1.This work selectively takes the difference between target and source attribute vectors as input. Selective transfer units(STU) are used with encoder-decoder to adaptively select and modify encoder feature for enhanced attribute editing.
2.In terms of selective, STGAN is suggested to (i) only consider the attributes to be changed, and (ii) selectively concatenate encoder feature in editing attribute irrelevant regions with decoder feature. In terms of transfer, STGAN is expected to adaptively modify encoder feature to match the requirement of varying editing task, thereby providing a unified model for handling both local and global attributes.
3.Tried to resolve problem with Skip Connections in AttGAN
4.For arbitrary image attribute editing, instead of full target attribute vector, only the attributes to be changed is considered.
5.STGAN Architecture:
6.Comparison:
7.13 attributes, including Bald, Bangs, Black Hair, Blond Hair, Brown Hair, Bushy Eyebrows, Eyeglasses, Male, Mouth Slightly Open, Mustache, No Beard, Pale Skin and Young were investigated.
8.Example of testing single attribute(Gender):
 |
 |
|---|---|
| Img-1 | Img-2 |
 |
 |
|---|---|
| Img-3 | Img-4 |