Luminark: Training-free, Probabilistically-Certified Watermarking for General Vision Generative Models

2026-01-03

arXiv.org

AI Summary

This work addresses the lack of reliable copyright protection mechanisms in general-purpose visual generative models by proposing a training-free, plug-and-play watermarking method with probabilistic certification. The approach embeds and detects watermarks based on block-level luminance statistics, utilizing predefined binary patterns and thresholds. A watermark-guided mechanism is introduced to ensure compatibility with diverse generative architectures, including diffusion, autoregressive, and hybrid models. Extensive experiments across nine state-of-the-art generative models demonstrate that the method achieves high detection accuracy, strong robustness, and provably low false positive rates, all while preserving high visual fidelity. This provides a universal and dependable solution for content provenance and copyright protection in generated imagery.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Data transparency and provenance

Abstract

In this paper, we introduce \emph{Luminark}, a training-free and probabilistically-certified watermarking method for general vision generative models. Our approach is built upon a novel watermark definition that leverages patch-level luminance statistics. Specifically, the service provider predefines a binary pattern together with corresponding patch-level thresholds. To detect a watermark in a given image, we evaluate whether the luminance of each patch surpasses its threshold and then verify whether the resulting binary pattern aligns with the target one. A simple statistical analysis demonstrates that the false positive rate of the proposed method can be effectively controlled, thereby ensuring certified detection. To enable seamless watermark injection across different paradigms, we leverage the widely adopted guidance technique as a plug-and-play mechanism and develop the \emph{watermark guidance}. This design enables Luminark to achieve generality across state-of-the-art generative models without compromising image quality. Empirically, we evaluate our approach on nine models spanning diffusion, autoregressive, and hybrid frameworks. Across all evaluations, Luminark consistently demonstrates high detection accuracy, strong robustness against common image transformations, and good performance on visual quality.

Problem

• watermarking
• generative models
• probabilistic certification
• training-free
• vision

Innovation

• training-free watermarking
• probabilistically-certified detection
• patch-level luminance statistics
• watermark guidance
• general vision generative models