File size: 7,493 Bytes
b911d20
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""ResNet in PyTorch.
ImageNet-Style ResNet
[1] Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun
    Deep Residual Learning for Image Recognition. arXiv:1512.03385
Adapted from: https://github.com/bearpaw/pytorch-classification
"""
import torch
import torch.nn as nn
import torch.nn.functional as F


class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, in_planes, planes, stride=1, is_last=False):
        super(BasicBlock, self).__init__()
        self.is_last = is_last
        self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)

        self.shortcut = nn.Sequential()
        if stride != 1 or in_planes != self.expansion * planes:
            self.shortcut = nn.Sequential(
                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(self.expansion * planes)
            )

    def forward(self, x):
        out = F.relu(self.bn1(self.conv1(x)))
        out = self.bn2(self.conv2(out))
        out += self.shortcut(x)
        preact = out
        out = F.relu(out)
        if self.is_last:
            return out, preact
        else:
            return out


class Bottleneck(nn.Module):
    expansion = 4

    def __init__(self, in_planes, planes, stride=1, is_last=False):
        super(Bottleneck, self).__init__()
        self.is_last = is_last
        self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)
        self.conv3 = nn.Conv2d(planes, self.expansion * planes, kernel_size=1, bias=False)
        self.bn3 = nn.BatchNorm2d(self.expansion * planes)

        self.shortcut = nn.Sequential()
        if stride != 1 or in_planes != self.expansion * planes:
            self.shortcut = nn.Sequential(
                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(self.expansion * planes)
            )

    def forward(self, x):
        out = F.relu(self.bn1(self.conv1(x)))
        out = F.relu(self.bn2(self.conv2(out)))
        out = self.bn3(self.conv3(out))
        out += self.shortcut(x)
        preact = out
        out = F.relu(out)
        if self.is_last:
            return out, preact
        else:
            return out


class ResNet(nn.Module):
    def __init__(self, block, num_blocks, in_channel=3, zero_init_residual=False, pool=False):
        super(ResNet, self).__init__()
        self.in_planes = 64

        if pool:
            self.conv1 = nn.Conv2d(in_channel, 64, kernel_size=7, stride=2, padding=3, bias=False)
        else:
            self.conv1 = nn.Conv2d(in_channel, 64, kernel_size=3, stride=1, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(64)

        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) if pool else nn.Identity()
        self.layer1 = self._make_layer(block, 64, num_blocks[0], stride=1)
        self.layer2 = self._make_layer(block, 128, num_blocks[1], stride=2)
        self.layer3 = self._make_layer(block, 256, num_blocks[2], stride=2)
        self.layer4 = self._make_layer(block, 512, num_blocks[3], stride=2)
        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
            elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)

        # Zero-initialize the last BN in each residual branch,
        # so that the residual branch starts with zeros, and each residual block behaves
        # like an identity. This improves the model by 0.2~0.3% according to:
        # https://arxiv.org/abs/1706.02677
        if zero_init_residual:
            for m in self.modules():
                if isinstance(m, Bottleneck):
                    nn.init.constant_(m.bn3.weight, 0)
                elif isinstance(m, BasicBlock):
                    nn.init.constant_(m.bn2.weight, 0)

    def _make_layer(self, block, planes, num_blocks, stride):
        strides = [stride] + [1] * (num_blocks - 1)
        layers = []
        for i in range(num_blocks):
            stride = strides[i]
            layers.append(block(self.in_planes, planes, stride))
            self.in_planes = planes * block.expansion
        return nn.Sequential(*layers)

    def forward(self, x, layer=100):
        out = self.maxpool(F.relu(self.bn1(self.conv1(x))))
        out = self.layer1(out)
        out = self.layer2(out)
        out = self.layer3(out)
        out = self.layer4(out)
        out = self.avgpool(out)
        out = torch.flatten(out, 1)
        return out


def resnet18(**kwargs):
    return ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)


def resnet34(**kwargs):
    return ResNet(BasicBlock, [3, 4, 6, 3], **kwargs)


def resnet50(**kwargs):
    return ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)


def resnet101(**kwargs):
    return ResNet(Bottleneck, [3, 4, 23, 3], **kwargs)


model_dict = {
    'resnet18': [resnet18, 512],
    'resnet34': [resnet34, 512],
    'resnet50': [resnet50, 2048],
    'resnet101': [resnet101, 2048],
}


class LinearBatchNorm(nn.Module):
    """Implements BatchNorm1d by BatchNorm2d, for SyncBN purpose"""

    def __init__(self, dim, affine=True):
        super(LinearBatchNorm, self).__init__()
        self.dim = dim
        self.bn = nn.BatchNorm2d(dim, affine=affine)

    def forward(self, x):
        x = x.view(-1, self.dim, 1, 1)
        x = self.bn(x)
        x = x.view(-1, self.dim)
        return x


class SupConResNet(nn.Module):
    """backbone + projection head"""

    def __init__(self, name='resnet50', head='mlp', feat_dim=128, pool=False):
        super(SupConResNet, self).__init__()
        model_fun, dim_in = model_dict[name]
        self.encoder = model_fun(pool=pool)
        if head == 'linear':
            self.head = nn.Linear(dim_in, feat_dim)
        elif head == 'mlp':
            self.head = nn.Sequential(
                nn.Linear(dim_in, dim_in),
                nn.ReLU(inplace=True),
                nn.Linear(dim_in, feat_dim)
            )
        else:
            raise NotImplementedError(
                'head not supported: {}'.format(head))

    def forward(self, x):
        feat = self.encoder(x)
        feat = F.normalize(self.head(feat), dim=1)
        return feat


class SupCEResNet(nn.Module):
    """encoder + classifier"""

    def __init__(self, name='resnet50', num_classes=10, pool=False):
        super(SupCEResNet, self).__init__()
        model_fun, dim_in = model_dict[name]
        self.encoder = model_fun(pool=pool)
        self.fc = nn.Linear(dim_in, num_classes)

    def forward(self, x):
        return self.fc(self.encoder(x))


class LinearClassifier(nn.Module):
    """Linear classifier"""

    def __init__(self, name='resnet50', num_classes=10):
        super(LinearClassifier, self).__init__()
        _, feat_dim = model_dict[name]
        self.fc = nn.Linear(feat_dim, num_classes)

    def forward(self, features):
        return self.fc(features)