深度学习笔记

卷积和池化大小变化

(图像尺寸-卷积核尺寸 + 2*填充值)/步长+1
(图像尺寸-池化窗尺寸 + 2*填充值)/步长+1

卷积核计算公式

SeNet源码

        self.se = nn.Sequential(
            nn.AdaptiveAvgPool2d((1,1)),
            nn.Conv2d(filter3,filter3//16,kernel_size=1),
            nn.ReLU(),
            nn.Conv2d(filter3//16,filter3,kernel_size=1),
            nn.Sigmoid()
        )

SimAM

class Simam_module(torch.nn.Module):
    def __init__(self, channels=None, e_lambda=1e-4):
        super(Simam_module, self).__init__()

        self.activaton = nn.Sigmoid()
        self.e_lambda = e_lambda

    def __repr__(self):
        s = self.__class__.__name__ + '('
        s += ('lambda=%f)' % self.e_lambda)
        return s

    def forward(self, x):
        b, c, h, w = x.size()

        n = w * h - 1

        x_minus_mu_square = (x - x.mean(dim=[2, 3], keepdim=True)).pow(2)
        y = x_minus_mu_square / (4 * (x_minus_mu_square.sum(dim=[2, 3], keepdim=True) / n + self.e_lambda)) + 0.5

        return x * self.activaton(y)