squeezenet1_0代码理解

def squeezenet1_0(pretrained=False, progress=True, **kwargs):
    r"""SqueezeNet model architecture from the `"SqueezeNet: AlexNet-level
    accuracy with 50x fewer parameters and <0.5MB model size"
    <https://arxiv.org/abs/1602.07360>`_ paper.

    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
    """
    return _squeezenet('1_0', pretrained, progress, **kwargs)

    return _squeezenet('1_1', pretrained, progress, **kwargs)

def _squeezenet(version, pretrained, progress, **kwargs):
    model = SqueezeNet(version, **kwargs)
    if pretrained:
        arch = 'squeezenet' + version
        state_dict = load_state_dict_from_url(model_urls[arch],
                                              progress=progress)
        model.load_state_dict(state_dict)
    return model

class SqueezeNet(nn.Module):

    def __init__(self, version='1_0', num_classes=1000):
        super(SqueezeNet, self).__init__()
        self.num_classes = num_classes
        if version == '1_0':
            self.features = nn.Sequential(
                nn.Conv2d(3, 96, kernel_size=7, stride=2),  # 7x7卷积
                nn.ReLU(inplace=True),
                nn.MaxPool2d(kernel_size=3, stride=2, ceil_mode=True),  # 最大池化
                Fire(96, 16, 64, 64),
                Fire(128, 16, 64, 64),
                Fire(128, 32, 128, 128),
                nn.MaxPool2d(kernel_size=3, stride=2, ceil_mode=True),  # 最大池化
                Fire(256, 32, 128, 128),
                Fire(256, 48, 192, 192),
                Fire(384, 48, 192, 192),
                Fire(384, 64, 256, 256),
                nn.MaxPool2d(kernel_size=3, stride=2, ceil_mode=True),  # 最大池化
                Fire(512, 64, 256, 256),
            )
        elif version == '1_1':
            self.features = nn.Sequential(
                nn.Conv2d(3, 64, kernel_size=3, stride=2),  # 3x3卷积
                nn.ReLU(inplace=True),
                nn.MaxPool2d(kernel_size=3, stride=2, ceil_mode=True),    # 最大池化
                Fire(64, 16, 64, 64),
                Fire(128, 16, 64, 64),
                nn.MaxPool2d(kernel_size=3, stride=2, ceil_mode=True),  # 最大池化
                Fire(128, 32, 128, 128),
                Fire(256, 32, 128, 128),
                nn.MaxPool2d(kernel_size=3, stride=2, ceil_mode=True),  # 最大池化
                Fire(256, 48, 192, 192),
                Fire(384, 48, 192, 192),
                Fire(384, 64, 256, 256),
                Fire(512, 64, 256, 256),
            )
        else:
            # FIXME: Is this needed? SqueezeNet should only be called from the
            # FIXME: squeezenet1_x() functions
            # FIXME: This checking is not done for the other models
            raise ValueError("Unsupported SqueezeNet version {version}:"
                             "1_0 or 1_1 expected".format(version=version))

        # Final convolution is initialized differently from the rest
        final_conv = nn.Conv2d(512, self.num_classes, kernel_size=1)  # 1x1卷积
        self.classifier = nn.Sequential(
            nn.Dropout(p=0.5),  # dropout层
            final_conv,
            nn.ReLU(inplace=True),
            nn.AdaptiveAvgPool2d((1, 1))  #自适应均值池化，输出大小为（1，1）
        )

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                if m is final_conv:
                    init.normal_(m.weight, mean=0.0, std=0.01)
                else:
                    init.kaiming_uniform_(m.weight)
                if m.bias is not None:
                    init.constant_(m.bias, 0)

    def forward(self, x):
        x = self.features(x)
        x = self.classifier(x)
        return torch.flatten(x, 1)

class Fire(nn.Module):

    def __init__(self, inplanes, squeeze_planes,
                 expand1x1_planes, expand3x3_planes):
        # inplanes=96, squeeze_planes=16, expand1x1_planes=64, expand3x3_planes=64
        super(Fire, self).__init__()
        self.inplanes = inplanes
        self.squeeze = nn.Conv2d(inplanes, squeeze_planes, kernel_size=1)  # 1x1卷积
        self.squeeze_activation = nn.ReLU(inplace=True)   # 激活函数
        self.expand1x1 = nn.Conv2d(squeeze_planes, expand1x1_planes,
                                   kernel_size=1)  # 1x1卷积
        self.expand1x1_activation = nn.ReLU(inplace=True)   # 激活函数
        self.expand3x3 = nn.Conv2d(squeeze_planes, expand3x3_planes,
                                   kernel_size=3, padding=1)  # 1x1卷积
        self.expand3x3_activation = nn.ReLU(inplace=True)   # 激活函数

    def forward(self, x):
        x = self.squeeze_activation(self.squeeze(x))  # 1x1卷积计算
        return torch.cat([
            self.expand1x1_activation(self.expand1x1(x)),  # 再1x1卷积计算
            self.expand3x3_activation(self.expand3x3(x))  # 并着3x3卷积计算
        ], 1)  # 按通道数合并