保存npz文件出现乱码问题，求解！

Question

运行以下代码，发现存入npz文件内的图片是乱码，输出图在下方，已检查“数据集路径”，“图片尺寸”，“图片格式“均无问题。安装的库均为匹配版本

以下是保存数据集代码

from os import listdir
import numpy as np
from numpy import asarray
from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.preprocessing.image import load_img
def load_images(path, size=(256, 256)):
    data_list = list()
    for filename in listdir(path):
        pixels = load_img(path + filename, target_size=size)
        pixels = img_to_array(pixels)
        data_list.append(pixels)
    return asarray(data_list)

# dataset path
path = '../hourse/horse2zebra/'
# load dataset A
dataA1 = load_images(path + 'trainA/')
dataA2 = load_images(path + 'testA/')
dataA = np.vstack((dataA1, dataA2))
print('Loaded dataA: ', dataA.shape)
# load dataset B
dataB1 = load_images(path + 'trainB/')
dataB2 = load_images(path + 'testB/')
dataB = np.vstack((dataB1, dataB2))
print('Loaded dataB: ', dataB.shape)
# save as compressed numpy array
filename = 'horse2zebra256.npz'
# np.savez_compressed(filename, dataA, dataB)
print('Saved dataset: ', filename)

# 查看horse2zebra_256.npz里的图片

from numpy import load
from matplotlib import pyplot
data = load(r'../hourse/horse2zebra256.npz')
dataA, dataB = data['arr_0'], data['arr_1']
print('Loaded: ', dataA.shape, dataB.shape)
n_samples = 3
for i in range(n_samples):
    pyplot.subplot(2, n_samples, 1 + i)
    pyplot.axis('off')
    pyplot.imshow(dataA[i])
for i in range(n_samples):
    pyplot.subplot(2, n_samples, 1 + n_samples + i)
    pyplot.axis('off')
    pyplot.imshow(dataB[i])
pyplot.show()

以下是创建npz文件代码

import numpy as np

# 创建需要存储的数据
dataA = np.random.rand(100, 256, 256, 3)
dataB = np.random.rand(100, 256, 256, 3)

# 存储数据到npz文件
np.savez('horse2zebra256.npz', arr_0=dataA, arr_1=dataB)

Answer 1

不知道你这个问题是否已经解决, 如果还没有解决的话:

• 以下回答来自chatgpt:

  对于保存图像数据，可以采用opencv库中的imread和imwrite函数，将图像读取为numpy array形式，再进行保存。可以先尝试使用以下代码进行保存：

  import numpy as np
  import cv2
  
  images = []
  for image_path in image_path_list: # image_path_list为图像的路径列表
      img = cv2.imread(image_path, cv2.IMREAD_UNCHANGED)
      images.append(img)
  
  np.savez('images.npz', *images)
  

  如果还是出现乱码，可以尝试使用pickle进行保存：

  import numpy as np
  import cv2
  import pickle
  
  images = []
  for image_path in image_path_list: # image_path_list为图像的路径列表
      img = cv2.imread(image_path, cv2.IMREAD_UNCHANGED)
      images.append(img)
  
  with open('images.pkl', 'wb') as f:
      pickle.dump(images, f)
  

  如果还是出现乱码，可能是其它原因导致，需要进一步排查。

如果你已经解决了该问题, 非常希望你能够分享一下解决方案, 写成博客, 将相关链接放在评论区, 以帮助更多的人 ^-^