search

sanghyun-son / EDSR-PyTorch

PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)
MIT License
2.42k stars 668 forks source link

Question regarding dir_data options in demo.sh, and image database layout #263

Open drcdr opened 4 years ago

drcdr commented 4 years ago

I noticed that there are several different --dir_data arguments in the demo.sh file:

../../datasetx4
../../datasetx3
../../datasetx2

What do these refer to? I have a single dataset set up as shown, and things seem to be running OK. But, the EDSR_paper template seems to be taking a long time; understandably, but I was wondering if I should be using a different database in some way.

<dirname>
|-benchmark
    |- B100
        |- HR
        |- LR_bicubic: under this, X2, X3, X4
    |- Set5
        |- HR
        |- LR_bicubic: under this, X2, X3, X4
    |- Set14
        |- HR
        |- LR_bicubic: under this, X2, X3, X4
    |- Urban100
        |- HR
        |- LR_bicubic: under this, X2, X3, X4
|-DIV2K
    |- DIV2K_train_HR
    |- DIV2K_train_LR_bicubic: under this, X2, X3, X4
    |- DIV2K_train_LR_unknown: under this, X2, X3, X4
    |- DIV2K_test_LR_bicubic: under this, X2, X3, X4
    |- DIV2K_test_LR_unknown: under this, X2, X3, X4
xiaoyi-st commented 3 years ago

Hello, do you have this error?ModuleNotFoundError: No module named 'data.div2k_train_hr' I also use DIK2 data