数据堂—5,147张日语手写体OCR数据

5,147张日语手写体OCR数据。文字载体为A4纸、横格纸、方格纸等。采集设备为手机，采集角度为平视视角。该数据内容涵盖日语作文、诗歌、散文、新闻、故事等领域。在标注方面，标注行级文本的四边形框，行级文本转写。本套数据可用于日语手写体OCR任务

This is a Japanese handwritten OCR dataset containing 5,147 samples. The text carriers include A4 paper, ruled paper, grid paper and other types of paper. The data was collected using mobile phones with an eye-level shooting perspective. The dataset covers various content domains such as Japanese compositions, poems, prose, news articles, stories and more. For annotation, quadrilateral bounding boxes for line-level text and corresponding line-level text transcriptions are provided. This dataset can be used for Japanese handwritten OCR tasks.

搜集汇总

背景与挑战

背景概述

该数据集包含5,147张日语手写体图像，用于手写OCR测试任务，内容涵盖散文、诗歌和新闻等多种体裁。图像通过手机从正面视角采集，标注包括行级四边形边界框和对应文本转录。