-
Notifications
You must be signed in to change notification settings - Fork 6
/
resnet_with_block.py
271 lines (201 loc) · 8.12 KB
/
resnet_with_block.py
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
import torch.nn as nn
from .layer_blocks import SELayer, SRMLayer
from torchvision.models import ResNet
def conv3x3(in_planes, out_planes, stride=1):
return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
padding=1, bias=False)
def conv1x1(in_planes, out_planes, stride=1):
return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride,
bias=False)
def basic_block_factory(layer_block=None):
# Factory for using torchvision ResNet class
class BasicBlock(nn.Module):
expansion = 1
def __init__(self, inplanes, planes, stride=1, downsample=None,
reduction=16):
super(BasicBlock, self).__init__()
self.conv1 = conv3x3(inplanes, planes, stride)
self.bn1 = nn.BatchNorm2d(planes)
self.relu = nn.ReLU(inplace=True)
self.conv2 = conv3x3(planes, planes)
self.bn2 = nn.BatchNorm2d(planes)
self.downsample = downsample
self.stride = stride
if layer_block is not None:
self.layer_block = layer_block(planes, reduction)
else:
self.layer_block = None
def forward(self, x):
residual = x
out = self.conv1(x)
out = self.bn1(out)
out = self.relu(out)
out = self.conv2(out)
out = self.bn2(out)
if self.layer_block is not None:
out = self.layer_block(out)
if self.downsample is not None:
residual = self.downsample(x)
out += residual
out = self.relu(out)
return out
return BasicBlock
def bottleneck_factory(layer_block=None):
# Factory for using torchvision ResNet class
class Bottleneck(nn.Module):
expansion = 4
def __init__(self, inplanes, planes, stride=1, downsample=None,
reduction=16):
super(Bottleneck, self).__init__()
self.conv1 = conv1x1(inplanes, planes)
self.bn1 = nn.BatchNorm2d(planes)
self.conv2 = conv3x3(planes, planes, stride=stride)
self.bn2 = nn.BatchNorm2d(planes)
self.conv3 = conv1x1(planes, planes * self.expansion)
self.bn3 = nn.BatchNorm2d(planes * self.expansion)
self.relu = nn.ReLU(inplace=True)
self.downsample = downsample
self.stride = stride
if layer_block is not None:
self.layer_block = layer_block(planes * self.expansion,
reduction)
else:
self.layer_block = None
def forward(self, x):
residual = x
out = self.conv1(x)
out = self.bn1(out)
out = self.relu(out)
out = self.conv2(out)
out = self.bn2(out)
out = self.relu(out)
out = self.conv3(out)
out = self.bn3(out)
if self.layer_block is not None:
out = self.layer_block(out)
if self.downsample:
residual = self.downsample(x)
out += residual
out = self.relu(out)
return out
return Bottleneck
class CifarResNetWithBlock(nn.Module):
def __init__(self, n_size, num_classes=10, layer_block=None,
reduction=None):
super(CifarResNetWithBlock, self).__init__()
self.inplanes = 16
self.conv1 = conv3x3(3, self.inplanes, stride=1)
self.bn1 = nn.BatchNorm2d(self.inplanes)
self.relu = nn.ReLU(inplace=True)
self.block = basic_block_factory(layer_block=layer_block)
self.layer1 = self._make_layer(16, blocks=n_size, stride=1,
reduction=reduction)
self.layer2 = self._make_layer(32, blocks=n_size, stride=2,
reduction=reduction)
self.layer3 = self._make_layer(64, blocks=n_size, stride=2,
reduction=reduction)
self.avgpool = nn.AdaptiveAvgPool2d(1)
self.fc = nn.Linear(64, num_classes)
self.initialize()
def initialize(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_normal_(m.weight, nonlinearity='relu')
elif isinstance(m, nn.BatchNorm2d):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
def _make_layer(self, planes, blocks, stride, reduction):
downsample = None
layers = []
if stride != 1 or self.inplanes != planes:
downsample = nn.Sequential(conv1x1(self.inplanes, planes,
stride=stride),
nn.BatchNorm2d(planes))
layers.append(self.block(self.inplanes, planes, stride, downsample,
reduction))
self.inplanes = planes
for i in range(1, blocks):
layers.append(self.block(self.inplanes, planes,
reduction=reduction))
self.inplanes = planes
return nn.Sequential(*layers)
def forward(self, x):
x = self.conv1(x)
x = self.bn1(x)
x = self.relu(x)
x = self.layer1(x)
x = self.layer2(x)
x = self.layer3(x)
x = self.avgpool(x)
x = x.view(x.size(0), -1)
x = self.fc(x)
return x
# ImageNet ResNets34
def resnet34(num_classes=1000):
model = ResNet(basic_block_factory(), [3, 4, 6, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
def se_resnet34(num_classes=1000):
model = ResNet(basic_block_factory(layer_block=SELayer), [3, 4, 6, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
def srm_resnet34(num_classes=1000):
model = ResNet(basic_block_factory(layer_block=SRMLayer), [3, 4, 6, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
# ImageNet ResNets50
def resnet50(num_classes=1000):
model = ResNet(bottleneck_factory(), [3, 4, 6, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
def se_resnet50(num_classes=1000):
model = ResNet(bottleneck_factory(layer_block=SELayer), [3, 4, 6, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
def srm_resnet50(num_classes=1000):
model = ResNet(bottleneck_factory(layer_block=SRMLayer), [3, 4, 6, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
# ImageNet ResNets101
def resnet101(num_classes=1000):
model = ResNet(bottleneck_factory(), [3, 4, 23, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
def se_resnet101(num_classes=1000):
model = ResNet(bottleneck_factory(layer_block=SELayer), [3, 4, 23, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
def srm_resnet101(num_classes=1000):
model = ResNet(bottleneck_factory(layer_block=SRMLayer), [3, 4, 23, 3],
num_classes=num_classes)
model.avgpool = nn.AdaptiveAvgPool2d(1)
return model
# Cifar10 ResNets32
def cifar_resnet20(**kwargs):
model = CifarResNetWithBlock(3, **kwargs)
return model
def cifar_se_resnet20(**kwargs):
model = CifarResNetWithBlock(3, layer_block=SELayer, reduction=16, **kwargs)
return model
def cifar_srm_resnet20(**kwargs):
model = CifarResNetWithBlock(3, layer_block=SRMLayer, **kwargs)
return model
# Cifar10 ResNets32
def cifar_resnet32(**kwargs):
model = CifarResNetWithBlock(5, **kwargs)
return model
def cifar_se_resnet32(**kwargs):
model = CifarResNetWithBlock(5, layer_block=SELayer, reduction=16,
**kwargs)
return model
def cifar_srm_resnet32(**kwargs):
model = CifarResNetWithBlock(5, layer_block=SRMLayer, **kwargs)
return model