Skip to content

Latest commit

 

History

History
7 lines (4 loc) · 403 Bytes

230626 Localized Text-to-Image Generation for Free via Cross Attention Control.md

File metadata and controls

7 lines (4 loc) · 403 Bytes

https://arxiv.org/abs/2306.14636

Localized Text-to-Image Generation for Free via Cross Attention Control (Yutong He, Ruslan Salakhutdinov, J. Zico Kolter)

text2img 모델에서 segmentation map 등을 사용한 localization 기능 추가. image-text cross attention을 segmentation map을 사용해 마스킹하는 방식으로 조작하는 접근이군요. 재미있네요.

#text2img #image_editing