File size: 11,661 Bytes
f549064
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# Copyright (c) OpenMMLab. All rights reserved.
import math

import torch
import torch.nn as nn
import torch.utils.checkpoint as cp
from mmcv.cnn import build_conv_layer, build_norm_layer
from mmengine.model import Sequential

from mmdet.registry import MODELS
from .resnet import Bottleneck as _Bottleneck
from .resnet import ResNet


class Bottle2neck(_Bottleneck):
    expansion = 4

    def __init__(self,
                 inplanes,
                 planes,
                 scales=4,
                 base_width=26,
                 base_channels=64,
                 stage_type='normal',
                 **kwargs):
        """Bottle2neck block for Res2Net.

        If style is "pytorch", the stride-two layer is the 3x3 conv layer, if
        it is "caffe", the stride-two layer is the first 1x1 conv layer.
        """
        super(Bottle2neck, self).__init__(inplanes, planes, **kwargs)
        assert scales > 1, 'Res2Net degenerates to ResNet when scales = 1.'
        width = int(math.floor(self.planes * (base_width / base_channels)))

        self.norm1_name, norm1 = build_norm_layer(
            self.norm_cfg, width * scales, postfix=1)
        self.norm3_name, norm3 = build_norm_layer(
            self.norm_cfg, self.planes * self.expansion, postfix=3)

        self.conv1 = build_conv_layer(
            self.conv_cfg,
            self.inplanes,
            width * scales,
            kernel_size=1,
            stride=self.conv1_stride,
            bias=False)
        self.add_module(self.norm1_name, norm1)

        if stage_type == 'stage' and self.conv2_stride != 1:
            self.pool = nn.AvgPool2d(
                kernel_size=3, stride=self.conv2_stride, padding=1)
        convs = []
        bns = []

        fallback_on_stride = False
        if self.with_dcn:
            fallback_on_stride = self.dcn.pop('fallback_on_stride', False)
        if not self.with_dcn or fallback_on_stride:
            for i in range(scales - 1):
                convs.append(
                    build_conv_layer(
                        self.conv_cfg,
                        width,
                        width,
                        kernel_size=3,
                        stride=self.conv2_stride,
                        padding=self.dilation,
                        dilation=self.dilation,
                        bias=False))
                bns.append(
                    build_norm_layer(self.norm_cfg, width, postfix=i + 1)[1])
            self.convs = nn.ModuleList(convs)
            self.bns = nn.ModuleList(bns)
        else:
            assert self.conv_cfg is None, 'conv_cfg must be None for DCN'
            for i in range(scales - 1):
                convs.append(
                    build_conv_layer(
                        self.dcn,
                        width,
                        width,
                        kernel_size=3,
                        stride=self.conv2_stride,
                        padding=self.dilation,
                        dilation=self.dilation,
                        bias=False))
                bns.append(
                    build_norm_layer(self.norm_cfg, width, postfix=i + 1)[1])
            self.convs = nn.ModuleList(convs)
            self.bns = nn.ModuleList(bns)

        self.conv3 = build_conv_layer(
            self.conv_cfg,
            width * scales,
            self.planes * self.expansion,
            kernel_size=1,
            bias=False)
        self.add_module(self.norm3_name, norm3)

        self.stage_type = stage_type
        self.scales = scales
        self.width = width
        delattr(self, 'conv2')
        delattr(self, self.norm2_name)

    def forward(self, x):
        """Forward function."""

        def _inner_forward(x):
            identity = x

            out = self.conv1(x)
            out = self.norm1(out)
            out = self.relu(out)

            if self.with_plugins:
                out = self.forward_plugin(out, self.after_conv1_plugin_names)

            spx = torch.split(out, self.width, 1)
            sp = self.convs[0](spx[0].contiguous())
            sp = self.relu(self.bns[0](sp))
            out = sp
            for i in range(1, self.scales - 1):
                if self.stage_type == 'stage':
                    sp = spx[i]
                else:
                    sp = sp + spx[i]
                sp = self.convs[i](sp.contiguous())
                sp = self.relu(self.bns[i](sp))
                out = torch.cat((out, sp), 1)

            if self.stage_type == 'normal' or self.conv2_stride == 1:
                out = torch.cat((out, spx[self.scales - 1]), 1)
            elif self.stage_type == 'stage':
                out = torch.cat((out, self.pool(spx[self.scales - 1])), 1)

            if self.with_plugins:
                out = self.forward_plugin(out, self.after_conv2_plugin_names)

            out = self.conv3(out)
            out = self.norm3(out)

            if self.with_plugins:
                out = self.forward_plugin(out, self.after_conv3_plugin_names)

            if self.downsample is not None:
                identity = self.downsample(x)

            out += identity

            return out

        if self.with_cp and x.requires_grad:
            out = cp.checkpoint(_inner_forward, x)
        else:
            out = _inner_forward(x)

        out = self.relu(out)

        return out


class Res2Layer(Sequential):
    """Res2Layer to build Res2Net style backbone.

    Args:
        block (nn.Module): block used to build ResLayer.
        inplanes (int): inplanes of block.
        planes (int): planes of block.
        num_blocks (int): number of blocks.
        stride (int): stride of the first block. Default: 1
        avg_down (bool): Use AvgPool instead of stride conv when
            downsampling in the bottle2neck. Default: False
        conv_cfg (dict): dictionary to construct and config conv layer.
            Default: None
        norm_cfg (dict): dictionary to construct and config norm layer.
            Default: dict(type='BN')
        scales (int): Scales used in Res2Net. Default: 4
        base_width (int): Basic width of each scale. Default: 26
    """

    def __init__(self,
                 block,
                 inplanes,
                 planes,
                 num_blocks,
                 stride=1,
                 avg_down=True,
                 conv_cfg=None,
                 norm_cfg=dict(type='BN'),
                 scales=4,
                 base_width=26,
                 **kwargs):
        self.block = block

        downsample = None
        if stride != 1 or inplanes != planes * block.expansion:
            downsample = nn.Sequential(
                nn.AvgPool2d(
                    kernel_size=stride,
                    stride=stride,
                    ceil_mode=True,
                    count_include_pad=False),
                build_conv_layer(
                    conv_cfg,
                    inplanes,
                    planes * block.expansion,
                    kernel_size=1,
                    stride=1,
                    bias=False),
                build_norm_layer(norm_cfg, planes * block.expansion)[1],
            )

        layers = []
        layers.append(
            block(
                inplanes=inplanes,
                planes=planes,
                stride=stride,
                downsample=downsample,
                conv_cfg=conv_cfg,
                norm_cfg=norm_cfg,
                scales=scales,
                base_width=base_width,
                stage_type='stage',
                **kwargs))
        inplanes = planes * block.expansion
        for i in range(1, num_blocks):
            layers.append(
                block(
                    inplanes=inplanes,
                    planes=planes,
                    stride=1,
                    conv_cfg=conv_cfg,
                    norm_cfg=norm_cfg,
                    scales=scales,
                    base_width=base_width,
                    **kwargs))
        super(Res2Layer, self).__init__(*layers)


@MODELS.register_module()
class Res2Net(ResNet):
    """Res2Net backbone.

    Args:
        scales (int): Scales used in Res2Net. Default: 4
        base_width (int): Basic width of each scale. Default: 26
        depth (int): Depth of res2net, from {50, 101, 152}.
        in_channels (int): Number of input image channels. Default: 3.
        num_stages (int): Res2net stages. Default: 4.
        strides (Sequence[int]): Strides of the first block of each stage.
        dilations (Sequence[int]): Dilation of each stage.
        out_indices (Sequence[int]): Output from which stages.
        style (str): `pytorch` or `caffe`. If set to "pytorch", the stride-two
            layer is the 3x3 conv layer, otherwise the stride-two layer is
            the first 1x1 conv layer.
        deep_stem (bool): Replace 7x7 conv in input stem with 3 3x3 conv
        avg_down (bool): Use AvgPool instead of stride conv when
            downsampling in the bottle2neck.
        frozen_stages (int): Stages to be frozen (stop grad and set eval mode).
            -1 means not freezing any parameters.
        norm_cfg (dict): Dictionary to construct and config norm layer.
        norm_eval (bool): Whether to set norm layers to eval mode, namely,
            freeze running stats (mean and var). Note: Effect on Batch Norm
            and its variants only.
        plugins (list[dict]): List of plugins for stages, each dict contains:

            - cfg (dict, required): Cfg dict to build plugin.
            - position (str, required): Position inside block to insert
              plugin, options are 'after_conv1', 'after_conv2', 'after_conv3'.
            - stages (tuple[bool], optional): Stages to apply plugin, length
              should be same as 'num_stages'.
        with_cp (bool): Use checkpoint or not. Using checkpoint will save some
            memory while slowing down the training speed.
        zero_init_residual (bool): Whether to use zero init for last norm layer
            in resblocks to let them behave as identity.
        pretrained (str, optional): model pretrained path. Default: None
        init_cfg (dict or list[dict], optional): Initialization config dict.
            Default: None

    Example:
        >>> from mmdet.models import Res2Net
        >>> import torch
        >>> self = Res2Net(depth=50, scales=4, base_width=26)
        >>> self.eval()
        >>> inputs = torch.rand(1, 3, 32, 32)
        >>> level_outputs = self.forward(inputs)
        >>> for level_out in level_outputs:
        ...     print(tuple(level_out.shape))
        (1, 256, 8, 8)
        (1, 512, 4, 4)
        (1, 1024, 2, 2)
        (1, 2048, 1, 1)
    """

    arch_settings = {
        50: (Bottle2neck, (3, 4, 6, 3)),
        101: (Bottle2neck, (3, 4, 23, 3)),
        152: (Bottle2neck, (3, 8, 36, 3))
    }

    def __init__(self,
                 scales=4,
                 base_width=26,
                 style='pytorch',
                 deep_stem=True,
                 avg_down=True,
                 pretrained=None,
                 init_cfg=None,
                 **kwargs):
        self.scales = scales
        self.base_width = base_width
        super(Res2Net, self).__init__(
            style='pytorch',
            deep_stem=True,
            avg_down=True,
            pretrained=pretrained,
            init_cfg=init_cfg,
            **kwargs)

    def make_res_layer(self, **kwargs):
        return Res2Layer(
            scales=self.scales,
            base_width=self.base_width,
            base_channels=self.base_channels,
            **kwargs)