摘要
根据行人步态具有唯一性,且不受客观因素影响的特点,利用OpenPose提取行人的骨架信息,结合时空图卷积神经网络模型,构建步态识别模型,在步态识别网络中融入注意力机制,可以有效地避免系统陷入局部最优解,使分类模型可以结合骨架的局部信息和全局信息,构建丰富的网络模型,提高识别系统的泛化能力,实验结果显示,达到了较为理想的步态识别效果。
According to the unique characteristics of pedestrian gait,which is not affected by objective factors,OpenPose is used to extract the skeleton information of pedestrians.The gait recognition model is constructed by OpenPose and the spatiotempo‑ral graph convolutional neural network,which can effectively avoid the system falling into the local optimal solution.The classifica‑tion model can combine the local and global information of the skeleton,build a rich network model,and improve the generalization ability of the recognition system.The experimental results show that the ideal gait recognition effect is achieved.
作者
刘艳丽
Liu Yanli(School of Public Administration,Chongqing Vocational College of Public Transportation,Chongqing 402247,China)
出处
《现代计算机》
2024年第3期66-70,共5页
Modern Computer
基金
重庆公共运输职业学院第二批青年骨干教师培养计划。
