A Visual Semantic Adaptive Watermark grounded by Prefix-Tuning for Large Vision-Language Model

TL;DR

This paper introduces VISA-Mark, a novel watermarking framework for large vision-language models that embeds detectable signals aligned with visual evidence, ensuring high visual fidelity, robustness, and efficiency.

Contribution

VISA-Mark employs a lightweight prefix-tuner and adaptive mechanisms to embed semantic-aware watermarks without disrupting visual grounding or incurring high inference latency.

Findings

7.8% improvement in visual consistency on Chair-I dataset

96.88% detection accuracy (AUC)

99.3% robustness against attacks

Abstract

Watermarking has emerged as a pivotal solution for content traceability and intellectual property protection in Large Vision-Language Models (LVLMs). However, vision-agnostic watermarks introduce visually irrelevant tokens and disrupt visual grounding by enforcing indiscriminate pseudo-random biases, while some semantic-aware methods incur prohibitive inference latency due to rejection sampling. In this paper, we propose the VIsual Semantic Adaptive Watermark (VISA-Mark), a novel framework that embeds detectable signals while strictly preserving visual fidelity. Our approach employs a lightweight, efficiently trained prefix-tuner to extract dynamic Visual-Evidence Weights, which quantify the evidentiary support for candidate tokens based on the visual input. These weights guide an adaptive vocabulary partitioning and logits perturbation mechanism, concentrating watermark strength specifically on visually-supported tokens. By actively aligning the watermark with visual evidence, VISA-Mark effectively maintains visual fidelity. Empirical results confirm that VISA-Mark outperforms conventional methods with a 7.8% improvement in visual consistency (Chair-I) and superior semantic fidelity. The framework maintains highly competitive detection accuracy (96.88% AUC) and robust attack resilience (99.3%) without sacrificing inference efficiency, effectively establishing a new standard for reliability-preserving multimodal watermarking.