search

zyainfal / One-Shot-Face-Swapping-on-Megapixels

One Shot Face Swapping on Megapixels.
Other
314 stars 40 forks source link

About the "Objective function of HieRFE"? #16

Closed jo-dean closed 3 years ago

jo-dean commented 3 years ago

hello, when i read the paper,i found the " a pair of input image x and its reconstruction image ˆx". Does the "x" mean the cat(S_img,T_img)? And what's the meaning of "reconstruction image ˆx"? How to generate it? thank you so much!

zyainfal commented 3 years ago

In reconstruction training, there is no concept of source and target. X means individual face image.

For generating X_hat, we use HieRFE to encode a face image X to its latent codes, then feed StyleGAN2 these latent codes to get reconstructed face image X_hat.

jo-dean commented 3 years ago

Thanks for your quick reply,it's a great job.