search

DeokyunKim / Progressive-Face-Super-Resolution

Official Pytorch Implementation of Progressive Face Super-Resolution (BMVC 2019 Accepted)
260 stars 60 forks source link

Distorted image outputs #19

Closed kushalchordiya216 closed 4 years ago

kushalchordiya216 commented 4 years ago

I have tried running the demo.py script to use the pretrained ckpt file included in your repo, to upscale images from the celeba dataset, but the output images are always highly distorted. Is there some kind of preprocessing that needs to be done on the images before running demo.py input 000010

output face_op

All images i have tried so far have had similar results

0xinit commented 4 years ago

did you find any solution to it??

DeokyunKim commented 4 years ago

Hi, thank you guys for your interest in our work.

Before resizing image to 128x128 pixels, We cropped images to 178x178 pixels to remove the background. (we use torchvision.transforms.CenterCrop((178, 178)) )

we note that we can not show you guys the exact hyper-parameters because of company research project.

Thanks in advance your understanding.

0xinit commented 4 years ago

@DeokyunKim Yes, that is what I was thinking. While I understand that it may not be possible, but do you think we can get the training code for this paper? I am trying to implement this for my research project. It would be of great help.

If not, then that is fine. :)

kushalchordiya216 commented 4 years ago

Hi, thank you guys for your interest in our work.

Before resizing image to 128x128 pixels, We cropped images to 178x178 pixels to remove the background. (we use torchvision.transforms.CenterCrop((178, 178)) )

we note that we can not show you guys the exact hyper-parameters because of company research project.

Thanks in advance your understanding.

I understand. Thank you for responding. I think I'll close this issue now.