Research On Pose Estimation And Behavior Recognition Based On Deep Learning In Logistics Warehousing

With the rapid development of the Internet,e-commerce has become an emerging form of economic development,and e-commerce operations are inseparable from supporting the logistics industry.Logistics enterprises realize informatization and intelligence through modern communication engineering technologies such as network sensing technology,information positioning technology,and video intelligent monitoring technology,making full use of information flow to dispatch and control logistics,and efficiently allocate logistics service resources.In order to ensure the core competitiveness of monitoring and local logistics,efficient and safe transfer of materials and products,logistics and warehousing safety are important aspects,which are also important links in the logistics chain.The content of this article is to apply deep learning human behavior recognition to logistics and warehousing safety management.The feature extraction and behavior recognition of people in the video are helpful to discover abnormal behaviors in time and take effective solutions.The current video surveillance system is used in various public places with a lot of people,but it usually does not have the function of behavior recognition.Intelligent video surveillance can be used not only to capture and record scenes,but also to identify human behavior and detect abnormal behavior in time.Relying only on the monitoring personnel in front of the screen to find abnormal information in the video is not only inefficient and accurate.Monitoring personnel watching the video for a long time will cause excessive fatigue,and false alarms and false alarms may occur.Video information extraction and analysis are based on deep learning algorithms,so that posture estimation and behavior recognition of people in video images can be performed.Behavior recognition through video can detect abnormal behaviors in time,and provide early warning when unexpected situations are discovered,and staff can detect and deal with them in time,which can effectively avoid accidental losses.This paper mainly uses OpenPose to study human pose estimation,and then uses YOLO's spatiotemporal graph convolutional network to recognize human behavior.Finally,an intelligent monitoring system suitable for logistics warehouses is designed,which can improve the accuracy of identifying abnormal behaviors in logistics warehouses and provide timely warnings,and improve the safety level of logistics warehouse management.Figure 32 table4 reference 67...