jee-king / tsan Goto Github PK

TSAN/Train/model/tsan.py
line: 103 out = x --> out = x.clone()

I have tried to train the TSAN network on other dataset. For that I have made folder named "dataset". In dataset folder, two sub-folders named "DIV2K_train_HR" and "DIV2K_train_LR_bicubic" are created under DIV2K folder and HR and LR images are stored in the respective folders. In option.py code, I have changed the path:
parser.add_argument('--dir_data', type=str, default='dataset/DIV2K/', help='dataset directory')

But, while compiling the train code using this following command:
python main.py --template TSAN --save TSAN_X2 --scale 2 --reset --save_results --patch_size 96 --ext sep_reset
I got this error:

Please help me on this regard @Jee-King

I got similar results on U100 and B100, however, degraded on Set14. maybe Set14 only has 14 images, right?

Hi!
In the paper you say about using RAB as the refine net which described in equation 17.
But in current (deleted) implementation you use 5 stacked conv+relu blocks instead of RAB.
Is there any particular reason for that? Does stacked conv-relu block perform better?