Finding Needles in a Haystack: A Black-Box Approach to Invisible Watermark Detection

Abstract

In this paper, we propose WaterMark Detector ($\textsc {WMD}$), the first invisible watermark detection method under a black-box and annotation-free setting. $\textsc {WMD}$ is capable of detecting arbitrary watermarks within a given detection dataset using a clean non-watermarked dataset as a reference, without relying on specific decoding methods or prior knowledge of the watermarking techniques. We develop $\textsc {WMD}$ using foundations of offset learning, where a clean non-watermarked dataset enables us to isolate the influence of only watermarked samples in the reference dataset. Our comprehensive evaluations demonstrate the effectiveness of $\textsc {WMD}$, which significantly outperforms naive detection methods with AUC scores around only 0.5. In contrast, $\textsc {WMD}$ consistently achieves impressive detection AUC scores, surpassing 0.9 in most single-watermark datasets and exceeding 0.7 in more challenging multi-watermark scenarios across diverse datasets and watermarking methods. As invisible watermarks become increasingly prevalent, while specific decoding techniques remain undisclosed, our approach provides a versatile solution and establishes a path toward increasing accountability, transparency, and trust in our digital visual content.

Work done during Minzhou Pan’s and Zhenting Wang’s internship at Sony AI.