Trying to reproduce Uformer_B model for deblurring

[jhkim0759]

I trained the model you provided on the GoPro dataset. I trained 3000 Epoch with CharbonnierLoss and Edge loss as MPRNet code and I also use same scheduler as MPRNet code

but I got the PSNR 31.1db on GoPro test dataset

Is this happens because of the absence of modulator? or is there any other problem?

[Dong1P]

I also had similar results in the GoPro test dataset.

Can you upload the train code for deblurring case?

Best,

[Einstone-rose]

I trained the model you provided on the GoPro dataset. I trained 3000 Epoch with CharbonnierLoss and Edge loss as MPRNet code and I also use same scheduler as MPRNet code

but I got the PSNR 31.1db on GoPro test dataset

Is this happens because of the absence of modulator? or is there any other problem?

Did you use the PSNR metric in Matlab or Python?

[jhkim0759]

I used the metric in Python

[ZhendongWang6]

Hi, thanks for your interest! We have uploaded codes and models for deblurring.

[gauenk]

I now get 33.0627 PSNR vs the reported 32.97 PSNR on the website. The SSIM both match at 0.967.

Why does the github repo do better than the reported value? I am using the test GoPro dataset to compute this number.