Closed nicolas-dufour closed 2 years ago
Hi, our implementation is basically the same as our paper, except we use several epochs for warming up training.
And actually, we did the division for normalization purposes before. But I think you can safely remove the division in our current implementation.
Hi, I see that the loss used for IRT in the code is different than the one in the paper. Can u pls explain why is there a difference pls?
Also, i see u divide the pixelwise difference of O_main and the net_gt by the image. What's the idea behind this? Thank you!