search

zhilin007 / FFA-Net

FFA-Net: Feature Fusion Attention Network for Single Image Dehazing
601 stars 141 forks source link

Is there any reason why you normalize only input, but not gt? #20

Open DongHwanJang opened 3 years ago

DongHwanJang commented 3 years ago

I believe that if you use the global residual connection, you need to normalize both of them. Unless, the residual term has to learn much more complicated space.

dingyan1478 commented 3 years ago

Hi! I have the same question. Do you understand now?

Alienvoid commented 3 years ago

Maybe the convolutional network automatically learns the reversal normalization such that it produces the clean image without normalizing the gt.

Wangbk-dl commented 2 years ago

Hi, I have the same question. Do you get any idea?

DongHwanJang commented 2 years ago

@Wangbk-dl No. I guess there was a mistake implementing the code.

LT1st commented 2 years ago

Error while implementing the code.