GuoLanqing / ShadowFormer

ShadowFormer (AAAI2023), Pytorch implementation
MIT License
129 stars 17 forks source link

I am unable to reach your PSNR. #22

Closed Hedlen closed 1 year ago

Hedlen commented 1 year ago

Hi, I follow your provided code and keep your setting. However, I am unable to reach your PSNR. Could you please provide me with the training code and pre-train model that can reach this number? Thank you very much !

image
GuoLanqing commented 1 year ago

Thanks for your interest. Our released pre-trained model can reach the reported results in the paper. As we mentioned in the experiment part, following previous works, our evaluation includes two settings, i.e., evaluated in 256256 and the original resolution. If you want to reach the results of 256 256, you should resize the results first and then calculate the PSNR and SSIM.