从识别到推理：在现实世界收据文档理解上对齐和增强MLLMs
From Recognition to Reasoning: Benchmarking and Enhancing MLLMs on Real-World Receipt Document Understanding
AI总结：
本文提出ReceiptBench基准，通过四个层次化子任务提升收据信息提取的结构一致性，并提出两阶段训练框架GRPO，通过强化学习信号提升模型性能，实验证明其在复杂推理任务上的优越性。
摘要：
从视觉文档中提取结构化信息（视觉信息提取，VIE）是业务自动化的核心。尽管最近的多模态大语言模型（MLLMs）展示了有前途的能力，但现有基准在规模和现实性方面存在关键限制，缺乏语义粒度，并未能覆盖多样化的文档类型。为弥合这一差距，我们引入ReceiptBench，一个大规模、人工标注的基准，包含10,000种多样化的收据，将信息提取组织成四个层次化子任务：（1）基础感知用于原始文本定位，（2）格式标准化用于严格遵循标准化指令，（3）语义推理用于从上下文中推断隐含属性，（4）结构解析用于处理嵌套的行项。此外，我们提出了一种两阶段训练框架，结合Metric-Aware Group Relative Policy Optimization（GRPO），将严格评估约束转化为强化学习信号以增强结构一致性。广泛的实验表明，我们的方法在复杂推理任务上实现了最先进的性能，超越了领先的专有模型。我们在此发布我们的数据集和代码：https://github.com/wwwT0ri/ReceiptBench。
英文摘要：
Extracting structured information from visual documents (Visual Information Extraction, VIE) is a cornerstone of business automation. While recent Multimodal Large Language Models (MLLMs) have shown promising capabilities, existing benchmarks suffer from critical limitations in scale and realism, lack semantic granularity, and fail to cover diverse document types. To bridge this gap, we introduce ReceiptBench, a large-scale, human-annotated benchmark consisting of 10k diverse receipts, organizing information extraction into four hierarchical sub-tasks: (1) Basic Perception for raw text spotting, (2) Format Normalization for strictly following standardization instructions, (3) Semantic Reasoning for inferring implicit attributes from context, and (4) Structure Parsing for handling nested line items. Furthermore, we propose a two-stage training framework incorporating Metric-Aware Group Relative Policy Optimization (GRPO), which translates rigorous evaluation constraints into reinforcement learning signals to enhance structural consistency. Extensive experiments demonstrate that our method yields state-of-the-art performance, surpassing leading proprietary models on complex reasoning tasks. We release our datasets and code at https://github.com/wwwT0ri/ReceiptBench.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
