Analysis Of Working Status Of Construction Workers Based On Computer Vision

At present,the construction industry is facing severe market environment and development challenges.There are many problems in the construction industry,such as extensive development mode,low skill quality of construction workers,imperfect supervision system and backward technical equipment.Computer vision technology is generally regarded as a powerful tool to promote a new round of intelligent revolution in the construction industry.Using computer vision technology to realize the automated management of on-site workers can effectively improve the related problems.This paper focuses on the automatic acquisition and analysis of the working status information of construction workers,and proposes a method of analyzing the working status of workers based on computer vision.The proposed working status analysis method includes four parts: analysis of labor resources,worker activity scope analysis,workers’ safety status warning,and \labor time efficiency calculation.In labor resources analysis,gaussian mixture model and support vector machine were used to detect construction workers from moving objects in video scene,and based on the characteristics of acceleration robustness,the fusion statistics of the number of workers from multiple perspectives were conducted.In the workers’ activity scope analysis process,the camera three-dimensional measurement method is used to obtain the location information of construction workers detected in the analysis of labor resources process,and according to the distribution density characteristics of workers’ activity location.Based on the distribution density of workers’ activity location,the unsafe level of the construction site environment and the unsafe area are inferred dynamically.In the aspect of workers’ safety warning,according to the worker’s position information obtained from the dynamic statistics of workload,Kalman filter tracking method is used to predict the worker’s trajectory,and then the worker’s safety status is predicted by combining the prior information of dangerous areas.In the aspect of working efficiency measurement,the visual measurement method of construction workers’ working time is constructed by the method of human body posture space vector modeling and Dynamic Time Warping algorithm,and then the average construction efficiency of construction workers is analyzed by combining the information of engineering quantity obtained from the dynamic statistics of workload.In this paper,field experiments are respectively carried out on the four parts mentioned above.The results show that the accuracy rate of labor resource analysis method is 90.6%,the maximum error of workers’ position observation in the analysis of workers’ activity range is 153.1cm;the accuracy rate of workers’ safety status is 87.6%,and the recognition accuracy rate of construction workers’ action recognition algorithm is 84.1% in the working efficiency measurement.The experimental results show that this method can automatically collect,statistics and analyze the working status information of construction workers,and provide objective quantitative data for personnel management decision-making and analysis of field managers,so as to reduce personnel management costs and improve construction management efficiency.