Hide&Seek: Remove Image Watermarks with Negligible Cost via Pixel-wise Reconstruction
Huajie Chen, Tianqing Zhu, Hailin Yang, Yuchen Zhong, Yang Zhang, Hui Sun, Heng Xu, Zuobin Ying, Lihua Yin, Wanlei Zhou

TL;DR
This paper introduces HIDE&SEEK, a set of efficient attacks that can effectively remove watermarks from images with minimal visual quality loss, highlighting the vulnerabilities of current watermarking defenses.
Contribution
The paper presents a novel, versatile attack suite that challenges the robustness of existing image watermarking methods with negligible computational cost.
Findings
HIDE&SEEK reliably removes watermarks from images.
The attacks maintain high visual fidelity post-removal.
Current watermarking defenses are vulnerable to these attacks.
Abstract
Watermarking has emerged as a key defense against the misuse of machine-generated images (MGIs). Yet the robustness of these protections remains underexplored. To reveal the limits of SOTA proactive image watermarking defenses, we propose HIDE&SEEK (HS), a suite of versatile and cost-effective attacks that reliably remove embedded watermarks while preserving high visual fidelity.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Steganography and Watermarking Techniques · Digital Media Forensic Detection · Adversarial Robustness in Machine Learning
