York-SDCNLab/Ghost-DeblurGAN

test result

Closed this issue · 2 comments

Does anyone reproduce the test results in the paper?PSNR:28.79 SSIM: 0.919
My test results on the GoPro dataset are PSNR: 27.53 SSIM: 0.917

Hi. Can you test the mobilenetv2 version with the same setup on the same computer? You will likely obtain results worse than those in the table again. I think this is a problem with the random seeds. The random seeds can affect both training and testing.

Thanks for your reply, I did get worse results in the mobilenetv2 experiment. Different computer environment and random seeds may lead to different results.