High-Altitude Operation Safety Monitoring System Based On Image Recognition

The safety of high-altitude operations has always been a key concern in power,construction,chemical,railway,and high-altitude services industries.Although there are currently cameras for monitoring the on-site conditions of high-altitude operations,this method can only perform simple shooting and display of the screen,and the analysis of the screen content still requires the judgment of supervisors,which is a low level method in automation.In order to solve the problem of how to intelligently monitor high-altitude operations to protect the personal safety of workers,this article mainly illustrates image recognition algorithms to recognize video images.The recognition of video images is divided into two parts: target detection and human posture estimation.The target recognition of personnel and safety tooling equipment,as well as the working status of the staff and the posture recognition of the safety tooling equipment are discussed.For the target recognition part,the backbone network of the YOLOv5 target detection algorithm is rebuilt,Ghost Net and SENet are integrated into the network,and the performance of the YOLOv5 algorithm in terms of model volume and algorithm accuracy is optimized,and the safety monitoring system is implemented efficiently and accurately for the identification of the personnel and the equipment in the video.The Alpha Pose human body pose estimation algorithm is discussed,and the Alpha Pose human body pose estimation algorithm based on Tensor RT inference acceleration optimization is proposed to improve the calculation speed of the algorithm and meet the real-time requirements of the system.At the same time,based on the improved Alpha Pose algorithm,the coordinates of the human body joint points in the video are extracted,and the human skeleton vector is constructed through the coordinate data,and the angle of the corresponding joint and the relative position between the joints are calculated.Using these data,we can further design a discrimination method for climbing and using double hooks to achieve the purpose of posture recognition in safety monitoring.In this paper,combining the above two improved recognition algorithms,considering the actual situation of high-altitude operation safety monitoring,a high-altitude operation safety monitoring system is designed,which realizes the functions of real-time monitoring and alarm information management,protects the personal safety of high-altitude operations personnel,and improves the working efficiency of the supervisory staff.Finally,the relevant functional modules of the system are designed,and the algorithms and functions of the system are tested to verify the feasibility of the system.