SGDNet
SGDNet copied to clipboard
Could you tell me the value of alpha for each dataset?
I have tried your code in Koniq10k and LIVEc. In Koniq10k, the result is close to your paper while LIVEc is not. I guess this is because of the alpha value. Since if mostype=ss(default), the mos range of Koniq10k is [1,5], but LIVEc is [0,100], if alpha remains same in LIVEc (0.25 as default), the loss of saliency map may be ignored since is too small than the loss of image quality. So could you tell me the value of alpha for each dataset or any other way to improve the performance? Every suggestion wolud be helpful, thanks. @ysyscool
Can you tell me how could get the sim image? I can not find it in the released code.
Can you tell me how could get the sim image? I can not find it in the released code.
Please check this link (https://drive.google.com/drive/folders/1EoXpuEBAUPNKPh-17u8WRyHCCg9PKKgT)
In truth, I mean how could you get these image , maybe grad_cam?
In truth, I mean how could you get these image , maybe grad_cam?
I use my saliency model (https://github.com/ysyscool/DINet) to get these images.