Safety Clothing Inspection System On Industrial Site

Safety is a responsibility,the correct wearing of safety helmet is a kind of protection and responsibility for themselves and others,in the construction site,it is difficult for us to avoid some potential or external safety hazards,such as: falling object impact and head touch other things.In recent years,based on the remote network real time video surveillance technology gradually rise,the real-time network video monitoring system used in industrial production field of monitoring staff whether consciously abide by the safety rules and regulations,to strengthen the industrial production field personnel safety supervision,thereby reducing head injuries caused by the impact of the scattered fallout of safety accident probability,At the same time,it is of great significance to industrial safety production.Based on the above situation,many researchers at home and abroad have designed and studied the safety clothing detection system.However,previous detection methods have problems such as single image occlusion,low accuracy of small target recognition and inability to adapt to complex scene environment,which make their detection effect of safety clothing in complex industrial scene environment poor.This paper proposes and designs an industrial field safety dress detection system,and in a complex application background,the system needs to meet the needs of remote control,image quality,real-time demand and accuracy demand.In order to meet the functional requirements and performance requirements of the system proposed above,the industrial site safety clothing testing system is composed of the following modules:Monitoring image acquisition module based on Nginx server,safety helmet detection module based on Harr cascade classifier,personnel detection and tracking module based on Py Torch platform,Web back-end data processing module based on Flask framework and Web front-end real-time display module based on Vue.In the monitoring image acquisition module based on Nginx server,the system uses Nginx server as the streaming media server,and uses FFmpeg video streaming processing tool to collect and process the real-time monitoring image of the industrial scene.In the safety helmet detection module based on Harr cascading classifier,Harr cascading classifier is used to recognize the safety helmet in video image frames.In the personnel detection and tracking module based on Py Torch platform,the multi-object tracking(MOT)model based on target detection and apparent feature sharing structure is used,which makes the industrial site safety clothing detection system significantly improved in terms of success rate and speed.At the same time,it can solve the problems of target occlusion and overlap.In the Web back-end data processing module based on Flask framework and the Web front-end real-time display module based on Vue,the Web framework Flask developed by Python is used to design and realize the processing and interaction of back-end data,and Vue is used as the front-end development framework to design the display interface,so as to display the detection results in real time.Finally,in order to evaluate the industrial site safety clothing detection system,this paper has carried out a large number of experimental tests from the aspects of the stability of monitoring image acquisition,real-time performance and accuracy of the system to prove that the system can meet the actual needs.