You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Use MAD (Mean Absolute Deviation) pooling, it's more accurate than mean pooling with SSIM metric (I, honestly, would trust PSNR more than SSIM with mean pooling).
This is what I'm using for evaluation (using >2 scale levels doesn't make much of a difference):
I use SSIM with mean pooling (and PSNR with peak value = 1) because it is the way these methods are normally used, so the results can be compared to any other. But I'll take a look at SSIM with MAD pooling for my personal reference.
On the other hand I wonder if it could be useful as a loss function.
Use MAD (Mean Absolute Deviation) pooling, it's more accurate than mean pooling with SSIM metric (I, honestly, would trust PSNR more than SSIM with mean pooling).
This is what I'm using for evaluation (using >2 scale levels doesn't make much of a difference):
The text was updated successfully, but these errors were encountered: