用于多尺度道路目标检测的优化定位置信度改进算法  

Improved Algorithm of Optimized Localization Confidence for Multi-scale Road Object Detection

作　　者：刘悦 张璐 罗文广[3] 叶洪涛[3] 石英[1] 林朝俊 LIU Yue;ZHANG Lu;LUO Wen-guang;YE Hong-tao;SHI Ying;LIN Chao-jun(School of Automation,Wuhan University of Technology,Wuhan 430070,China;Wuhan NARI Limited Liability Company of State Grid Electric Power Research Institute,Wuhan 430070,China;Guangxi Key Laboratory of Automobile Components and Vehicle Technology,Guangxi University of Science and Technology,Liuzhou 545006,China)

机构地区：[1]武汉理工大学自动化学院,武汉430070 [2]国网电力科学研究院武汉南瑞有限责任公司,武汉430070 [3]广西汽车零部件与整车技术重点实验室(广西科技大学),广西柳州545006

出　　处：《小型微型计算机系统》2023年第9期2030-2037,共8页Journal of Chinese Computer Systems

基　　金：国家重点研发计划项目(2020YFB1506802)资助;广西汽车部件与整车技术重点实验室开放研究项目(2020GKLACVTKF03)资助.

摘　　要：为了提高多尺度道路目标的检测性能,本文针对目标检测算法在非极大值抑制阶段的检测质量表征不合理问题,提出了一种优化定位置信度改进算法.首先基于RepPoints构建研究框架,研究定位置信度对多尺度道路目标的敏感性.在敏感性研究结果的基础上,本文提出了混合定位置信度.然后针对IoU定位置信度无法区分重叠程度相同的包围框的缺陷,提出了CIoU定位置信度.最后将这两种定位置信度结合得到优化定位置信度改进算法,解决了检测质量表征不合理问题.在道路场景数据集Cityscapes上的实验结果表明,混合定位置信度和CIoU定位置信度单一作用时均有效,共同作用时精度提高2.4%,多尺度目标检测精度均有显著提升,且实时性没有下降.相较于主流道路场景检测算法如Cascade-RCNN、FCOS等,本文算法取得了最高的mAP、AP M和AP L.In order to improve the detection performance of multi-scale road objects,this paper proposes an improved algorithm of optimized localization confidence to solve the problem that the detection quality is represented unreasonably in NMS of detection algorithms.Firstly,a research framework is constructed based on RepPoints to study the sensitivity of localization confidence to multi-scale road targets.A mixed localization confidence was proposed according to the research results.Then,CIoU localization confidence is proposed to solve the problem that IoU cannot distinguish bounding boxes with the same overlapping degree.Finally,an improved algorithm of optimized localization confidence is obtained by combining these two confidence,which solves the problem of unreasonable representation of detection quality.The experimental results on Cityscapes show that both the mixed localization confidence and CIoU localization confidence are effective when used separately,and the accuracy is improved by 2.4%when used together.The detection accuracy of multi-scale objects is significantly improved,and the real-time performance is not decreased.Compared with the mainstream detection algorithms for road scene such as Cascade-RCNN,FCOS,etc,the proposed algorithm achieves the highest mAP,AP M and AP L.

关 键 词：目标检测 多尺度道路目标 定位置信度 RepPoints 非极大值抑制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]