把cifar数据转换为图片

参考

https://gist.github.com/fzliu/64821d31816bce595a4bbd98588b37f5

"""
make_cifar10.py: Create training data from raw CIFAR-10 batches.
"""
 
import cPickle as pkl
import glob
import os
 
import numpy as np
from skimage.io import imsave
 
PIXELS_DIR = "pixel_data"
LABEL_FILE = "labels.txt"
 
def unpack_file(fname):
    """
        Unpacks a CIFAR-10 file.
    """
 
    with open(fname, "r") as f:
        result = pkl.load(f)
 
    return result
 
def save_as_image(img_flat, fname):
    """
        Saves a data blob as an image file.
    """
 
    # consecutive 1024 entries store color channels of 32x32 image
    img_R = img_flat[0:1024].reshape((32, 32))
    img_G = img_flat[1024:2048].reshape((32, 32))
    img_B = img_flat[2048:3072].reshape((32, 32))
    img = np.dstack((img_R, img_G, img_B))
 
    imsave(os.path.join(PIXELS_DIR, fname), img)
 
def main():
    """
        Entry point.
    """
 
    labels = {}
 
    # use "data_batch_*" for just the training set
    for fname in glob.glob("*_batch*"):
        data = unpack_file(fname)
 
        for i in range(10000):
            img_flat = data["data"][i]
            fname = data["filenames"][i]
            label = data["labels"][i]
 
            # save the image and store the label
            save_as_image(img_flat, fname)
            labels[fname] = label
 
    # write out labels file
    with open(LABEL_FILE, "w") as f:
        for (fname, label) in labels.iteritems():
            f.write("{0} {1}\n".format(fname, label))
 
if __name__ == "__main__":
main()