.png){kind=icon}
返回
Validity-aware context modeling for gradient-guided image inpainting
.png){kind=icon}
PRE
AI
DOI：10.1016/j.inffus.2026.104162.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
• We propose a gradient-guided image inpainting method that achieves state-of-the-art results. • The repaired gradient information provides structure and detail priors for image inpainting. • We introduce a gradient-driven attention module that uses the attention property of gradients. • We introduce a context validity-aware module to progressively fill the hole regions of the image. • We enable user-driven image editing by controlling the generation of gradients through sketches.
