图片集生成h5文件 体积反而变大

[DarkPocket]

512张 64x64的图片 原始大小0.4mb，转换h5文件后 有25m 多了60倍 使用的示例代码

    data_image_set =  np.random.rand(len(data_image_path), 64, 64, 3).astype('float32')
    data_label_set = np.random.rand(len(data_label), 1).astype('float32')
    list_classes_set= np.random.rand(10, 1).astype('int')

    for i in range(len(data_image_path) ):
        data_image_set[i] = np.array(plt.imread(os.path.join(image_dir_path,data_image_path[i])))
        data_label_set[i] = np.array(data_label[i])

    for i in range( 10 ):
        list_classes_set[i] = np.array(i)

    f = h5py.File(dataset_name, 'w')
    f.create_dataset('data_set_x', data=data_image_set)
    f.create_dataset('data_set_y', data=data_label_set)
    f.create_dataset('list_classes', data=list_classes_set)
    
    f.close()

读取出来好像没发现问题 data_set_x (512, 64, 64, 3) data_set_y (512, 1) list_classes (10, 1)

[dq0309]

会不会是类型问题？ 图像像素应该是0-255，试试类型为‘uint8’