search

taotaonice / FaceShifter

Try to reproduce FaceShifter
250 stars 62 forks source link

Different loss functions #29

Open lo2aayy opened 4 years ago

lo2aayy commented 4 years ago

Why are the L_rec and L_att loss functions implemented different than that of the paper?

in the paper L_att, and L_rec are the L2 distance but you do squaring followed by meaning Shouldn't they be for L2 distance: torch.norm((X - Y),2) instead of torch.mean(torch.pow(X - Y, 2))? The difference in the loss between them is large.

tangxuvis commented 4 years ago

i have implemented some perfomance, can we chat on email about more details? @lo2aayy my email address: tangxuvis@163.com

wvinzh commented 4 years ago

Why are the L_rec and L_att loss functions implemented different than that of the paper?

in the paper L_att, and L_rec are the L2 distance but you do squaring followed by meaning Shouldn't they be for L2 distance: torch.norm((X - Y),2) instead of torch.mean(torch.pow(X - Y, 2))? The difference in the loss between them is large.

I think torch.mean should be torch.sum,torch.norm((X - Y),2) is equal to torch.sqrt(torch.sum(torch.pow(X - Y, 2)))