基于多特征的非接触式学习疲劳检测分析  

Analysis of Non-contact Learning Fatigue Detection Based on Multiple Features

作　　者：范凌云[1] FAN Lingyun(Chongqing City Vocational College,Chongqing 402160,China)

机构地区：[1]重庆城市职业学院,重庆402160

出　　处：《电子技术（上海）》2024年第9期345-347,共3页Electronic Technology

基　　金：重庆市教育委员会科学技术研究计划资助项目(KJQN202203905)。

摘　　要：阐述一种多特征的非接触式学习疲劳检测方法。通过OpenCV对视频流进行人脸检测,基于Dlib定位面部特征点,再计算左右眼的平均EAR值、嘴部的MAR值,进而以30s为单位时间计算PERCLOS值,并统计单位时间内打哈欠持续图像帧数T,最后将PERCLOS值、打哈欠持续图像帧数T分别与对应阀值进行比较,实现对学习疲劳的判断。This paper describes a multi feature non-contact learning fatigue detection method. Firstly, face detection is performed on the video stream using OpenCV. Then, facial feature points are located based on Dlib. Based on this, the average EAR value of the left and right eyes and the MAR value of the mouth are calculated. Then, the PERCROS value is calculated in units of 30 seconds, and the number of frames per unit time for yawning is calculated. Finally, the PERCROS value and the number of frames per unit time for yawning are compared with the corresponding threshold to achieve learning fatigue judgment.

关 键 词：图像识别 疲劳检测 多特征 学习疲劳检测 EAR MAR PERCLOS 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]