Image Editing
Image editing research focuses on developing efficient and controllable methods for modifying images based on user instructions, whether textual, visual (e.g., drag-and-drop), or a combination. Current efforts concentrate on improving the precision and speed of editing, particularly using diffusion models and incorporating techniques like prompt engineering, geometric loss functions, and attention mechanisms to enhance realism and background preservation. These advancements are significant for various applications, including content creation, medical imaging analysis, and the detection of image forgeries, improving both the quality and accessibility of image manipulation tools.
Papers
Prompt Augmentation for Self-supervised Text-guided Image Manipulation
Rumeysa Bodur, Binod Bhattarai, Tae-Kyun Kim
Unsupervised Region-Based Image Editing of Denoising Diffusion Models
Zixiang Li, Yue Song, Renshuai Tao, Xiaohong Jia, Yao Zhao, Wei Wang
Defending LVLMs Against Vision Attacks through Partial-Perception Supervision
Qi Zhou, Tianlin Li, Qing Guo, Dongxia Wang, Yun Lin, Yang Liu, Jin Song Dong