arXiv cs.AI· Amirmohammad Izadi, Hosein Hasani, Fatemeh Askari, Mobin Bagherian, Sadegh Mohammadian, Mohammad Izadi, Mahdieh Soleymani Baghshah·· 2 天前AI 评分23
揭示 Grounding IDs：外部线索如何塑造多模态绑定
Uncovering Grounding IDs: How External Cues Shape Multi-Modal Binding
AI 导读
研究发现，符号、网格线等外部视觉线索能帮助大型视觉语言模型（LVLM）在物体与其文本描述之间建立更准确的连接。作者提出 Grounding IDs 概念，即外部线索结构化视觉与文本模态后在模型内部产生的潜在标识符，它能增强相关组件间的注意力，改善跨模态对齐并减少模型幻觉。
正文 · 原文
Abstract:Large vision-language models (LVLMs) perform well on multimodal tasks, but their ability to reason and precisely align visual and textual information still has room for improvement. In this study, we show that external visual cues, such as symbols or grid lines, help LVLMs form more accurate connections between visual components, such as objects, and their corresponding textual descriptions, improving their grounding and reasoning abilities. We introduce the concept of Grounding IDs, which are latent identifiers that arise within the model as a result of external cues structuring both visual and textual modalities. Our analysis reveals that partition-inducing external cues lead to Grounding IDs that make better alignment between corresponding visual and text representations, helping the model focus on relevant information. We find that Grounding IDs enhance attention between related components, improving cross-modal grounding and reducing hallucinations. Overall, our results show that Grounding IDs are a key mechanism that enables external cues to improve cross-modal alignment, reduce errors, and enhance the overall performance of LVLMs across a range of multimodal tasks.
| Comments: | Accepted at ICML 2026 |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2509.24072 [cs.CV] |
| (or arXiv:2509.24072v5 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2509.24072 arXiv-issued DOI via DataCite |
Submission history
From: Hosein Hasani [view email]
[v1]
Sun, 28 Sep 2025 21:15:07 UTC (1,997 KB)
[v2]
Fri, 3 Oct 2025 22:57:00 UTC (1,998 KB)
[v3]
Fri, 5 Dec 2025 17:19:01 UTC (2,124 KB)
[v4]
Wed, 25 Feb 2026 01:59:45 UTC (2,115 KB)
[v5]
Wed, 14 Oct 2026 20:57:03 UTC (2,124 KB)
来源：arXiv cs.AI · arxiv.org