Design Of Riverbank Waterfront Personnel Safety Inspection System Based On UAV

The traditional Yellow River inspection mostly adopts the foot way to patrol the riverbank and the waterfront personnel.This method has the problems of low efficiency and poor flexibility,and can not warn the waterfront personnel near the riverbank in time.Unmanned Aerial Vehicle(UAV)has the advantages of saving manpower,high efficiency and flexibility.In order to find the riverbank waterfront personnel in time and shout warnings to reduce the risk of drowning,this paper designs a set of safety inspection system for riverbank waterfront personnel based on UAV to realize the automatic inspection of UAV,and focuses on the design of the safety detection algorithm for waterfront personnel and the inspection system software platform,which comprehensively improves the efficiency of riverbank inspection.The main research contents of this paper are as follows:(1)Based on the analysis of the requirements of riverbank inspection,the overall design of the inspection system is carried out from the aspects of hardware platform,waterfront personnel safety detection algorithm and software platform.According to the work flow of the inspection system,it is necessary to control the UAV to inspect along the bank of the Yellow River.The software platform displays the inspection video screen in real time and performs personnel safety detection.After detecting the dangerous waterfront personnel,the UAV is used for remote call warning and the specific location is displayed on the map.(2)Design of safety detection algorithm for waterfront personnel.Firstly,the River-Person data set is built by using the UAV riverbank inspection video.Then,the real-time detection of people in the video is carried out,and the YOLOv7 target detection algorithm is improved by using GhostNet network,Normalization-based Attention Module(NAM)and SIoU loss function.The accuracy is balanced and the real-time performance of the algorithm is effectively improved.The final algorithm runs at 64 frames per second(FPS),which is 12.28%higher than the original YOLOv7 algorithm.Secondly,the real-time segmentation of the riverbank scene is carried out and the dangerous water area is obtained.The Coordinate Attention(CA)is added while reducing the complexity of the DeepLabV3+model,which improves the training efficiency of the model.The algorithm takes 0.08s on average.Finally,the pixel conversion method is used to expand the dangerous water area and delineate the dangerous waterfront area.The safety detection of personnel is completed by judging whether the personnel are in the waterfront area.(3)The corresponding inspection system software platform is designed for the riverbank inspection application.As the carrier of the safety detection algorithm for waterfront personnel,the platform adopts B/S(Browser/Server)architecture and builds a software platform based on SpringMVC.According to the modular development idea,the map module is based on the Cesium framework to realize the characteristic functions of coordinate system and camera angle control,riverbank protection engineering terrain view and UAV route planning.The audio and video module realizes the video real-time display and remote call scheme.The management module realizes the management of inspection tasks,system equipment and user rights.(4)Build inspection system platform and system verification.The UAV field flight function is stable,the safety detection function of the waterfront personnel is normal,and the functions of each part of the software platform are perfect,which achieves the design purpose of the inspection system.Based on the background of the Yellow River bank inspection,this paper designs a safety inspection system for riverbank waterfront personnel based on UAV.After system verification and application in actual scenarios,it provides a good foundation and development direction for the application of the Yellow River bank inspection.