Design of Joint Control System for Target Detection UAV and Ground Vehicle Based on YOLO Algorithm
Weijie Li, Weicheng Huang, Yin-Hua Li, Li Wang · 2023
Nowadays, there is an increasing need for faster collection and processing of environmental information in technology, battlefield, and urban construction. In complex environments where human activity is difficult, drones can be dispatched to search for targets. Whether in military or civilian applications, the research on detection and positioning algorithms for search targets is crucial in the process of drone search. This also requires us to continuously improve the drone's ability to detect and locate targets. The breakthrough developments in computer vision and deep learning have greatly promoted the application of visual image technology in intelligent target recognition, localization, tracking, and target trajectory prediction. Due to its flexible flight characteristics, unmanned aerial vehicles (UAVs) are capable of flying in various complex environments. Therefore, the combination of drones and machine vision is increasingly sought after by people. Therefore, the research on unmanned aerial vehicle target detection and joint control technology based on YOLO algorithm has practical requirements and significance. Based on the characteristics of mainstream object detection algorithms in image technology and their performance comparison, a suitable object detection algorithm was selected for the scenario in this article. Based on the analysis of the trained results and the retested detection results, the YOLO detection algorithm that is most suitable for this article was selected. Then, the prior boxes of the algorithm were re clustered, and the clustered algorithm was re trained and tested to improve detection accuracy. The method used to improve detection accuracy is to analyze the feature layer information of each layer again, improve the structure of the clustered detection algorithm, and retrain and test the algorithm. Finally, drones provide more application scenarios for the YOLO algorithm, which can help drones complete more novel tasks. Through this approach, drone technology and YOLO algorithms can further promote people's daily lives while contributing to the productivity of their respective industries. The experiment shows that the design of the target detection unmanned aerial vehicle and ground vehicle joint control system based on YOLO algorithm in this paper is feasible, achieving the expected results, and has the advantages of stability and simple operation.