标题： DRISHTIKON：文档中多粒度的视觉定位 显示英文标题

标题： DRISHTIKON: Visual Grounding at Multiple Granularities in Documents

摘要： 文本丰富文档图像中的视觉定位是文档智能和视觉问答（VQA）系统中一个关键但研究不足的挑战。 我们提出了DRISHTIKON，一种多粒度和多块的视觉定位框架，旨在增强复杂、多语言文档中VQA的可解释性和信任度。 我们的方法结合了多语言OCR、大语言模型和一种新颖的区域匹配算法，以在块、行、词和点级别定位答案片段。 我们引入了 多粒度视觉定位（MGVG）基准，这是一个精心整理的测试集，包含来自不同领域的多样化圆形通知，每个都经过人工标注，在多个粒度上具有细粒度的人工验证标签。 大量实验表明，我们的方法实现了最先进的定位准确性，其中行级粒度在精度和召回率之间提供了最佳平衡。 消融研究进一步突显了多块和多行推理的好处。 比较评估显示，领先的视觉语言模型在精确定位方面存在困难，这强调了我们结构化、对齐方法的有效性。 我们的研究结果为在具有多粒度定位支持的真实世界、以文本为中心的场景中更强大和可解释的文档理解系统铺平了道路。 代码和数据集已提供给未来的研究。 显示英文摘要

摘要： Visual grounding in text-rich document images is a critical yet underexplored challenge for Document Intelligence and Visual Question Answering (VQA) systems. We present DRISHTIKON, a multi-granular and multi-block visual grounding framework designed to enhance interpretability and trust in VQA for complex, multilingual documents. Our approach integrates multilingual OCR, large language models, and a novel region matching algorithm to localize answer spans at the block, line, word, and point levels. We introduce the Multi-Granular Visual Grounding (MGVG) benchmark, a curated test set of diverse circular notifications from various sectors, each manually annotated with fine-grained, human-verified labels across multiple granularities. Extensive experiments show that our method achieves state-of-the-art grounding accuracy, with line-level granularity providing the best balance between precision and recall. Ablation studies further highlight the benefits of multi-block and multi-line reasoning. Comparative evaluations reveal that leading vision-language models struggle with precise localization, underscoring the effectiveness of our structured, alignment-based approach. Our findings pave the way for more robust and interpretable document understanding systems in real-world, text-centric scenarios with multi-granular grounding support. Code and dataset are made available for future research.