arXiv cs.AI· Eunchan Jo, Seounghyun Jang, Sangwoo Mo·· 2 天前AI 评分31
Loud and Clear（LoC）：让视觉证据在多模态文档 RAG 中真正被用上
Loud and Clear: Making Visual Evidence Count in Multimodal Document RAG
AI 导读
多模态文档 RAG 的瓶颈不在检索，而在生成器如何读取检索到的页面。研究发现，在 SciEGQA 上把页面视觉 token 预算从 1K 提到 32K 可提升 15.22 分，固定 16K 时加入 ground-truth 裁剪再提升 3.75 分。
正文 · 原文
Abstract:Multimodal document retrieval-augmented generation (RAG) has focused largely on retrieval, yet its answers depend on how well a vision-language model reads the retrieved pages. We find that evidence in retrieved pages remains underused even when the correct pages are retrieved. On SciEGQA oracle pages, five prior methods give the generator 0.7K to 2.7K visual tokens per query on average under a common image processor, yet raising the page-only budget from 1K to 32K improves accuracy by 15.22 points. At a fixed 16K budget, adding ground-truth crops improves accuracy by 3.75 points over page-only input. We study what the generator needs from retrieved pages through two requirements: Loud, making full use of visual token capacity, and Clear, refining the page so that its evidence becomes prominent. For Loud, increasing either the visual token budget or visual fidelity, the image resolution those tokens preserve, improves accuracy with the other held fixed, and full-page scaling outperforms repetition and tiling at matched budgets. For Clear, suppressing surrounding content improves accuracy without enlarging selected regions, and adding OCR of these regions to page images outperforms either input alone. We implement both as Loud and Clear (LoC), which combines higher-resolution pages with OCR from query-relevant regions without ground-truth annotations or additional training. LoC outperforms eight prior methods on SciEGQA, ViDoRe-v3, and MMLongBench-Doc, improving over the strongest, ColQwen3, by 8.88, 3.09, and 4.22 points with the same retriever and generator. It also outperforms its page-only variant under the same visual token budget. These results locate a bottleneck beyond retrieval: how the retrieved pages are given to the generator.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.21318 [cs.CV] |
| (or arXiv:2610.21318v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.21318 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Eunchan Jo [view email]
[v1]
Thu, 15 Oct 2026 06:08:04 UTC (979 KB)
来源：arXiv cs.AI · arxiv.org