pytorch实现LeNet-5网络

pytorch实现LeNet-5网络

• 定义网络
• 数据处理与加载
• 主程序

1、训练模型

	包括损失函数，优化器，可视化等操作

2、测试

定义网络：

input

import torch.nn as nn
import torch.nn.functional as F

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(3, 6, 5) 
        self.conv2 = nn.Conv2d(6, 16, 5)  
        self.fc1   = nn.Linear(16*5*5, 120)  
        self.fc2   = nn.Linear(120, 84)
        self.fc3   = nn.Linear(84, 10)

    def forward(self, x): 
        x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2)) 
        x = F.max_pool2d(F.relu(self.conv2(x)), 2) 
        x = x.view(x.size()[0], -1)   #展开成一维的
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)        
        return x

net = Net()
print(net)

output

Net(
  (conv1): Conv2d(1, 6, kernel_size=(5, 5), stride=(1, 1))
  (conv2): Conv2d(6, 16, kernel_size=(5, 5), stride=(1, 1))
  (fc1): Linear(in_features=400, out_features=120, bias=True)
  (fc2): Linear(in_features=120, out_features=84, bias=True)
  (fc3): Linear(in_features=84, out_features=10, bias=True)
)

数据处理与加载：

input

import torchvision as tv
import torchvision.transforms as transforms
from torchvision.transforms import ToPILImage
show = ToPILImage() # 可以把Tensor转成Image，方便可视化

定义对数据的预处理

#torchvision.transforms是pytorch中的图像预处理包
#一般用Compose把多个步骤整合到一起
transform = transforms.Compose([
        transforms.ToTensor(), # 转为Tensor
        transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)), # 归一化
                             ])

训练集

trainset = tv.datasets.CIFAR10(
                    root='./data/', 
                    train=True, 
                    download=True,
                    transform=transform)

trainloader = t.utils.data.DataLoader( #训练数据加载，将每4个拼成一个批次
                    trainset, 
                    batch_size=4,
                    shuffle=True,  # shuffle打乱数据顺序
                    num_workers=2)  # 加载数据时使用多少子进程。默认值为0，表示在主进程中加载数据。 

测试集

testset = tv.datasets.CIFAR10(
                    './data/',
                    train=False, 
                    download=True, 
                    transform=transform)

testloader = t.utils.data.DataLoader(
                    testset,
                    batch_size=4, 
                    shuffle=False,
                    num_workers=2)

classes = (‘plane’, ‘car’, ‘bird’, ‘cat’,
‘deer’, ‘dog’, ‘frog’, ‘horse’, ‘ship’, ‘truck’)

output

Files already downloaded and verified
Files already downloaded and verified

主程序:

from torch import optim
criterion = nn.CrossEntropyLoss() # 交叉熵损失函数
optimizer = optim.SGD(net.parameters(), lr=0.001, momentum=0.9) # 优化器

1、训练模型

• 输入数据
• 前向传播+反向传播
• 更新参数

t.set_num_threads(8) # 设置cpu利用率
for epoch in range(2):  
    
    running_loss = 0.0 #初始化loss
    for i, data in enumerate(trainloader, 0): #enumerate枚举训练器中数据，参考[enumerate](https://www.runoob.com/python/python-func-enumerate.html)
        
        # 输入数据
        inputs, labels = data
        inputs, labels = Variable(inputs), Variable(labels)
        
        # 梯度清零
        optimizer.zero_grad()
        
        # forward + backward 
        outputs = net(inputs)
        loss = criterion(outputs, labels)
        loss.backward()   
        
        # 更新参数 
        optimizer.step()
        
        # 打印log信息
        # loss 是一个scalar,需要使用loss.item()来获取数值，不能使用loss[0]
        running_loss += loss.item()
        if i % 2000 == 1999: # 每2000个batch打印一下训练状态
            print('[%d, %5d] loss: %.3f', % (epoch+1, i+1, running_loss / 2000))
            running_loss = 0.0
print('Finished Training')

2、测试

#**测试图像的实际labels**
dataiter = iter(testloader) #把测试数据放在迭代器iter
images, labels = dataiter.next() # 一个batch返回4张图片，依次获取下一个数据
print('实际的label: ', ' '.join( '%08s'%classes[labels[j]] for j in range(4)))

# make_grid将若干图像拼接成一幅，（images/2 - 0.5）是为了还原被归一化的数据
show(tv.utils.make_grid(images / 2 - 0.5)).resize((400,100))

# **网络测试结果的labels**
# 计算图片在每个类别上的分数
outputs = net(Variable(images))
# 得分最高的那个类
_, predicted = t.max(outputs.data, 1) # max(output.data, 1)是将4*10维的数据，取每一行上的最大值，返回最大值和序号

print('预测结果: ', ' '.join('%5s'% classes[predicted[j]] for j in range(4)))

#整个测试集上预测
correct = 0 # 预测正确的图片数
total = 0 # 总共的图片数

# 由于测试的时候不需要求导，可以暂时关闭autograd，提高速度，节约内存
with t.no_grad():
    for data in testloader:
        images, labels = data
        
        outputs = net(images)
        _, predicted = t.max(outputs, 1)
        
        total += labels.size(0)
        correct += (predicted == labels).sum()

print('10000张测试集中的准确率为: %d %%' % (100 * correct / total))

每一类分类情况

class_correct = list(0. for i in range(10))
class_total = list(0. for i in range(10))
with t.no_grad():
    for data in testloader:
        images,labels = data
        outputs = net(images)
        _,predicted = t.max(outputs,1)
        c = (predicted==labels).squeeze()
        for i in range(4):
            label = labels[i]
            class_correct[label] += c[i].item()
            class_total[label] += 1
for i in range(10):
    print('Accuracy of %5s : %2d %%' % (classes[i],100 * class_correct[i] / class_total[i]))