一种结合YOLOv8和DeepOCSORT的排球比赛球员自动检测与追踪算法  

An Algorithm Combining YOLOv8 and DeepOCSORT for Automatic Detection and Tracking of Players in Volleyball Games

作　　者：尹邦政 彭泽林 朱静[3] 唐耿 YIN Bangzheng;PENG Zelin;ZHU Jing;TANG Geng(School of Information Engineering,Guangzhou Railway Polytechnic,Guangzhou 511300,China;School of Mathematics and Information Sciences,Guangzhou University,Guangzhou 510420,China;Experimental Centre,Guangzhou University,Guangzhou 510420,China)

机构地区：[1]广州铁路职业技术学院信息工程学院,广东广州511300 [2]广州大学数学与信息科学学院,广东广州510420 [3]广州大学实验中心,广东广州510420

出　　处：《东莞理工学院学报》2024年第5期26-32,共7页Journal of Dongguan University of Technology

基　　金：广州市科技局2022年基础研究项目(202201011668)。

摘　　要：在体育视频中的战术分析中,运动员的检测和跟踪至关重要,球员检测和跟踪已成为关键研究领域。目前,目标检测和跟踪算法在排球视频分析中存在着球员快速移动、同一队球员聚集以及目标重叠等问题,导致检测和跟踪错误。本文提出一种结合YOLOv8和DeepSORT的改进算法,并结合时空注意力机制模块,能够自动准确地检测和跟踪运动员,进行运动轨迹和数据分析。实验结果显示,该算法在球员检测和追踪方面表现出良好的跟踪精度和目标检测准确性。The detection and tracking of volleyball players are crucial for tactical analysis in sports videos.Player detection and tracking have become key research areas.The current object detection and tracking algorithms in volleyball video analysis suffer from problems such as rapid player movement,clustering of players on the same team,and overlapping targets,leading to detection and tracking errors.This article proposes an improved algorithm combining YOLOv8 and DeepSORT,combined with a spatiotemporal attention mechanism module,which can automatically and accurately detect and track athletes for motion trajectory and data analysis.The experimental results show that the algorithm exhibits good tracking accuracy and object detection accuracy in player detection and tracking.

关 键 词：目标检测 目标追踪 注意力机制 YOLOv8 DeepOCSORT 

分 类 号：TP918.1[自动化与计算机技术]