CIFAR-10数据集是一个包含了60000张32×32的RGB图像数据集，所有图像共被划分为10个类别，分别为：

['飞机', '汽车', '鸟', '猫', '鹿', '狗', '青蛙', '马', '船', '卡车']
['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck']

 

程序编写

1 文件夹建立

步骤一：首先我们在桌面（也可为其他路径）路径下建立一个名为cifar10的空文件夹；

步骤二：在cifar10文件夹下建立train和test文件夹；

步骤三：分别在train和test文件夹下建立10个与图像类别名称相同的空文件夹。

2 程序编写

对训练集数据进行图像保存的程序如下所示：

from keras.datasets import cifar10
from PIL import Image

"""
数据集下载与加载（利用KerasAPI）
"""
# x_train_original和y_train_original代表训练集的图像与标签, x_test_original与y_test_original代表测试集的图像与标签
(x_train_original, y_train_original), (x_test_original, y_test_original) = cifar10.load_data()

root_path = 'C:\\Users\\Lenovo\\Desktop\\cifar10\\'


def mnist_train_save():
    for i in range(60000):
        print(i)
        if (y_train_original[i] == 0):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\airplane\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 1):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\automobile\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 2):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\bird\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 3):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\cat\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 4):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\deer\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 5):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\dog\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 6):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\frog\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 7):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\horse\\' + str(i) + '.jpg')
        elif (y_train_original[i] == 8):
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\ship\\' + str(i) + '.jpg')
        else:
            img = Image.fromarray(x_train_original[i])
            img.save(root_path + 'train\\truck\\' + str(i) + '.jpg')


mnist_train_save()
print('图片保存完成')

对测试集数据进行图像保存的程序只需要把程序中的train改成test即可。