ImportError: No module named 'PIL''.

[Catheriana]

I use python3.5.4. Anaconda3-5.0.1-Windows-x86_64. Tensorflow1.4.0.
When I run 'tid2013.py' ,there is the error 'ImportError: No module named 'PIL''.
How can this happen? Should I change the version of tensorflow to 1.0.0 ?
Besides,I have installed 'pillow',it is said 'PIL'and 'pillow' are incompatible since 'PIL' only can support python 2.7.However, your program is run with python3.5.
All in all，how to configure ‘PIL’ and python3.5 ? Should I uninstall 'pillow' first?
THANKS VERY MUCH!

[HC-2016]

In tid2013_input.py :

import PIL.Image as Image
def load_img(path, gray_scale=False):
""" Load image and convert to Image object.

Args:
    :param path:str. the path of the image file.
    :param gray_scale:bool. gray or color.

Return:
    :return img:Image object. an instance of Image object.
"""
img = Image.open(path)
if gray_scale:
    img = img.convert('L')
else:
    img = img.convert('RGB')

return img

PIL is only used to load images.

Solution:

1. https://stackoverflow.com/questions/8863917/importerror-no-module-named-pil
2. You can use any other image processing library instead.

[Catheriana]

THANK YOU !
It works! I re-installed tensorflow using cmd instead of pip.
I am a new learner and interested in no-reference method. During this time ,I have tried to change the tid2013_input code and remove the feature fusion step of full-reference model , however，there are always errors in it. I wonder if it is convenient for you to release the no-reference method,Thank you very much!