机电工程技术2024，Vol.53Issue(11)：143-147,5.DOI:10.3969/j.issn.1009-9492.2024.11.031
基于YOLOv5的微小飞行器自主识别及云-边-端视觉跟踪监控系统
Autonomous Recognition and Cloud Edge End Vision Tracking and Monitoring System for Micro Aircraft Based on YOLOv5
摘要
Abstract
As drones increasingly permeate,the issue of unauthorized flights emerges as a critical concern for community security.A visual identification system utilizing a cloud-edge-node framework to address the monitoring hurdles of illicit drone operations,such as target-background differentiation,swift and precise tracing,recognition amidst complex scenarios,and attribute extraction.The mechanism harnesses the YOLO schema's deep learning methodologies for drone detection and fine-tunes the YOLOv5 structure,enhancing its fidelity and efficiency in drone discernment based on a lightweight ethos.The experimental arrangement comprises a gimbal equipped with a camera,managed by a Raspberry Pi as the end client,and a server established in the cloud,interconnected via the MQTT protocol.The findings indicate that the visual identification system,capitalizing on the advanced YOLOv5 algorithm,can recognize drones with an average accuracy of 96%in real-time within intricate environments,precisely monitor their movement trajectories,and display commendable resilience to varying lighting intensities.关键词
视觉识别/无人机安防/深度学习/轻量化Key words
visual recognition/unmanned aerial vehicle(UAV)security/deep learning/lightweight分类
信息技术与安全科学引用本文复制引用
李艺良,宋海鹰,黄彦嶂,谢卓凯,颜依颖,陈填韬,吴淇森,吴江南,陆淳鑫..基于YOLOv5的微小飞行器自主识别及云-边-端视觉跟踪监控系统[J].机电工程技术,2024,53(11):143-147,5.基金项目
国家大学生创新训练项目(202310588033) （202310588033）
国家大学生创新训练项目(202210588028) （202210588028）
