Anomaly Detection And System Implementation Of Logistics Pipeline Transportation Node Area Under Video Monitoring

In recent years,the level of urbanization in China has been increasing,and the logistics pipeline transportation system has been developing rapidly.As a key link,the stable operation of logistics pipeline transportation nodes is not only related to the pipeline transportation system but also affects the safety and stability of cities and towns.With the increasing phenomenon of human interference and destruction of pipeline transportation nodes,their safety issues have become the focus of social concern.Video surveillance as a common means of protection,logistics pipeline transport node area video surveillance system generally uses personnel guarded approach,only part of the realization of intelligent monitoring and detectable categories are relatively single.For the above problems,this paper proposes and builds an abnormal detection system for logistics pipeline transportation node area,which can detect the category and location of personnel in the node area and realize the classification and early warning of abnormal behavior of personnel.The work is as follows.(1)A person recognition and classification model based on YOLOv4Tiny is constructed to realize the detection of persons in the area of logistics pipeline transportation nodes and to determine the specific types of persons based on clothing features.The model adds large-size feature layers and adjusts the number and size of anchors by K-means++clustering algorithm to enhance the detection level of the model for small targets.The attention mechanism is also introduced for different scale feature fusion to enhance the model’s ability to distinguish person category features.Experiments show that the improved model improves the mAP by 3.1%compared with the original model,and the speed indicator FPS reached 32.74,which meets the application requirements in terms of speed and accuracy.(2)A framework and solution for detecting abnormal behaviors of personnel in the logistics pipeline transportation node area are proposed,and four typical abnormal behaviors are defined for the logistics pipeline transportation node area,including key area intrusion,prolonged contact with facilities,wandering and foreign object carrying behaviors,and the detection of the four abnormal behaviors is realized.The experiments show that the proposed solution for detecting abnormal behavior of personnel in the logistics pipeline transportation node area meets the task requirements and effectively guarantees the safety of the logistics pipeline transportation node.(3)Finally,the abnormality detection system of logistics pipeline transportation node area under video monitoring is built,and the above research is engineered through research and analysis of the actual application environment,and the system realizes abnormal detection and classification warning of people entering the logistics pipeline transportation node area,which makes the research of this paper have application value and practical significance.