUPlace, float>,
    ops::GemmConvTransposeGradKernel<paddle::platform::CPUPlace, double>);
C
chengduoZH 已提交
194

C
chengduoZH 已提交
195 196
REGISTER_OP(conv3d_transpose, ops::ConvTransposeOp, ops::Conv3DTransposeOpMaker,
            conv3d_transpose_grad, ops::ConvTransposeOpGrad);
C
chengduoZH 已提交
197 198

REGISTER_OP_CPU_KERNEL(
C
chengduoZH 已提交
199
    conv3d_transpose,
C
chengduoZH 已提交
200 201
    ops::GemmConvTransposeKernel<paddle::platform::CPUPlace, float>,
    ops::GemmConvTransposeKernel<paddle::platform::CPUPlace, double>);
C
chengduoZH 已提交
202
REGISTER_OP_CPU_KERNEL(
C
chengduoZH 已提交
203
    conv3d_transpose_grad,
C
chengduoZH 已提交
204 205
    ops::GemmConvTransposeGradKernel<paddle::platform::CPUPlace, float>,
    ops::GemmConvTransposeGradKernel<paddle::platform::CPUPlace, double>);