python - 尽管导入操作系统，_get_batches_of_transformed_samples 中的 NameError : global name 'os' is not defined in keras_preprocessing/image/iterator. py

Question

这是我的代码顶部的导入

import os
from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img

这是回溯:

Using TensorFlow backend.
Found 33 images belonging to 2 classes.
Traceback (most recent call last):
  File "imggen.py", line 29, in <module>
    save_format = 'jpeg'):
  File "/home/psdanielxu/.local/lib/python2.7/site-packages/keras_preprocessing/image/iterator.py", line 111, in next
    return self._get_batches_of_transformed_samples(index_array)
  File "/home/psdanielxu/.local/lib/python2.7/site-packages/keras_preprocessing/image/iterator.py", line 245, in _get_batches_of_transformed_samples
    img.save(os.path.join(self.save_to_dir, fname))
NameError: global name 'os' is not defined

我不确定这个错误是如何持续存在的，因为我已经导入了操作系统

Answer 1

我查看了 keras-preprocessing 的 github，似乎他们错过了那里的导入。尝试在该文件位于 /home/psdanielxu/.local/lib/python2.7/site-packages 的 iterator.py 文件上添加 import os/keras_preprocessing/image/iterator.py 在您的系统上

如果这对您有用，您还应该转到项目的 github 并针对此错误提出问题。 (https://github.com/keras-team/keras-preprocessing/issues)