llm-jp/jawildtext

JaWildText是一个日语场景文本理解基准，用于评估视觉语言模型在文本丰富的真实世界图像上的表现。它旨在诊断日语光学字符识别、场景文本视觉问答以及在实际条件下（如密集标牌、手写文本和移动设备拍摄的收据）的结构化信息提取。该数据集包含三个互补的评估任务：密集场景文本视觉问答/标牌视觉问答、收据关键信息提取和手写日语光学字符识别。数据集包含3,241个示例，涵盖三个任务配置，其中default配置是所有任务子集的并集。

JaWildText is a Japanese scene text understanding benchmark for evaluating vision-language models (VLMs) on text-rich real-world images. It is designed to diagnose Japanese OCR, scene text visual question answering, and structured information extraction in practical conditions such as dense signboards, handwritten text, and mobile-captured receipts. The dataset contains 3,241 examples across the three task configurations. The `default` configuration is the union of all task subsets.