NotImplementedError: There were no tensor arguments to this function about estrnn

[zzh-tech]

Is this your command?

python inference.py

[Entretoize]

Yes but I modified it not to have to write parameters each time:

if __name__ == '__main__':
    parser = ArgumentParser()
    parser.add_argument('--src', type=str, default="jardin", help="the path of input video or video dir")
    parser.add_argument('--ckpt', type=str, default="checkpoints/ESTRNN_C80B15_BSD_2ms16ms.tar", help="the path of checkpoint of pretrained model")
    parser.add_argument('--dst', type=str, default="result", help="where to store the results")
    args = parser.parse_args()

[zzh-tech]

What is the image file format in this dir "jardin"?
If you don't mind, you can share the contents of this folder with me.

[Entretoize]

After your message I tried with the BSD data and it doesn't worked, then as I already successfully tried your script I redownloaded all py files and it works, then sorry I think I had modified something somewhere that corrupted the code...

Maybe I should open a new issue, tell me, but I have a small issue, the script works great with blurred images but when there's no blur at all the original image is sharpen than the result, is there a way to have a threshold to prevent adding blur on sharp images or on sharp portion of an image ?
(sample images : https://linkall.pro/ESTRNN/000001.png and 000002.png...)

[zzh-tech]

Maybe you can design some solutions with blur detection.

[Entretoize]

Yes, I'll compare original and modifed and keep the best after computation.
Thanks