针对低空微小型无人机的轻量型YOLOv5检测算法  

Lightweight YOLOv5 detection algorithm for low-altitude micro UAV

作　　者：魏峰[1] 周建平[1] 谭翔 林静[2] 田莉 王虎[1] WEI Feng;ZHOU Jianping;TAN Xiang;LIN Jing;TIAN Li;WANG Hu(School of Intelligent Manufacturing Modern Industry,Xinjiang University,Urumchi,Xinjiang Uygur Autonomous Region 830000,China;Institute of Geographic Sciences and Natural Resources Research,Chinese Academy of Sciences,Beijing 100101,China;The Research Center for UAV Application and Regulation,Chinese Academy of Sciences,Beijing 100101,China)

机构地区：[1]新疆大学智能制造现代产业学院,新疆维吾尔自治区乌鲁木齐830000 [2]中国科学院地理科学与资源研究所,北京100101 [3]中国科学院无人机应用与管控研究中心,北京100101

出　　处：《光电子．激光》2024年第6期641-649,共9页Journal of Optoelectronics·Laser

基　　金：黑土地保护与利用科技创新工程专项资助(XDA28060400);中科吉安生态环境研究院院长基金(ZJIEES-2020-026)资助项目。

摘　　要：针对低空微小型无人机对公共安全造成威胁的问题,本文基于YOLOv5(you only look once v5)网络提出了一种适用于移动端的轻量型目标检测模型YOLOv5_SS。该模型以轻量型网络ShuffleNetv2替换YOLOv5原有的主干网络,引入SENet(squeeze-and-excitation networks)注意力机制,并采用Soft-NMS(soft non-maximum suppression)算法提升对密集重叠目标的检测效果。实验结果表明,该模型在数据集上对低空微小无人机进行检测的平均精确率均值(mean average precision@0.5,mAP_(50))为92.75%,精度为90.49%,参数量为0.2374 M,浮点运算数为0.9千兆浮点运算(giga floating-point operations,GFLOPS)。具有检测精度高、内存占用率低的特点,有利于在移动终端上部署且在复杂背景及密集目标的场景下均有较好的检测效果。Aiming at the problem that low-altitude micro-UAVs pose a threat to public safety,this paper proposes a lightweight target detection model YOLOv5_SS suitable for mobile terminals based on the you only look once v5(YOLOv5)network.In this model,the lightweight network ShuffleNetv2 replaces the original backbone network of YOLOv5,introduces squeeze-and-excitation networks(SENet)attention mechanism,and uses soft non-maximum suppression(Soft-NMS)algorithm to improve the detection effect of dense overlapping targets.The experimental results show that the mean average precision@0.5(mAP_(50))of the model for the detection of low-altitude micro-UAV on the dataset is 92.75%,the accuracy is 90.49%,and the number of parameters is 0.2374 M.The number of floating-point operations is 0.9GFLOPS(giga floating-point operations).

关 键 词：无人机检测 深度学习 轻量型网络 注意力机制 非极大值抑制(NMS) 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]