| With the increasing competition in the ocean,it is necessary to effectively monitor the sea situation to maintain the safety of maritime traffic and ensure national interests.One important aspect of this monitoring is the ability to achieve real-time detection of targets on the sea.In this paper,based on this application scenario,we have conducted research on sea surface target detection solutions,and designed and implemented an intelligent sea surface target detection and terminal management system based on Internet technology,computer vision technology,digital stabilization algorithm,and unmanned terminal devices.This system aims to solve the problems of inflexible detection,high cost of use,inability to highlight target category features,and inaccurate recognition caused by wind and waves in existing solutions,and to provide new solutions for the field of sea surface target detection.The main research of this paper is as follows:(1)This article investigates the existing detection solutions and proposes a sea surface target detection system architecture based on the actual situation of sea detection scenarios.The system is divided into three parts: monitoring and management platform,service provider module,and detection task execution terminal.The system utilizes Yolo v5 object detection algorithm to achieve recognition and monitoring of sea surface targets through optical imaging.In order to reduce data processing latency,improve system response speed,and reduce labor costs,the system performs sea patrol detection tasks through edge deployment such as unmanned ships and AGX intelligent computing modules.To solve the problem of decreased recognition accuracy caused by ship sway due to sea waves,this article introduces the optical flow video stabilization technology.In order to adapt to the sea search scene of unmanned terminals,the system also uses the wireless self-organizing network module qca9531 to provide reliable network connection services.In order to facilitate users to achieve unified monitoring and management,this article develops a corresponding operating interface and service program using internet technology,making it easy for users to quickly complete sea surface target detection tasks in a user-friendly manner.(2)The paper conceptualizes and produces the details of each module,including analysis of the system’s overall requirements,mastery and combination of various technology features,module function division,system workflow design,interface design of each module,and specific content implementation of each module.This article constructs the entire system,with the front-end operation page written in Vue technology,allowing users to log in and use the system through a web browser.The back-end is written in the Spring Boot framework,which not only provides service support to the entire system through microservices,but also manages and controls the detection task execution terminal through HTTP protocol requests.Additionally,the system also provides functions such as gateways,streaming services,performance monitoring,data persistence,and external interfaces.(3)This paper verifies the functional integrity and performance of this system by means of a two-part test,including theoretical performance experiments and actual scenario tests.The results show that the detection accuracy of this paper reached 90.2% of m AP,and the detection rate and accuracy were improved compared with those before the anti-shake treatment,and the detection speed reached 45 FPS.According to the experimental data,this intelligent sea surface target detection system has good recognition performance and usability,and can meet the needs of actual detection scenarios... |
