Reproduce “ProxyEraseAgent: Blind Watermark Removal in the Wild” 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
Jun Yao, Chao Wang, Yupeng Qiu, Zehua Ma, Weiming Zhang, Bin Liu, Han Fang
ProxyEraseAgent uses publicly released watermark decoders as proxies to guide image changes while limiting perceptual distortion. In a blind setting, it ranks decoders by confidence and searches transformations. Across 100 images and 11 watermark systems, it achieves 94.8% cASR versus 79.5% for Regen-VAE, with target-decoder BER at least 0.1 and AlexNet LPIPS at most 0.05.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.