Research on Structured Information Extraction of Second Generation ID Cards Based on Multi-task Learning
This paper designs and implements an end-to-end ID card recognition system.The system improves the traditional OCR recognition process,adopts a multi-task deep learning algorithm,integrates target detection and structural analysis into the same deep network,then recognizes the detected text,and finally outputs the structured information of the ID card.The experimen-tal results show that compared with the traditional ID card template matching method,this method greatly improves the final recogni-tion effect,and also achieves good results for ID card recognition in complex scenarios.
国家科技期刊平台
NSTL
万方数据
