基于计算机视觉的考勤系统设计与实现

Design and Implementation of Attendance System Based on Computer Vision

针对传统课堂考勤中耗时长、效率低等问题,提出了一种基于计算机视觉的考勤系统,利用深度学习进行人脸识别与手机入袋检测,记录学生的到课情况与手机上交情况。为将考勤信息可视化,设计了3种登录模式的综合考勤系统。实验结果表明,该系统不仅能在毫秒级的时间内完成检测,而且平均准确率(mean Average Precision,mAP)0.5达到0.990,保证了精确率和召回率。

To improve the time-consuming and inefficient problems in traditional classroom attendance,a Computer Vision-based attendance system is proposed that uses Deep Learning for face recognition and mobile phone pocket detection to record students'attendance and mobile phone hand-in.To visualise the attendance information,an integrated attendance system with three login modes is designed.Experimental results show that the system is able to complete the detection in milliseconds with an mean Average Precision(mAP)0.5 of 0.990,while ensuring accuracy and recall.