3,506张印地语OCR标注及转写数据【数据堂】

3,506张印地语OCR标注及转写数据包括自然场景图片2,056张，互联网图像1,103张，文本图像347张。在标注方面，行级内容：行级四边形框标注、行级内容转写；竖列内容：竖列四边形框标注、竖列内容转写。 本套印地语OCR标注及转写数据可用于多场景下的印地语识别、印地语拍照翻译等任务

This dataset contains 3,506 Hindi OCR annotated and transcribed samples, consisting of 2,056 natural scene images, 1,103 internet images, and 347 text images. In terms of annotation, line-level content is annotated with quadrilateral bounding boxes and paired with corresponding transcriptions; for vertically arranged column content, quadrilateral bounding box annotation and matching transcriptions are also provided. This Hindi OCR annotated and transcribed dataset can be applied to tasks including multi-scenario Hindi text recognition, Hindi photo translation and other relevant tasks.

搜集汇总

背景与挑战

背景概述

该数据集包含3,506张印地语图像，涵盖自然场景、互联网和文本三类，提供行级和竖列四边形框标注及内容转写，适用于印地语OCR识别和翻译任务。数据以.jpg图片和.json标注格式提供，检测和转写精度均不低于97%。