Something wrong when I test my own data
huizhiw opened this issue · 0 comments
Thank you for your excellent work.
I'm doing some research about preprocessing images before feeding them into the network. So I', excited to try your approach.
I found some problems and may need help and advice from you to make it work
-
There are two lines of codes that needs be to adapted in util/visulizer.py if others use the lastest version of scipy.
the solution:
from PIL import Image
import numpy as npline 46: im = imresize(im, (h, int(w * aspect_ratio)), interp='bicubic') -> im = np.array(Image.fromarray(im).resize((h, int(w*aspect_ratio)))) line 48: im = imresize(im, (int(h / aspect_ratio), w), interp='bicubic') -> im = np.array(Image.fromarray(im).resize((int(h/aspect_ratio), w)))
-
I'm really confused about the test of this projects. The test data you provided work well and can remove the reflection from the image. However, when I try to use my own images, it returns all black images.
So I wonder if I haven't done something before I test my own images? I use your pre-trained model and I think I strictly follow the instructions you give to test the code.