search

alterzero / DBPN-Pytorch

The project is an official implement of our CVPR2018 paper "Deep Back-Projection Networks for Super-Resolution" (Winner of NTIRE2018 and PIRM2018)
https://alterzero.github.io/projects/DBPN.html
MIT License
565 stars 164 forks source link

Evaluate test data #31

Closed RaminAbbaszadi closed 5 years ago

RaminAbbaszadi commented 5 years ago

Thanks for your source code. When I try eval code with 2x, 4x, and 8x rescale parameters. I found that 2x, 4x results have some problems in color. I attached SR result. I am appreciate you if you reply me.

For 2x: woman_x2

For 4x: woman_x4

For 8x:

woman_x8

alterzero commented 5 years ago

Carefully check the parameter for --model. There are two model: DBPN and DBPNLL. Also set --residual to False.

I have just updated the code, removing if clause on the weights loading. So, if the --model and --weight has different structure, it will cause error.

RaminAbbaszadi commented 5 years ago

Thank you. problem solved. Best Regards

Lelouch422 commented 5 years ago

Hi@alterzero, could you please tell the difference between the two models DBPN and DBPNLL, thanks a lot!

alterzero commented 5 years ago

Hi @Lelouch422 DBPNLL is the deeper model. While DBPN uses T=7 (T is number of bp stage), DBPNLL uses T=10