基于3DCNN的驾驶员细微动作识别  

Driver subtle action recognition based on 3DCNN

作　　者：秦康 张小俊[1] 张明路[1] 杨亚昆 Qin Kang;Zhang Xiaojun;Zhang Minglu;Yang Yakun(College of Mechanical Engineering,Hebei University of Technology,Tianjin 300000,China)

机构地区：[1]河北工业大学机械工程学院,天津300000

出　　处：《电子测量技术》2023年第8期51-58,共8页Electronic Measurement Technology

基　　金：天津市新一代人工智能科技重大专项(18ZXZNGX00230)资助。

摘　　要：针对驾驶员相似的背景下的细微动作的动作识别,提出了一种基于X3D卷积神经网络X3D-M-GC-AE。通过引入轻量级的自注意力网络GCNet,提高对时间和空间关键特征的关注度,不引入参数量的情况下,提高检测精度;设计了一种运动增强模块,使网络对时序上的运动信息更加敏感;引入知识蒸馏,将X3D-XL作为教师网络,X3D-M-GC-AE作为学生网络,可以使用较少的参数量和计算量,达到可以实车应用的程度。实验结果表明教师网络测试精度最高可以达到75.56%,学生网络最高可以达到71.13%,该框架在车载硬件设备要求较低的情况下能够实现较高精度的检测效果。Aiming at the action recognition of subtle actions in the similar background of drivers,X3D-M-GC-AE based on X3Dnetwork is proposed.By introducing the lightweight self-attention network GCnet,the attention to key features in time and space is improved,and the detection accuracy is improved without increasing parameter quantities.Action enhancement block is designed to make the network more sensitive to the action information in time series.Introducing knowledge distillation,taking X3D-XL as the teacher network and X3D-M-GC-AE as the student network,so that X3D-M-GC-AE can be used in real vehicles with less parameters and calculations.The experimental results show that the maximum test accuracy of teacher network can reach 75.56%,and that of student network can reach 71.13%.This framework can achieve high-precision detection results in the case of low requirements for vehicle hardware equipment.

关 键 词：驾驶员动作识别 时间序列 X3D 注意力机制 知识蒸馏 深度学习 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] U463.6[自动化与计算机技术—计算机科学与技术]