File size: 8,075 Bytes
2a8871c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import torch.nn as nn
from basicsr.utils.registry import ARCH_REGISTRY


def conv3x3(inplanes, outplanes, stride=1):
    """A simple wrapper for 3x3 convolution with padding.

    Args:
        inplanes (int): Channel number of inputs.
        outplanes (int): Channel number of outputs.
        stride (int): Stride in convolution. Default: 1.
    """
    return nn.Conv2d(inplanes, outplanes, kernel_size=3, stride=stride, padding=1, bias=False)


class BasicBlock(nn.Module):
    """Basic residual block used in the ResNetArcFace architecture.

    Args:
        inplanes (int): Channel number of inputs.
        planes (int): Channel number of outputs.
        stride (int): Stride in convolution. Default: 1.
        downsample (nn.Module): The downsample module. Default: None.
    """
    expansion = 1  # output channel expansion ratio

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        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

    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.downsample is not None:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)

        return out


class IRBlock(nn.Module):
    """Improved residual block (IR Block) used in the ResNetArcFace architecture.

    Args:
        inplanes (int): Channel number of inputs.
        planes (int): Channel number of outputs.
        stride (int): Stride in convolution. Default: 1.
        downsample (nn.Module): The downsample module. Default: None.
        use_se (bool): Whether use the SEBlock (squeeze and excitation block). Default: True.
    """
    expansion = 1  # output channel expansion ratio

    def __init__(self, inplanes, planes, stride=1, downsample=None, use_se=True):
        super(IRBlock, self).__init__()
        self.bn0 = nn.BatchNorm2d(inplanes)
        self.conv1 = conv3x3(inplanes, inplanes)
        self.bn1 = nn.BatchNorm2d(inplanes)
        self.prelu = nn.PReLU()
        self.conv2 = conv3x3(inplanes, planes, stride)
        self.bn2 = nn.BatchNorm2d(planes)
        self.downsample = downsample
        self.stride = stride
        self.use_se = use_se
        if self.use_se:
            self.se = SEBlock(planes)

    def forward(self, x):
        residual = x
        out = self.bn0(x)
        out = self.conv1(out)
        out = self.bn1(out)
        out = self.prelu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        if self.use_se:
            out = self.se(out)

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

        out += residual
        out = self.prelu(out)

        return out


class Bottleneck(nn.Module):
    """Bottleneck block used in the ResNetArcFace architecture.

    Args:
        inplanes (int): Channel number of inputs.
        planes (int): Channel number of outputs.
        stride (int): Stride in convolution. Default: 1.
        downsample (nn.Module): The downsample module. Default: None.
    """
    expansion = 4  # output channel expansion ratio

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(Bottleneck, self).__init__()
        self.conv1 = nn.Conv2d(inplanes, 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, planes * self.expansion, kernel_size=1, bias=False)
        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

    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.downsample is not None:
            residual = self.downsample(x)

        out += residual
        out = self.relu(out)

        return out


class SEBlock(nn.Module):
    """The squeeze-and-excitation block (SEBlock) used in the IRBlock.

    Args:
        channel (int): Channel number of inputs.
        reduction (int): Channel reduction ration. Default: 16.
    """

    def __init__(self, channel, reduction=16):
        super(SEBlock, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool2d(1)  # pool to 1x1 without spatial information
        self.fc = nn.Sequential(
            nn.Linear(channel, channel // reduction), nn.PReLU(), nn.Linear(channel // reduction, channel),
            nn.Sigmoid())

    def forward(self, x):
        b, c, _, _ = x.size()
        y = self.avg_pool(x).view(b, c)
        y = self.fc(y).view(b, c, 1, 1)
        return x * y


@ARCH_REGISTRY.register()
class ResNetArcFace(nn.Module):
    """ArcFace with ResNet architectures.

    Ref: ArcFace: Additive Angular Margin Loss for Deep Face Recognition.

    Args:
        block (str): Block used in the ArcFace architecture.
        layers (tuple(int)): Block numbers in each layer.
        use_se (bool): Whether use the SEBlock (squeeze and excitation block). Default: True.
    """

    def __init__(self, block, layers, use_se=True):
        if block == 'IRBlock':
            block = IRBlock
        self.inplanes = 64
        self.use_se = use_se
        super(ResNetArcFace, self).__init__()

        self.conv1 = nn.Conv2d(1, 64, kernel_size=3, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(64)
        self.prelu = nn.PReLU()
        self.maxpool = nn.MaxPool2d(kernel_size=2, stride=2)
        self.layer1 = self._make_layer(block, 64, layers[0])
        self.layer2 = self._make_layer(block, 128, layers[1], stride=2)
        self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
        self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
        self.bn4 = nn.BatchNorm2d(512)
        self.dropout = nn.Dropout()
        self.fc5 = nn.Linear(512 * 8 * 8, 512)
        self.bn5 = nn.BatchNorm1d(512)

        # initialization
        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.xavier_normal_(m.weight)
            elif isinstance(m, nn.BatchNorm2d) or isinstance(m, nn.BatchNorm1d):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)
            elif isinstance(m, nn.Linear):
                nn.init.xavier_normal_(m.weight)
                nn.init.constant_(m.bias, 0)

    def _make_layer(self, block, planes, num_blocks, stride=1):
        downsample = None
        if stride != 1 or self.inplanes != planes * block.expansion:
            downsample = nn.Sequential(
                nn.Conv2d(self.inplanes, planes * block.expansion, kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(planes * block.expansion),
            )
        layers = []
        layers.append(block(self.inplanes, planes, stride, downsample, use_se=self.use_se))
        self.inplanes = planes
        for _ in range(1, num_blocks):
            layers.append(block(self.inplanes, planes, use_se=self.use_se))

        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.prelu(x)
        x = self.maxpool(x)

        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)
        x = self.bn4(x)
        x = self.dropout(x)
        x = x.view(x.size(0), -1)
        x = self.fc5(x)
        x = self.bn5(x)

        return x