Reproduce “Transferable Black-Box One-Shot Forging of Watermarks via Image Preference Models” 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

Tomáš Souček, Sylvestre-Alvise Rebuffi, Pierre Fernandez, Nikola Jovanović, Hady Elsahar, Valeriu Lacatusu, Tuan Tran, Alexandre Mourachko

WmForger estimates a post-hoc image watermark from one watermarked example. A preference model trained on synthetic artifacts guides pixel optimization; the estimated signal transfers to another image. Across four watermarkers, it reached 0.83 bit accuracy on Video Seal, versus 0.59 from averaging 100 images and 0.74 from Warfare using 1,000 images; findings are limited to tested systems.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.