（8）catanddog_自定义卷积神经网络分类和测试

1、采用的自定义卷积神经网络

设计参考了

深度学习（二）之猫狗分类 - 段小辉 - 博客园

 1.1神经网络结构

输入尺寸为112*112，用了4层卷积网络和2层输出，每层卷积陈的结构如下

1.2代码实现 

(1)注意输出层用损失函数CrossEntropyLoss()时不需要

criteon = nn.CrossEntropyLoss()

......................
.....................
loss = criteon(logits, y)

(2)完整代码

# import torch.nn.functional as F
import  torch
from    torch import optim, nn
class OurConv(nn.Module):
    def __init__(self):
        super(OurConv,self).__init__()
        self.conv1 = nn.Sequential(
            nn.Conv2d(3,32,kernel_size=3,padding=1),
            nn.ReLU(),
            nn.BatchNorm2d(32),
            nn.MaxPool2d(2,2),
            nn.Dropout(0.25)
        )
        self.conv2 = nn.Sequential(
            nn.Conv2d(32,64,kernel_size=3,padding=1),
            nn.ReLU(),
            nn.BatchNorm2d(64),
            nn.MaxPool2d(2,2),
            nn.Dropout(0.25)
        )

        self.conv3 = nn.Sequential(
            nn.Conv2d(64,128,kernel_size=3,padding=1),
            nn.ReLU(),
            nn.BatchNorm2d(128),
            nn.MaxPool2d(2,2),
            nn.Dropout(0.25)
        )

        self.conv4 = nn.Sequential(
            nn.Conv2d(128,256,kernel_size=3,padding=1),
            nn.ReLU(),
            nn.BatchNorm2d(256),
            nn.MaxPool2d(2,2),
            nn.Dropout(0.25)
        )


        self.fc = nn.Sequential(
            nn.Linear(256*7*7,256),
            nn.Dropout(0.2),
            nn.Linear(256,2),
        )
    def forward(self,x):
        x = self.conv1(x)
        print("shape1",x.shape)
        x = self.conv2(x)
        print("shape2",x.shape)
        x = self.conv3(x)
        print("shape3",x.shape)
        x = self.conv4(x)
        print("shape4",x.shape)

        x = x.view(x.size(0),-1)
        x = self.fc(x)
        # return F.log_softmax(x,dim=1)
        return x

def main():
    tmp = torch.randn(64, 3, 112, 112)
    out_mode = OurConv()
    out = out_mode(tmp)
    print('block:', out.shape)


if __name__ == '__main__':
    main()

2训练

import  torch
from    torch import optim, nn
# import  visdom
# fr