基于YOLO目标检测算法的人群多目标识别跟踪方法  

A crowd multiple target recognition and tracking method based on YOLO object detection algorithm

作　　者：张四平[1] ZHANG Siping(School of Economics and Management,Hunan College of Information,Changsha 410200,China)

机构地区：[1]湖南信息职业技术学院经济管理学院,长沙410200

出　　处：《智能计算机与应用》2024年第1期152-155,共4页Intelligent Computer and Applications

基　　金：湖南省教育厅科研项目(22C1171)。

摘　　要：目前对人群跟踪方法主要是建立跟踪模型,实现人群群体检查和跟踪,但是无法实现人群中多行人的识别和个体跟踪,造成人群跟踪与多目标识别存在效率低下和不准确。本文提出基于YOLO目标检测算法的人群多目标识别跟踪方法,通过对人群多目标的可见特征提取人群的行人轨迹和外观特征,实现人群多目标识别的跟踪。实验结果表明,该方法提高了人群多目标的识别效率,具有一定的实用性。The current tracking method for crowd mainly establishes a tracking model to realize crowd group checking and tracking,but it cannot realize the identification of multiple pedestrians in the crowd and individual tracking,resulting in inefficiency and inaccuracy of crowd tracking and multi-target recognition.This paper proposes a crowd multi-target recognition tracking method based on YOLO target detection algorithm,which realizes the tracking of crowd multi-target recognition by extracting the pedestrian trajectory and appearance features of the crowd through the visible features of the crowd multi-target.The experimental results show that the method improves the efficiency of crowd multi-target recognition and has certain practicality.

关 键 词：YOLO目标检测算法 人群多目标跟踪 识别方法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]