摘要
针对传统人脸识别系统画面异常卡顿和准确率低等问题,将目标追踪技术运用到基于深度学习的人脸识别系统来提高性能。该系统包括人脸图像录入、人脸识别以及人脸数据库管理三部分,图像经过人脸采集后录入,提取特征值到数据库中,然后使用图像识别、视频识别和实时摄像头识别功能从不同光照、距离和遮挡角度对系统进行人脸识别测试。测试结果表明,该系统达到了预期要求,能够满足实际需要。
Aiming at the problems of abnormal picture delay and low accuracy of traditional face recognition system,target tracking technology is applied to deep learning-based face recognition system to improve the performance.The system includes face image entry,face recognition and face database management three parts,the image after face acquisition input,extract the feature value to the database,and then use the image recognition,video recognition and real-time camera recognition function from different lighting,distance and occlusion angle to face recognition test.The test results showed that the system achieved the expected requirements and could meet the actual needs.
作者
何婧媛
李沐阳
田原
田琴琴
HE Jingyuan;LI Muyang;TIAN Yuan;TIAN Qinqin(School of Mathematics and Computer Science,Yan’an University,716000,Yan’an,Shanxi,PRC)
出处
《江西科学》
2023年第2期400-404,共5页
Jiangxi Science
基金
陕西省教育厅2022年度一般专项科研计划项目(22JK0616)。
关键词
深度学习
人脸识别
人脸检测
人脸匹配
目标追踪
deep learning
face recognition
face detection
face matching
target tracking
