基于改进YOLOv10n的电动车头盔佩戴检测算法  

Electric bike helmet wearing detection algorithm based on improved YOLOv10n

作　　者：周翔 王可庆 周新翔 韩基泰 Zhou Xiang;Wang Keqing;Zhou Xinxiang;Han Jitai(College of Automation,Nanjing University of Information Science&Technology,Nanjing 210044,China;College of Automation,Wuxi University,Wuxi 214105,China)

机构地区：[1]南京信息工程大学自动化学院,南京210044 [2]无锡学院自动化学院,无锡214105

出　　处：《电子测量技术》2025年第5期40-49,共10页Electronic Measurement Technology

基　　金：“太湖之光”科技攻关(基础研究)(K20221051)项目资助。

摘　　要：针对电动车头盔佩戴检测方法存在着复杂路况下头盔小目标检测精度较低、目标相互遮挡漏检率较高、检测模型大运算复杂等问题。本研究提出一种基于改进YOLOv10n的目标检测算法,以解决在实际应用中的这些问题。首先,在MAFPN的基础上融合了BiFPN的优点,创新性地提出了BIMAFPN结构,提高了模型在复杂路况场景下的检测性能。其次,构建Inner-Wise-MPDIoU损失函数以替代传统的CIoU损失函数,在提高网络的检测精度的同时,还加速了模型的收敛过程。最后,引入LSCD检测头进一步减少模型参数量并提升检测性能。实验结果表明,改进模型相比于原模型在mAP@0.5精度上提升了2.7%,同时参数量降低了25%,模型大小减少了35%。本研究使用的检测方法不仅显著提高了复杂路况下的头盔检测精度,同时在兼顾轻量化的基础上保持了良好的实时性,便于将模型部署于小型嵌入式交通设备中。For the electric vehicle helmet wearing detection method exists in complex road conditions helmet small target detection accuracy is low,the target mutual occlusion leakage rate is high,detection model large arithmetic complexity and other problems.This question proposes a target detection algorithm based on improved YOLOv10n to solve these problems in practical applications.Firstly,the advantages of BiFPN are integrated on the basis of MAFPN,and the BIMAFPN structure is innovatively proposed,which improves the detection performance of the model in complex road scenarios.Secondly,the Inner-Wise-MPDIoU loss function is constructed to replace the traditional CIoU loss function,which improves the detection accuracy of the network while accelerating the convergence process of the model.Finally,the LSCD detection head is introduced to further reduce the number of model parameters and improve the detection performance.Experimental results show that compared with the original model,the improved model improves the accuracy of mAP@0.5 by 2.7%,the number of parameters is reduced by 25%,and the model size is reduced by 35%.The detection method used in this paper not only significantly improves the accuracy of helmet detection under complex road conditions,but also maintains good real-time performance while taking into account lightweight,which makes it easy to deploy the model in small embedded transport devices.

关 键 词：YOLOv10n 头盔佩戴检测 BIMAFPN 损失函数 LSCD 模型轻量化 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TN06[自动化与计算机技术—计算机科学与技术]