Python 万级图片数据集成到HDF5文件中，绝对路径

1、实验目的
13470 * 2 张图片集成到 HDF5文件中
标签 train_imag train_labels train_mean

2、读取路径图片

img1=[]
#dir='D:\\download\\dataset\\Bernard0\\Computer_Vision\\spliced_copymove_NIST\\rgb_imgs'
dir='D:\\download\\dataset\\Bernard0\\Computer_Vision\\spliced_NIST\\rgb_imgs'
#img1=np.zeros()
for filename in os.listdir(dir):
        img1.append(dir+'/'+filename)

3、定义HDF5文件结构

imagedata_shape=(len(img1), 256, 256, 3)
    maskdata_shape=(len(img1), 256, 256)
    mean = np.zeros(imagedata_shape[1:], np.float32)

f = h5py.File('spliced_NIST.hdf5', mode='w')
f.create_dataset("train_img", imagedata_shape, np.int8)
f.create_dataset("train_labels", maskdata_shape, np.int8)
f.create_dataset("train_mean", imagedata_shape[1:], np.float32)

4、图片数据读取

for i in range(len(img1))