pytorch-deepglr icon indicating copy to clipboard operation
pytorch-deepglr copied to clipboard

Published 20 hours ago • verified publisher icon huyvd7

about the robust

Open robotzheng opened this issue 5 years ago • 3 comments

10_n.zip when I use the above images to test, I only got the result like bellow: python3 validate_DGLR.py dataset/test/ -m model/deepglr.pretrained -w 324 -o ./ CUDA: False, device: cpu torch.Size([3, 324, 324]) torch.Size([3, 324, 324]) torch.Size([9, 3, 36, 36]), 9/9 Prediction time: 51.84631419181824 PSNR: 23.136705906862762 SSIM: 0.8942293219652124 Saved ./0.png python3 validate_GLR.py dataset/test/ -m model/glr.pretrained -w 324 -o ./ CUDA: False, device: cpu torch.Size([3, 324, 324]) torch.Size([3, 324, 324]) torch.Size([9, 3, 36, 36]), 9/9 Prediction time: 12.586071968078613 PSNR: 22.999027375184106 SSIM: 0.8679046020071071 Saved ./0.png

It is not robust for my images, it should be 30(PSNR).

robotzheng avatar Apr 07 '20 07:04 robotzheng

Hi @robotzheng Do you have a ground truth of your images? Your uploaded zip only contains the noisy one. You need a ground truth for computing those metrics. Best.

huyvd7 avatar Apr 07 '20 23:04 huyvd7

10_r.zip Thanks.

robotzheng avatar Apr 08 '20 01:04 robotzheng

Hi @huyvd7, have you test this picture, maybe robust is not so easy, any clue for future?

robotzheng avatar Apr 17 '20 06:04 robotzheng