基于汽车前排驾乘人员的面部与坐姿智能识别方法研究与应用  

Research and Application of Intelligent Recognition Method of Face and Sitting Position Based on Front-seat Driver and Passenger

作　　者：谢智峰 项菲菲[1] 许德衡 甘文谨 XIE Zhifeng;XIANG Feifei;XU Deheng;GAN Wenjin(Jiangxi University of Technology,Nanchang 330098,China)

机构地区：[1]江西科技学院,江西南昌330098

出　　处：《现代信息科技》2024年第23期87-92,99,共7页Modern Information Technology

基　　金：江西省教育厅科学技术研究项目(GJJ2202620)。

摘　　要：为了能够实时准确地检测出驾驶员的疲劳状态及驾驶行为,保障汽车前排驾乘人员的安全驾驶,设计了一种面部与坐姿智能识别方法,用于实时判定前排驾乘人员的疲劳状态及驾驶行为。基于不同结构的神经网络,捕捉面部和行为特征,定位头部位置和行为状态,推断疲劳程度和全身坐姿。实验结果表明,打哈欠疲劳测试结果准确率高于长时间闭眼;My_Dense模型测试识别优于其他两种坐姿检测模型。因此,面部与坐姿智能识别方法能够有效地判定驾驶员的疲劳状态及驾驶行为,提供精准辅助与安全预警。In order to detect the fatigue state and driving behavior of drivers in real time and accurately,and ensure the safe driving of front-seat drivers and passengers,an intelligent recognition method of face and sitting position is designed to determine the fatigue state and driving behavior of front-seat drivers and passengers in real time.Neural networks based on different structures capture facial and behavioral features,locate head position and behavioral state,and infer fatigue level and whole body sitting position.The experimental results show that the accuracy of yawning fatigue test is higher than that of prolonged closed eyes.My_Dense model test recognition is better than the other two sitting position detection models.Therefore,the intelligent recognition method of face and sitting position can effectively determine the driver's fatigue state and driving behavior,and provide accurate assistance and safety early warning.

关 键 词：前排驾乘人员 面部识别 坐姿识别 驾驶行为 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]