search

qhan1028 / Fully-Convolutional-Networks

Fully Convolutional Networks for Portrait Matting
MIT License
34 stars 12 forks source link

Didn't work in tensorflow1.6 with python 3.6 in win7 #1

Open fourk-ty opened 6 years ago

fourk-ty commented 6 years ago

ques I want it to work in my environment. What should I do. Which part should I changed? Or maybe there are some mistakes I made. I would be appreciated if anyone could help. Now it shows a ValueError: can not feed value of shape (0,) for Tensor 'input_image:0', which has shape '(?, 224, 224, 3)'

theadventureoflinko commented 5 years ago

Similar error occurs in my environment (MacOS 10.14, python3.6.8, tensorflow-gpu 1.1.0 ). In my case, error message was as following:

Traceback (most recent call last):
  File "fcn.py", line 317, in <module>
    main()
  File "fcn.py", line 246, in main
    images[i] = augment(im, flip_prob, aug_type, randoms)
ValueError: could not broadcast input array from shape (13,13,3) into shape (224,224,3)
Johncheng1 commented 5 years ago

@theadventureoflinko Have you solved this problem?

theadventureoflinko commented 5 years ago

@Johncheng1 No, still in trouble.

DarkFunct commented 5 years ago

执行 python3.5 fcn.py -m test 后 Res目录下生成的inp_0.png pred_0.png 为什么会一片黑呢？？ inp_0 pred_0