基于强化学习的多模态场景人体危险行为识别方法  

Recognition Method of Human Dangerous Behavior in Multimodal Scenes Using Reinforcement Learning

作　　者：张晓龙 王庆伟[2] 李尚滨[3] ZHANG Xiaolong;WANG Qingwei;LI Shangbin(P.E.Department,Northeast Forestry University,Harbin 150040,Heilongjiang,China;Physical Education Department,Harbin Huade University,Harbin 150025,Heilongjiang,China;Physical Education Department,Harbin Engineering University,Harbin 150001,Heilongjiang,China)

机构地区：[1]东北林业大学体育部,黑龙江哈尔滨150040 [2]哈尔滨华德学院体育教研部,黑龙江哈尔滨150025 [3]哈尔滨工程大学体育部,黑龙江哈尔滨150001

出　　处：《应用科学学报》2021年第4期605-614,共10页Journal of Applied Sciences

基　　金：国家自然科学基金(No.61163025)资助。

摘　　要：在多模态场景下,常规人体危险行为识别方法对人体危险行为的识别精度较低,于是提出了基于强化学习的多模态场景人体危险行为识别方法。首先根据强化学习的特征提取算法获取多模态场景人体危险行为特征集,其次基于强化学习数据决策提取多模态场景人体危险行为,构建人体危险行为模糊识别模型。最后将上述人体危险行为特征子集代入模型,计算不同感官下危险行为的隶属度,实现多模态场景人体危险行为的识别。实验结果表明:该方法对危险行为的识别准确率较高,其识别延迟时间低于300 ms。In multimodal scenes,conventional human dangerous behavior recognition methods generally perform low recognition accuracy.Therefore,this paper proposes a human dangerous behavior recognition method based on reinforcement learning.Firstly,a feature extraction algorithm of reinforcement learning is used to obtain feature subsets of human dangerous behavior in multimodal scenes.Secondly,human dangerous behaviors in multimodal scenes are extracted by reinforcement learning data decision-making,and a fuzzy recognition model of human dangerous behavior is constructed.Finally,by bringing the obtained feature subsets of human dangerous behavior into the model and calculating the membership degree of dangerous behavior under different senses,the recognition of human dangerous behavior in multimodal scenes can be realized.Experimental results show that the method in this paper has a high recognition accuracy and a recognition delay of less than 300 ms.

关 键 词：强化学习 多模态 场景 行为识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]