File size: 13,170 Bytes
1ab1a09
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
# copyright (c) 2020 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.

import paddle
import paddle.nn as nn
import paddle.nn.functional as F

from paddleseg.cvlibs import manager
from paddleseg.models import layers
from paddleseg.utils import utils

__all__ = [
    "ResNet18_vd", "ResNet34_vd", "ResNet50_vd", "ResNet101_vd", "ResNet152_vd"
]


class ConvBNLayer(nn.Layer):
    def __init__(self,
                 in_channels,
                 out_channels,
                 kernel_size,
                 stride=1,
                 dilation=1,
                 groups=1,
                 is_vd_mode=False,
                 act=None,
                 data_format='NCHW'):
        super(ConvBNLayer, self).__init__()
        if dilation != 1 and kernel_size != 3:
            raise RuntimeError("When the dilation isn't 1," \
                "the kernel_size should be 3.")

        self.is_vd_mode = is_vd_mode
        self._pool2d_avg = nn.AvgPool2D(
            kernel_size=2,
            stride=2,
            padding=0,
            ceil_mode=True,
            data_format=data_format)
        self._conv = nn.Conv2D(
            in_channels=in_channels,
            out_channels=out_channels,
            kernel_size=kernel_size,
            stride=stride,
            padding=(kernel_size - 1) // 2 \
                if dilation == 1 else dilation,
            dilation=dilation,
            groups=groups,
            bias_attr=False,
            data_format=data_format)

        self._batch_norm = layers.SyncBatchNorm(
            out_channels, data_format=data_format)
        self._act_op = layers.Activation(act=act)

    def forward(self, inputs):
        if self.is_vd_mode:
            inputs = self._pool2d_avg(inputs)
        y = self._conv(inputs)
        y = self._batch_norm(y)
        y = self._act_op(y)

        return y


class BottleneckBlock(nn.Layer):
    def __init__(self,
                 in_channels,
                 out_channels,
                 stride,
                 shortcut=True,
                 if_first=False,
                 dilation=1,
                 data_format='NCHW'):
        super(BottleneckBlock, self).__init__()

        self.data_format = data_format
        self.conv0 = ConvBNLayer(
            in_channels=in_channels,
            out_channels=out_channels,
            kernel_size=1,
            act='relu',
            data_format=data_format)

        self.dilation = dilation

        self.conv1 = ConvBNLayer(
            in_channels=out_channels,
            out_channels=out_channels,
            kernel_size=3,
            stride=stride,
            act='relu',
            dilation=dilation,
            data_format=data_format)
        self.conv2 = ConvBNLayer(
            in_channels=out_channels,
            out_channels=out_channels * 4,
            kernel_size=1,
            act=None,
            data_format=data_format)

        if not shortcut:
            self.short = ConvBNLayer(
                in_channels=in_channels,
                out_channels=out_channels * 4,
                kernel_size=1,
                stride=1,
                is_vd_mode=False if if_first or stride == 1 else True,
                data_format=data_format)

        self.shortcut = shortcut
        # NOTE: Use the wrap layer for quantization training
        self.add = layers.Add()
        self.relu = layers.Activation(act="relu")

    def forward(self, inputs):
        y = self.conv0(inputs)
        conv1 = self.conv1(y)
        conv2 = self.conv2(conv1)

        if self.shortcut:
            short = inputs
        else:
            short = self.short(inputs)

        y = self.add(short, conv2)
        y = self.relu(y)
        return y


class BasicBlock(nn.Layer):
    def __init__(self,
                 in_channels,
                 out_channels,
                 stride,
                 dilation=1,
                 shortcut=True,
                 if_first=False,
                 data_format='NCHW'):
        super(BasicBlock, self).__init__()
        self.conv0 = ConvBNLayer(
            in_channels=in_channels,
            out_channels=out_channels,
            kernel_size=3,
            stride=stride,
            dilation=dilation,
            act='relu',
            data_format=data_format)
        self.conv1 = ConvBNLayer(
            in_channels=out_channels,
            out_channels=out_channels,
            kernel_size=3,
            dilation=dilation,
            act=None,
            data_format=data_format)

        if not shortcut:
            self.short = ConvBNLayer(
                in_channels=in_channels,
                out_channels=out_channels,
                kernel_size=1,
                stride=1,
                is_vd_mode=False if if_first or stride == 1 else True,
                data_format=data_format)

        self.shortcut = shortcut
        self.dilation = dilation
        self.data_format = data_format
        self.add = layers.Add()
        self.relu = layers.Activation(act="relu")

    def forward(self, inputs):
        y = self.conv0(inputs)
        conv1 = self.conv1(y)

        if self.shortcut:
            short = inputs
        else:
            short = self.short(inputs)
        y = self.add(short, conv1)
        y = self.relu(y)

        return y


class ResNet_vd(nn.Layer):
    """
    The ResNet_vd implementation based on PaddlePaddle.

    The original article refers to Jingdong
    Tong He, et, al. "Bag of Tricks for Image Classification with Convolutional Neural Networks"
    (https://arxiv.org/pdf/1812.01187.pdf).

    Args:
        layers (int, optional): The layers of ResNet_vd. The supported layers are (18, 34, 50, 101, 152, 200). Default: 50.
        output_stride (int, optional): The stride of output features compared to input images. It is 8 or 16. Default: 8.
        multi_grid (tuple|list, optional): The grid of stage4. Defult: (1, 1, 1).
        pretrained (str, optional): The path of pretrained model.

    """

    def __init__(self,
                 layers=50,
                 output_stride=8,
                 multi_grid=(1, 1, 1),
                 pretrained=None,
                 data_format='NCHW'):
        super(ResNet_vd, self).__init__()

        self.data_format = data_format
        self.conv1_logit = None  # for gscnn shape stream
        self.layers = layers
        supported_layers = [18, 34, 50, 101, 152, 200]
        assert layers in supported_layers, \
            "supported layers are {} but input layer is {}".format(
                supported_layers, layers)

        if layers == 18:
            depth = [2, 2, 2, 2]
        elif layers == 34 or layers == 50:
            depth = [3, 4, 6, 3]
        elif layers == 101:
            depth = [3, 4, 23, 3]
        elif layers == 152:
            depth = [3, 8, 36, 3]
        elif layers == 200:
            depth = [3, 12, 48, 3]
        num_channels = [64, 256, 512,
                        1024] if layers >= 50 else [64, 64, 128, 256]
        num_filters = [64, 128, 256, 512]

        # for channels of four returned stages
        self.feat_channels = [c * 4 for c in num_filters
                              ] if layers >= 50 else num_filters

        dilation_dict = None
        if output_stride == 8:
            dilation_dict = {2: 2, 3: 4}
        elif output_stride == 16:
            dilation_dict = {3: 2}

        self.conv1_1 = ConvBNLayer(
            in_channels=3,
            out_channels=32,
            kernel_size=3,
            stride=2,
            act='relu',
            data_format=data_format)
        self.conv1_2 = ConvBNLayer(
            in_channels=32,
            out_channels=32,
            kernel_size=3,
            stride=1,
            act='relu',
            data_format=data_format)
        self.conv1_3 = ConvBNLayer(
            in_channels=32,
            out_channels=64,
            kernel_size=3,
            stride=1,
            act='relu',
            data_format=data_format)
        self.pool2d_max = nn.MaxPool2D(
            kernel_size=3, stride=2, padding=1, data_format=data_format)

        # self.block_list = []
        self.stage_list = []
        if layers >= 50:
            for block in range(len(depth)):
                shortcut = False
                block_list = []
                for i in range(depth[block]):
                    if layers in [101, 152] and block == 2:
                        if i == 0:
                            conv_name = "res" + str(block + 2) + "a"
                        else:
                            conv_name = "res" + str(block + 2) + "b" + str(i)
                    else:
                        conv_name = "res" + str(block + 2) + chr(97 + i)

                    ###############################################################################
                    # Add dilation rate for some segmentation tasks, if dilation_dict is not None.
                    dilation_rate = dilation_dict[
                        block] if dilation_dict and block in dilation_dict else 1

                    # Actually block here is 'stage', and i is 'block' in 'stage'
                    # At the stage 4, expand the the dilation_rate if given multi_grid
                    if block == 3:
                        dilation_rate = dilation_rate * multi_grid[i]
                    ###############################################################################

                    bottleneck_block = self.add_sublayer(
                        'bb_%d_%d' % (block, i),
                        BottleneckBlock(
                            in_channels=num_channels[block]
                            if i == 0 else num_filters[block] * 4,
                            out_channels=num_filters[block],
                            stride=2 if i == 0 and block != 0 and
                            dilation_rate == 1 else 1,
                            shortcut=shortcut,
                            if_first=block == i == 0,
                            dilation=dilation_rate,
                            data_format=data_format))

                    block_list.append(bottleneck_block)
                    shortcut = True
                self.stage_list.append(block_list)
        else:
            for block in range(len(depth)):
                shortcut = False
                block_list = []
                for i in range(depth[block]):
                    dilation_rate = dilation_dict[block] \
                        if dilation_dict and block in dilation_dict else 1
                    if block == 3:
                        dilation_rate = dilation_rate * multi_grid[i]

                    basic_block = self.add_sublayer(
                        'bb_%d_%d' % (block, i),
                        BasicBlock(
                            in_channels=num_channels[block]
                            if i == 0 else num_filters[block],
                            out_channels=num_filters[block],
                            stride=2 if i == 0 and block != 0 \
                                and dilation_rate == 1 else 1,
                            dilation=dilation_rate,
                            shortcut=shortcut,
                            if_first=block == i == 0,
                            data_format=data_format))
                    block_list.append(basic_block)
                    shortcut = True
                self.stage_list.append(block_list)

        self.pretrained = pretrained
        self.init_weight()

    def forward(self, inputs):
        y = self.conv1_1(inputs)
        y = self.conv1_2(y)
        y = self.conv1_3(y)
        self.conv1_logit = y.clone()
        y = self.pool2d_max(y)

        # A feature list saves the output feature map of each stage.
        feat_list = []
        for stage in self.stage_list:
            for block in stage:
                y = block(y)
            feat_list.append(y)

        return feat_list

    def init_weight(self):
        utils.load_pretrained_model(self, self.pretrained)


@manager.BACKBONES.add_component
def ResNet18_vd(**args):
    model = ResNet_vd(layers=18, **args)
    return model


def ResNet34_vd(**args):
    model = ResNet_vd(layers=34, **args)
    return model


@manager.BACKBONES.add_component
def ResNet50_vd(**args):
    model = ResNet_vd(layers=50, **args)
    return model


@manager.BACKBONES.add_component
def ResNet101_vd(**args):
    model = ResNet_vd(layers=101, **args)
    return model


def ResNet152_vd(**args):
    model = ResNet_vd(layers=152, **args)
    return model


def ResNet200_vd(**args):
    model = ResNet_vd(layers=200, **args)
    return model