PyTorch CNN – 加载CIFAR10并标准化

使用torchvision，很容易加载CIFAR10数据集。

首先导入库：

import torch
import torchvision
import torchvision.transforms as transforms

torchvision数据集的输出为范围[0,1]的PILImage图像，需要把它们转换成张量，并作归一化(范围[-1,1])处理。

transform = transforms.Compose(
    [transforms.ToTensor(),
     transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])

trainset = torchvision.datasets.CIFAR10(root='./data', train=True,
                                        download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=4,
                                          shuffle=True, num_workers=0) # windows系统设置num_workers=0

testset = torchvision.datasets.CIFAR10(root='./data', train=False,
                                       download=True, transform=transform)
testloader = torch.utils.data.DataLoader(testset, batch_size=4,
                                         shuffle=False, num_workers=0) # windows系统设置num_workers=0

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

下载数据集要花一点时间，请耐心等待：

Downloading https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz to ./data\cifar-10-python.tar.gz
88.7%

接下来，让我们从数据集中显示一些训练图像：

import matplotlib.pyplot as plt
import numpy as np

# 显示图像函数


def imshow(img):
    img = img / 2 + 0.5     # unnormalize
    npimg = img.numpy()
    plt.imshow(np.transpose(npimg, (1, 2, 0)))
    plt.show()


# 随机获取一些训练图像
dataiter = iter(trainloader)
images, labels = dataiter.next()

# 显示图片
imshow(torchvision.utils.make_grid(images))
# 打印标签
print(' '.join('%5s' % classes[labels[j]] for j in range(4)))

输出

dog bird horse horse