研究人员开发了CFGPatch，一个新颖的对抗性框架，旨在揭示可见-红外视觉语言模型（VLMs）的漏洞。该方法利用曲边分形几何和特定模态的渲染机制来创建对抗性补丁，从而干扰VLMs的形状和纹理感知。实验表明，CFGPatch能有效地欺骗这些模型，并在图像字幕和视觉问答等不同任务中表现出强大的可迁移性。 AI
影响 这项研究突显了在复杂环境中运行的多模态AI系统潜在的安全风险，表明需要更强大的对抗性防御措施。
排序理由 该集群包含一篇学术论文，详细介绍了一种用于计算机视觉模型的新型对抗性框架。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.22273v1 Announce Type: new Abstract: Vision-language models (VLMs) have achieved strong performance across diverse multimodal tasks, but their adversarial robustness in visible-infrared (VIS-IR) scenarios remains underexplored. This gap is critical because VIS-IR sensi…
Vision-language models (VLMs) have achieved strong performance across diverse multimodal tasks, but their adversarial robustness in visible-infrared (VIS-IR) scenarios remains underexplored. This gap is critical because VIS-IR sensing is widely used in real-world perception syste…