arXiv：cs.AI· Hyeonggeun Han, Taehyun Cho, Sangwoo Hong·· 6 小时前
EGVCA：面向大型视觉语言模型的无训练幻觉缓解框架
From Token Correction to Claim Admission: Training-Free Hallucination Mitigation for Large Vision-Language Models
AI 导读
研究者提出无训练框架 EGVCA（Evidence-Guided Visual Claim Admission），通过对比图像输入与视觉空输入下的声明似然，仅在语义声明获得足够图像相关支持时才予以采纳。
正文
Abstract:Large vision-language models often exhibit object hallucination, generating claims that are not supported by the input image. To mitigate this issue, we propose Evidence-Guided Visual Claim Admission (EGVCA), a training-free framework that admits a semantic claim only when it receives sufficient image-dependent support. EGVCA measures this support by comparing the claim likelihood under the observed image and a visual-null input. For captioning, it controls semantic expansion without altering token generation. For binary visual question answering, it combines atomic visual support with the image-induced change in the model's native Yes--No margin. Experiments on CHAIR, POPE, and MME across three LVLM backbones show that EGVCA substantially reduces object hallucination while preserving caption quality. It achieves the highest POPE F1 in eight of nine backbone--sampling settings and improves the native MME total score on all three backbones. These results demonstrate the effectiveness of claim-level admission as a simple alternative to token-level distribution correction.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.17706 [cs.CV] |
| (or arXiv:2610.17706v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.17706 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Sangwoo Hong [view email]
[v1]
Tue, 13 Oct 2026 04:35:53 UTC (1,792 KB)
来源：arXiv：cs.AI · arxiv.org