When Surfaces Lie: Exploiting Wrinkle-Induced Attention Shift to Attack Vision-Language Models
It reveals a new vulnerability in VLMs to physically plausible non-rigid deformations, which is important for robustness research but is an incremental contribution as it applies existing adversarial attack concepts to a new perturbation type.
This paper introduces a parametric structural perturbation method that generates photorealistic wrinkles on flexible surfaces to attack Vision-Language Models (VLMs). The method significantly degrades VLM performance in zero-shot classification, image captioning, and visual question answering, outperforming baselines.
Visual-Language Models (VLMs) have demonstrated exceptional cross-modal understanding across various tasks, including zero-shot classification, image captioning, and visual question answering. However, their robustness to physically plausible non-rigid deformations-such as wrinkles on flexible surfaces-remains poorly understood. In this work, we propose a parametric structural perturbation method inspired by the mechanics of three-dimensional fabric wrinkles. Specifically, our method generates photorealistic non-rigid perturbations by constructing multi-scale wrinkle fields and integrating displacement field distortion with surface-consistent appearance variations. To achieve an optimal balance between visual naturalness and adversarial effectiveness, we design a hierarchical fitness function in a low-dimensional parameter space and employ an optimization-based search strategy. We evaluate our approach using a two-stage framework: perturbations are first optimized on a zero-shot classification proxy task and subsequently assessed for transferability on generative tasks. Experimental results demonstrate that our method significantly degrades the performance of various state-of-the-art VLMs, consistently outperforming baselines in both image captioning and visual question-answering tasks.