CBAM[2018-ECCV]

论文细节可以参考link：https://zhuanlan.zhihu.com/p/65529934

• CBA-Module

                   

• Channel Attention Module

                   

class CAModule(nn.Module):
    '''Channel Attention Module'''
    def __init__(self, channels, reduction):
        super(CAModule, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.max_pool = nn.AdaptiveMaxPool2d(1)
        self.shared_mlp = nn.Sequential(nn.Conv2d(channels, channels // reduction, kernel_size=1, padding=0, bias=False),
                                        nn.ReLU(inplace=True),
                                        nn.Conv2d(channels // reduction, channels, kernel_size=1, padding=0, bias=False))
        self.sigmoid = nn.Sigmoid()

    def forward(self, x):
        input = x
        avg_pool = self.avg_pool(x)
        max_pool = self.max_pool(x)
        x = self.shared_mlp(avg_pool) + self.shared_mlp(max_pool)
        x = self.sigmoid(x)

        return input * x

• Spatial Attention Module

                                  

class SAModule(nn.Module):
    '''Spatial Attention Module'''
    def __init__(self):
        super(SAModule, self).__init__()
        self.conv = nn.Conv2d(2, 1, kernel_size=3, padding=1, bias=False)  # default kernel_size = 7 x 7
        self.sigmoid = nn.Sigmoid()

    def forward(self, x):
        input = x
        avg_c = torch.mean(x, 1, True)
        max_c, _ = torch.max(x, 1, True)
        x = torch.cat((avg_c, max_c), 1)
        x = self.conv(x)
        x = self.sigmoid(x)
        return input * x

• ResBlock + CBAM

                          

class BasicBlock(nn.Module):
    
    expansion = 1

    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.ca    = CAModule(planes)
        self.sa    = SAModule()

        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.ca(out) * out
        out = self.sa(out) * out

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

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

        return out

• Conclusion

1. 将CBA-Module集成到不同的backbone中，模型的性能(在分类、检测)都有一致地提升；
2. 从对比实验的结果来看，先进行Channel attention，后进行Spatial attention的效果会好一些；
3. 对于Spatial attention而言，采用7x7的kernel，结果似乎更好一些；