shufflenetv2.py 7.27 KB
Newer Older
Bar's avatar
Bar committed
1
2
import torch
import torch.nn as nn
ekka's avatar
ekka committed
3
from .utils import load_state_dict_from_url
Bar's avatar
Bar committed
4

5
6
7
8
9

__all__ = [
    'ShuffleNetV2', 'shufflenet_v2_x0_5', 'shufflenet_v2_x1_0',
    'shufflenet_v2_x1_5', 'shufflenet_v2_x2_0'
]
Bar's avatar
Bar committed
10
11

model_urls = {
12
13
    'shufflenetv2_x0.5': 'https://download.pytorch.org/models/shufflenetv2_x0.5-f707e7126e.pth',
    'shufflenetv2_x1.0': 'https://download.pytorch.org/models/shufflenetv2_x1-5666bf0f80.pth',
Bar's avatar
Bar committed
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
    'shufflenetv2_x1.5': None,
    'shufflenetv2_x2.0': None,
}


def channel_shuffle(x, groups):
    batchsize, num_channels, height, width = x.data.size()
    channels_per_group = num_channels // groups

    # reshape
    x = x.view(batchsize, groups,
               channels_per_group, height, width)

    x = torch.transpose(x, 1, 2).contiguous()

    # flatten
    x = x.view(batchsize, -1, height, width)

    return x


class InvertedResidual(nn.Module):
    def __init__(self, inp, oup, stride):
        super(InvertedResidual, self).__init__()

        if not (1 <= stride <= 3):
            raise ValueError('illegal stride value')
        self.stride = stride

        branch_features = oup // 2
        assert (self.stride != 1) or (inp == branch_features << 1)

        if self.stride > 1:
            self.branch1 = nn.Sequential(
48
                self.depthwise_conv(inp, inp, kernel_size=3, stride=self.stride, padding=1),
Bar's avatar
Bar committed
49
                nn.BatchNorm2d(inp),
50
                nn.Conv2d(inp, branch_features, kernel_size=1, stride=1, padding=0, bias=False),
Bar's avatar
Bar committed
51
52
53
54
55
                nn.BatchNorm2d(branch_features),
                nn.ReLU(inplace=True),
            )

        self.branch2 = nn.Sequential(
56
57
            nn.Conv2d(inp if (self.stride > 1) else branch_features,
                      branch_features, kernel_size=1, stride=1, padding=0, bias=False),
Bar's avatar
Bar committed
58
59
            nn.BatchNorm2d(branch_features),
            nn.ReLU(inplace=True),
60
            self.depthwise_conv(branch_features, branch_features, kernel_size=3, stride=self.stride, padding=1),
Bar's avatar
Bar committed
61
            nn.BatchNorm2d(branch_features),
62
            nn.Conv2d(branch_features, branch_features, kernel_size=1, stride=1, padding=0, bias=False),
Bar's avatar
Bar committed
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
            nn.BatchNorm2d(branch_features),
            nn.ReLU(inplace=True),
        )

    @staticmethod
    def depthwise_conv(i, o, kernel_size, stride=1, padding=0, bias=False):
        return nn.Conv2d(i, o, kernel_size, stride, padding, bias=bias, groups=i)

    def forward(self, x):
        if self.stride == 1:
            x1, x2 = x.chunk(2, dim=1)
            out = torch.cat((x1, self.branch2(x2)), dim=1)
        else:
            out = torch.cat((self.branch1(x), self.branch2(x)), dim=1)

        out = channel_shuffle(out, 2)

        return out


class ShuffleNetV2(nn.Module):
Bar's avatar
Bar committed
84
    def __init__(self, stages_repeats, stages_out_channels, num_classes=1000):
Bar's avatar
Bar committed
85
86
        super(ShuffleNetV2, self).__init__()

Bar's avatar
Bar committed
87
88
89
90
91
        if len(stages_repeats) != 3:
            raise ValueError('expected stages_repeats as list of 3 positive ints')
        if len(stages_out_channels) != 5:
            raise ValueError('expected stages_out_channels as list of 5 positive ints')
        self._stage_out_channels = stages_out_channels
ekka's avatar
ekka committed
92

Bar's avatar
Bar committed
93
94
        input_channels = 3
        output_channels = self._stage_out_channels[0]
Bar's avatar
Bar committed
95
96
97
98
99
100
101
102
103
104
105
        self.conv1 = nn.Sequential(
            nn.Conv2d(input_channels, output_channels, 3, 2, 1, bias=False),
            nn.BatchNorm2d(output_channels),
            nn.ReLU(inplace=True),
        )
        input_channels = output_channels

        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)

        stage_names = ['stage{}'.format(i) for i in [2, 3, 4]]
        for name, repeats, output_channels in zip(
Bar's avatar
Bar committed
106
                stage_names, stages_repeats, self._stage_out_channels[1:]):
Bar's avatar
Bar committed
107
108
109
110
111
112
            seq = [InvertedResidual(input_channels, output_channels, 2)]
            for i in range(repeats - 1):
                seq.append(InvertedResidual(output_channels, output_channels, 1))
            setattr(self, name, nn.Sequential(*seq))
            input_channels = output_channels

Bar's avatar
Bar committed
113
        output_channels = self._stage_out_channels[-1]
Bar's avatar
Bar committed
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
        self.conv5 = nn.Sequential(
            nn.Conv2d(input_channels, output_channels, 1, 1, 0, bias=False),
            nn.BatchNorm2d(output_channels),
            nn.ReLU(inplace=True),
        )

        self.fc = nn.Linear(output_channels, num_classes)

    def forward(self, x):
        x = self.conv1(x)
        x = self.maxpool(x)
        x = self.stage2(x)
        x = self.stage3(x)
        x = self.stage4(x)
        x = self.conv5(x)
        x = x.mean([2, 3])  # globalpool
        x = self.fc(x)
        return x


Bar's avatar
Bar committed
134
135
def _shufflenetv2(arch, pretrained, progress, *args, **kwargs):
    model = ShuffleNetV2(*args, **kwargs)
Bar's avatar
Bar committed
136
137

    if pretrained:
ekka's avatar
ekka committed
138
        model_url = model_urls[arch]
Bar's avatar
Bar committed
139
        if model_url is None:
ekka's avatar
ekka committed
140
141
            raise NotImplementedError('pretrained {} is not supported as of now'.format(arch))
        else:
142
            state_dict = load_state_dict_from_url(model_url, progress=progress)
ekka's avatar
ekka committed
143
            model.load_state_dict(state_dict)
Bar's avatar
Bar committed
144
145
146
147

    return model


148
def shufflenet_v2_x0_5(pretrained=False, progress=True, **kwargs):
149
150
151
152
153
154
155
156
157
    """
    Constructs a ShuffleNetV2 with 0.5x output channels, as described in
    `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design"
    <https://arxiv.org/abs/1807.11164>`_.

    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
        progress (bool): If True, displays a progress bar of the download to stderr
    """
Bar's avatar
Bar committed
158
159
    return _shufflenetv2('shufflenetv2_x0.5', pretrained, progress,
                         [4, 8, 4], [24, 48, 96, 192, 1024], **kwargs)
Bar's avatar
Bar committed
160
161


162
def shufflenet_v2_x1_0(pretrained=False, progress=True, **kwargs):
163
164
165
166
167
168
169
170
171
    """
    Constructs a ShuffleNetV2 with 1.0x output channels, as described in
    `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design"
    <https://arxiv.org/abs/1807.11164>`_.

    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
        progress (bool): If True, displays a progress bar of the download to stderr
    """
Bar's avatar
Bar committed
172
173
    return _shufflenetv2('shufflenetv2_x1.0', pretrained, progress,
                         [4, 8, 4], [24, 116, 232, 464, 1024], **kwargs)
Bar's avatar
Bar committed
174
175


176
def shufflenet_v2_x1_5(pretrained=False, progress=True, **kwargs):
177
178
179
180
181
182
183
184
185
    """
    Constructs a ShuffleNetV2 with 1.5x output channels, as described in
    `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design"
    <https://arxiv.org/abs/1807.11164>`_.

    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
        progress (bool): If True, displays a progress bar of the download to stderr
    """
Bar's avatar
Bar committed
186
187
    return _shufflenetv2('shufflenetv2_x1.5', pretrained, progress,
                         [4, 8, 4], [24, 176, 352, 704, 1024], **kwargs)
Bar's avatar
Bar committed
188
189


190
def shufflenet_v2_x2_0(pretrained=False, progress=True, **kwargs):
191
192
193
194
195
196
197
198
199
    """
    Constructs a ShuffleNetV2 with 2.0x output channels, as described in
    `"ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design"
    <https://arxiv.org/abs/1807.11164>`_.

    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
        progress (bool): If True, displays a progress bar of the download to stderr
    """
Bar's avatar
Bar committed
200
201
    return _shufflenetv2('shufflenetv2_x2.0', pretrained, progress,
                         [4, 8, 4], [24, 244, 488, 976, 2048], **kwargs)