Research And Design Of Personnel Monitoring System In Complex Industrial Environment

In the complex industrial production environment,the safety of operators is very important.Most industrial enterprises use the way of human eye observation video to judge the position of operators.This way will make managers appear fatigue problems for a long time,and the efficiency is very low,and the safety problems can not be handled in time.Therefore,this paper uses image processing and deep learning technology to make the computer replace the human eye to detect and locate personnel,which can not only improve the accuracy of detection,but also reduce the workload of supervisors.In this paper,the main work of this paper is firstly focused on the types of difficult to detect images,image acquisition in industrial environment and the number of people in a variety of postures and states.At the same time,the personnel in the data set are annotated to form the corresponding format data set for algorithm training.Secondly,for the problem of difficult detection of personnel in the complex industrial environment,the HOG + SVM personnel detection algorithm is used Faster?RCNN personnel detection algorithm and centernet personnel detection algorithm are trained,tested and analyzed in this paper,and the conclusion is drawn that the detection accuracy and speed of centernet personnel detection algorithm are better than other algorithms;finally,for the problem that large area of personnel occlusion cannot be detected in the practical application of centernet personnel detection algorithm,an improvement based on background difference method is proposed The centrernet personnel detection algorithm uses the background subtraction method to quickly determine the approximate position of personnel,adjusts the size of the centrernet score threshold,so that the large area occlusion personnel with less score can be detected correctly,and improve the average detection accuracy of centernet personnel detection algorithm in application.In this paper,the system is built and tested in the industrial field environment.Using Lab VIEW and python programming language,the improved centernet personnel detection algorithm based on background difference is applied to the system.At the same time,the personnel detection is realized,combined with the safety hazard plan of the work area and camera calibration to determine whether the personnel have cross boundary behavior.In the final system test,the average accuracy and speed of personnel detection and positioning can meet the requirements of field system performance,which can meet the needs of actual production environment.