Closed xuyuan-qd closed 1 day ago
Sure. I would upload the results and training file asap.
Hello,
It's been over a month. Will you upload it ?
Thanks.
I have the same problem.
Hello,
It's been over a month. Will you upload it ?
Thanks.
I updated the paper, which should be visible the next week. New model weights and codes have been uploaded.
I have the same problem.
I updated the paper, which should be visible the next week. New model weights and codes have been uploaded.
Thank you @HanzhouLiu ! very appreciated.
Edit:
......................DeblurDiNAT/models/losses.py", line 5, in <module> from models.torch_wavelets import DWT_2D ModuleNotFoundError: No module named 'models.torch_wavelets'
Any input regarding this error? Which package contains DWT_2D?
Edit 2: can you fix https://drive.google.com/drive/folders/1BdV2l7A5MRXLWszGonMxR88eV27geb_n?usp=sharing there is no GoPro dataset there (404 error)
Also GoPro from official website does not have trainA and trainB folders as stated in your YAML files - can you explain how did you generate them?
Thanks
Thank you @HanzhouLiu ! very appreciated.
Edit:
......................DeblurDiNAT/models/losses.py", line 5, in <module> from models.torch_wavelets import DWT_2D ModuleNotFoundError: No module named 'models.torch_wavelets'
Any input regarding this error? Which package contains DWT_2D?
Edit 2: can you fix https://drive.google.com/drive/folders/1BdV2l7A5MRXLWszGonMxR88eV27geb_n?usp=sharing there is no GoPro dataset there (404 error)
Also GoPro from official website does not have trainA and trainB folders as stated in your YAML files - can you explain how did you generate them?
Thanks
Thanks for letting me know. I already removed those lines of codes in the latest version, or you can just comment them out. trainA is just the blur. I am still uploading the gopro dataset, which should be available at 13:00 06/28 CST.
Thanks for the update.
One more issue:
Your dataset code contains sub-folders inside trainA/trainB folders, however your data loader does not account for these sub-folders and expects all the files to be under trianA/trainB without being separated to sub-folders (GOPR0372_07_00/GOPR0372_07_01/etc...)
This way the loader code or the dataset structure needs to be fixed in order for training code to run properly.
Which way do you suggest?
Thanks
Thanks for the update.
One more issue:
Your dataset code contains sub-folders inside trainA/trainB folders, however your data loader does not account for these sub-folders and expects all the files to be under trianA/trainB without being separated to sub-folders (GOPR0372_07_00/GOPR0372_07_01/etc...)
This way the loader code or the dataset structure needs to be fixed in order for training code to run properly.
Which way do you suggest?
Thanks
The dataset codes does not affect training codes. We only have two subsets of images, blurred and sharp. If your dataset does not contain sub-folders, you just need to remove one of the for loops.
could you please upload this file?