OCR识别-港澳居民来往内地通行证OCR识别

依托前沿深度学习，利用光学字符识别（OCR）技术。通过采集自然场景下港澳居民来往内地通行证OCR识别的扫描图片，可自动捕捉通行证的边缘、校正透视变形图像，支持任意方向、正反图像自动识别。对通行证的有效信息行智能检测、定位、识别，准确率可达99%以上。Leveraging cutting-edge deep learning technologies and Optical Character Recognition (OCR) techniques, this dataset is developed by collecting scanned images for OCR recognition of Home Return Permits for Hong Kong and Macao residents traveling to the Chinese mainland under natural scenes. It can automatically capture the edges of the permits, correct perspective-distorted images, and support automatic recognition of images in any orientation as well as both front and back sides. The dataset intelligently detects, locates and recognizes the valid information lines on the permits, achieving an accuracy rate of over 99%.

搜集汇总

背景与挑战

背景概述

该数据集采用深度学习OCR技术，通过自然场景下的港澳通行证扫描图像，实现自动边缘捕捉、图像校正及多方向识别，对证件信息的检测定位准确率超过99%。