arXiv cs.CL· Xiaomeng Wang, Martha Larson, Zhengyu Zhao·· 12 小时前AI 评分31
视觉文本样式如何影响大型视觉语言模型的属性描述
Revealing the Impact of Visual Text Style on Attribute-based Descriptions Produced by Large Visual Language Models
AI 导读
研究发现，即使大型视觉语言模型（LVLM）正确识别出图像中文字所指的概念，文字的视觉样式仍会影响其对该概念的属性描述。实验对比了功能性样式（如黑色无衬线字体）与装饰性样式（如彩色手写体），证实存在非平凡的样式泄漏，即文本样式渗入语义推理。该成果已被 ICMR 2026 接收，作者呼吁对基于 LVLM 的多媒体系统开展样式感知的评估与缓解。
正文 · 原文
Abstract:When the visual style of text is considered, a wide variety can be observed in font, color, and size. However, when a word is read, its meaning is independent of the style in which it has been written or rendered. In this paper, we investigate whether, and how, the style in which a word is visualized in an image impacts the description that a Large Visual Language Model (LVLM) provides for the concept to which that word refers. Specifically, we investigate how functional text styles (readability-oriented, e.g., black sans-serif) versus decorative styles (display-oriented, e.g., colored cursive/script) affect LVLMs' descriptions of a concept in terms of the attributes of that concept. Our experiments study the situation in which the LVLM is able to correctly identify the concept referred to by a visual text, i.e., by a word or words rendered as an image, and in which the visual text style should not influence the attribute-based description that the LVLM produces. Our experimental results reveal that even when the concept is correctly identified, text style influences the model's attribute-based descriptions of the concept. Our findings demonstrate non-trivial style leakage from text style into semantic inference and motivate style-aware evaluation and mitigation for LVLM-based multimedia systems.
| Comments: | Accepted by ICMR 2026. Code is available at this https URL |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL) |
| Cite as: | arXiv:2604.27553 [cs.CV] |
| (or arXiv:2604.27553v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2604.27553 arXiv-issued DOI via DataCite |
Submission history
From: Xiaomeng Wang [view email]
[v1]
Thu, 30 Apr 2026 08:01:45 UTC (795 KB)
来源：arXiv cs.CL · arxiv.org