dingkeyan93 / iqa-optimization Goto Github PK

Can I get MATLAB version code of "recover.py"?

Dear author:
I add DISTS loss to my task but the model training speed is very slow, Is it normal?
code:
def forward():
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
D = DISTS().to(device)
dists_loss = D(X, Y, as_loss=True)

The output of MAD gives the distance between two images between [0, infinity). How do you convert this value in the similarity of range [0,1], so that it can be compared with other metrics (ssim, ms-ssim)?

I got the error below while running the example given in the definition CWSSIM class. The issue raised in line 52 of file SteerPyrComplex.

x = torch.index_select( x, -2, self.indF[0] )

IndexError: index out of range in self

Whether to use recover.py to restore the reference image from the initial image and calculate the distance, if the value was show 0, indicating that the IQA method is injective？

wrong: module 'torch' has no attribute 'rfft'
pytorch ：1.8.1+cu101