Reproduce “SpiderMark: Hybrid Watermarking for Generative AI via Algorithmic Injection and Learning-Based Verification” 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

Chihcheng Hsieh, Yang Song, Flora Salim, Martin Bell, Jacinto C. Nascimento, Catarina Moreira

SpiderMark embeds a web-shaped pattern in a diffusion generator’s starting noise and trains a verifier to detect it after image changes. Injection requires no generator retraining, but verification needs a compatible generator and sampling schedule. On Stable Diffusion 2.1, it has the highest fixed-threshold Accuracy in eight of nine tested conditions; Tree-Ring leads on clean images.

Summary by alphaXiv.

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