图片三：numpy制作雪碧图（如何将多个图片拼接成一张图片）

 下面是我自己写的代码，每个图片直接紧挨这另个一个图片，图片直接没有任何填充，所以你的图片应该是同等规格的，大小和尺寸必须一致

#制作雪碧图
def make_grid(data, size=(4,4)):
    rows, columns = size[0], size[1]
    data = np.array(data)

    if (data<=1).all():
        data = (data*255).astype(np.uint8)
        
    data_row, data_columns = [], []
    for row in range(rows):
        data_columns.clear()
        for column in range(columns):
            data_columns.append(data[row*columns+column])
        data_row.append(np.hstack(copy.deepcopy(data_columns)))

    data = np.vstack(data_row)
    return data

 下面是是使用pytorch加载的mnist,fashion_mnist,cifar10数据，其中，mnist和fashion_mnist图片的规格为(28*28),  cifar10的图片规格为(32*32*3)

cifar_train = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transforms.ToTensor())
cifar_feature, cifar_label = cifar_train.data, cifar_train.targets

mnist_train = torchvision.datasets.MNIST(root='./data', train=True, download=True, transform=transforms.ToTensor())
mnist_feature, mnist_label = mnist_train.data, mnist_train.targets

fashion_mnist_train = torchvision.datasets.FashionMNIST(root='./data', train=True,download=True, transform=transforms.ToTensor())
fashion_mnist_feature, fashion_mnist_label = fashion_mnist_train.data, fashion_mnist_train.targets

制作的结果：