预测输出模块

Savannah913

已于 2023-05-11 14:42:38 修改

阅读量123

点赞数

文章标签：深度学习 cnn 神经网络

于 2023-05-11 14:42:23 首次发布

本文链接：https://blog.csdn.net/weixin_45847320/article/details/130621744

版权

一个卷积+全连接+sigmoid+全连接

# 定义模型
class Output_Layer(nn.Module):
    def __init__(self, in_channels, out_channels):
        super(Output_Layer, self).__init__()
        '''
            Input torch.Size([32, 64, 307, 12])
            Output: torch.Size([32, 307, 12, 1])
        '''
        self.conv = nn.Conv2d(in_channels, out_channels, kernel_size=(1,3), padding=(0, 1))
        self.fc1 = nn.Linear(out_channels, out_channels)
        self.sigmoid = nn.Sigmoid()
        self.fc2 = nn.Linear(out_channels, out_channels)

    def forward(self, x):
        x = self.conv(x)
        #x = x.view(x.size(0), -1)  # 将卷积层的输出展平
        x = x.permute(0,2,3,1)
        x = self.fc1(x)
        x = self.sigmoid(x)
        x = self.fc2(x)
        return x

两个1*1卷积构成的输出模块

class OutputModule(nn.Module):
    def __init__(self, in_channels, out_channels):
        super(OutputModule, self).__init__()
        '''
           Input torch.Size([32, 64, 307, 12])
           Output: torch.Size([32, 307, 12, 1])
        '''
        self.conv1 = nn.Conv2d(in_channels, out_channels, kernel_size=1)
        self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=1)
        self.relu = nn.ReLU()

    def forward(self, x):
        x = self.conv1(x)
        x = self.relu(x)
        x = self.conv2(x)
        x = self.relu(x)
        return x.permute(0,2,3,1)