Reproduce “RecoverMark: Robust Watermarking for Localization and Recovery of Manipulated Faces” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Haonan An, Xiaohui Ye, Guang Hua, Yihang Tao, Hangcheng Cao, Xiangyu Yu, Yuguang Fang
RecoverMark stores a recoverable face representation in the image background to localize edits and restore the face. It segments the face, compresses it into a latent payload, and trains extraction against image distortions including regeneration. On CelebA after regeneration removal, its localization F1 is 0.721, versus OmniGuard’s 0.064; however, OmniGuard scores 0.911 under patch removal, exceeding RecoverMark’s 0.569.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.