Add se channel attention module to the network

 First step ： First, in the network structure py Add SELayer This class 
class SELayer(nn.Module):
    def __init__(self, channel, reduction=16):
        super(SELayer, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.fc = nn.Sequential(
            nn.Linear(channel, channel // reduction, bias=False),
            nn.ReLU(inplace=True),
            nn.Linear(channel // reduction, channel, bias=False),
            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.expand_as(x)

The second step ：

  On the Internet body Class SE Module properties .

Be careful ： This 75 Is the number to be modified , For example, I use it here yolov3, It uses voc Data sets , So this should be 3*（4+1+20）=75. If it is two categories, it should be changed to 3*（4+1+2）=21.

The third step ：

At the end of the network output Place to use SE modular .

effect :

 

 voc After the data set is used once map Promoted 0.86.

 

  Own a very simple data set map Can also be improved .