Hi,
In the paper "ArcFace: Additive Angular Margin Loss for Deep
Face Recognition" is written that " the pre-trained ArcFace
model can generate identity-preserved and visually reasonable
face images only by using the gradient and BN priors."
How it can be done ? It can generate the face from the embeddings even if the face was not in training data ?
Thanks in advance
Hi, In the paper "ArcFace: Additive Angular Margin Loss for Deep Face Recognition" is written that " the pre-trained ArcFace model can generate identity-preserved and visually reasonable face images only by using the gradient and BN priors." How it can be done ? It can generate the face from the embeddings even if the face was not in training data ? Thanks in advance