DimV-Ai/kuzushiji-character-dataset-v1

该数据集包含历史日本文本中的Kuzushiji字符，提供了每个字符实例的结构化注释，包括字符的Unicode代码点、书籍/手稿ID、页面图像标识符以及字符在页面上的位置。元数据文件（metadata.csv）链接了文本标签和图像数据，支持字符级分类、页面布局重建以及视觉和文本表示的对齐。数据集适用于Kuzushiji OCR和识别、字符级图像分类、视觉-语言/多模态模型、对比或少量学习以及数字人文学科研究。

This dataset contains Kuzushiji characters from historical Japanese texts, providing structured annotations for each character instance, including the Unicode code point of the character, book/manuscript ID, page image identifier, and the characters position on the page. The metadata file (metadata.csv) links text labels to image data, supporting character-level classification, reconstruction of page layout, and alignment between visual and textual representations. The dataset is suitable for Kuzushiji OCR and recognition, character-level image classification, vision-language/multimodal models, contrastive or few-shot learning, and Digital Humanities research on historical Japanese texts.