There is an obvious difference between the results of reproduction and the paper

sanghyun-son / EDSR-PyTorch

PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)

MIT License

2.44k stars 671 forks source link

There is an obvious difference between the results of reproduction and the paper #361

Open HongSmit opened 3 months ago

HongSmit commented 3 months ago

Hello ! I am replicating the results of your paper, but I tested the data set according to the pt file you provided, and there is an obvious gap between the results and the paper. May I ask what went wrong?

python main.py --dir_data /mnt/workspace/ --model EDSR --data_test Set5 --chop --save_results --rgb_range 1 --data_range 801-900 --scale 4 --n_feats 256 --n_resblocks 32 --res_scale 0.1  --pre_train ../model_zoo/EDSR_x4.pt --test_only --res_scale 0.1