57,645张竖排文字场景OCR数据【数据堂】

57,645张竖排文字场景OCR数据。采集场景包括街景、牌匾、广告牌、海报、装饰、艺术字、杂志封面等。语言分布为中文和英文（少量）。在标注方面，列级矩形框（多边形框、平行四边形框）标注、列级内容转写；非竖排矩形框（多边形框、平行四边形框）标注、非竖排内容转写 。本套数据可用于多种竖排文字场景OCR任务

57,645 vertical text scene OCR samples. The data collection scenarios cover street scenes, plaques, billboards, posters, decorations, artistic fonts, magazine covers, and other scenarios. The languages involved are Chinese and a small amount of English. For annotation, column-level rectangular bounding boxes (including polygonal boxes and parallelogram boxes) are utilized, along with transcription of the corresponding column-level content; similarly, non-vertical rectangular bounding boxes (including polygonal boxes and parallelogram boxes) are utilized, along with transcription of the corresponding non-vertical content. This dataset can be applied to various vertical text scene OCR tasks.

搜集汇总

背景与挑战

背景概述

该数据集包含57,645张竖排文字场景图像，总计528,553个标注框，覆盖街景、牌匾、广告牌等多种场景，以中文为主、英文为辅。标注内容包括列级和非竖排的矩形、多边形、平行四边形框及文本转写，检测和转写精度均不低于97%，为商业数据仅限企业合作购买。