摘要
针对传统方法的人体动态行为智能识别方法存在识别率较低等问题,提出基于神经网络的人体动态行为智能识别方法。对人体动态行为数据预处理,并构建人工神经网络模型,实现人工神经网络训练以及特征提取;将视频的光流图像放置于卷积神经网络模型中,获取图像的时域特征;融合人工神经网络特征与时域特征,并将其放入SVM中进行类别划分,实现基于神经网络的人体动态行为智能识别。仿真实验研究结果表明,所提方法能够有效提升人体动态行为识别准确率,并且整个方法的综合性能较好。
Aiming at the problems of low recognition rate of human body dynamic behavior intelligent recognition method based on traditional methods,a neural network based intelligent recognition method for human body dynamic behavior is proposed.Preprocessing human dynamic behavior data and constructing artificial neural network model to realize artificial neural network training and feature extraction.The optical flow image of the video is placed in a convolutional neural network model to acquire the temporal characteristics of the image.Combining the characteristics of artificial neural network and time domain features,and putting them into SVM for class division,realize intelligent identification of human body dynamic behavior based on neural network.The simulation experiment results show that the proposed method can effectively improve the accuracy of human body dynamic behavior recognition,and the overall performance of the whole method is better.
作者
贾双成
杨凤萍
Jia Shuangcheng;Yang Fengping(Alibaba Network Technology Co.,Ltd.,Beijing 100102,China;College of Internet of Things,Jiangnan University,Wuxi 214122,China)
出处
《科技通报》
2020年第1期60-63,共4页
Bulletin of Science and Technology
关键词
神经网络
人体动态行为
智能识别方法
neural network
human body dynamic behavior
intelligent recognition method