基于改进YOLOv3的运动目标分类检测算法研究  

Moving Target Classification and Detection Algorithm Based on Improved YOLOv3

作　　者：梁秦嘉 刘怀[1] 陆飞 Liang Qinjia;Liu Huai;Lu Fei(School of Electrical and Automation Engineering,Nanjing Normal University,Nanjing 210023,China)

机构地区：[1]南京师范大学电气与自动化工程学院,江苏南京210023

出　　处：《南京师范大学学报（工程技术版）》2021年第4期27-32,共6页Journal of Nanjing Normal University(Engineering and Technology Edition)

基　　金：国家自然科学基金项目(61603194).

摘　　要：提出一种基于改进YOLOv3算法的一类运动目标检测算法.为进一步提高YOLOv3的检测精度,采用基于DIoU优化的边界框回归损失函数进行计算;优化非极大值抑制,有效减少了目标框重叠的现象,提高检测精度;针对运动目标检测,提出一种基于目标框多中心点位移的检测算法.经UA-DETRAC数据集上的实验表明,改进后的算法在提高检测精度的同时保证了较快的速度,准确率和召回率相比原始YOLOv3分别提高了8.07%和3.87%,对运动目标的检测速度可达20 fps/s,可满足实时检测的要求.A kind of moving target detection algorithm based on improved YOLOv3 algorithm is proposed in this paper.In order to improve the detection accuracy of YOLOv3,the boundary box regression loss function based on DIoU optimization is used.Non-maximum suppression is optimized to effectively reduce the overlap of target boxes and improve the detection accuracy.Aiming at moving target detection,a multi-center displacement detection algorithm based on target frame is proposed.The experimental results on UA-DETRAC dataset show that the detection accuracy and the fast speed can be improved by the improved algorithm.Compared with the original YOLOv3,the accuracy and recall rate are increased by 8.07%and 3.87%,respectively.The detection speed of moving target can reach 20 fps/s,which can meet the requirements of real-time detection.

关 键 词：交通监控 卷积神经网络 运动目标检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]