基于YOLOv5改进算法的足球检测识别研究  

Research on Football Detection and Recognition based on improved YOLOv5 Algorithm

作　　者：张诗欣 李鸿强[1] 李子熠 张天宇 ZHANG Shixin;LI Hongqiang;LI Ziyi;ZHANG Tianyu(Hebei University of Architecture,Zhangjiakou,Hebei 075000)

机构地区：[1]河北建筑工程学院,河北张家口075000

出　　处：《河北建筑工程学院学报》2024年第4期235-240,共6页Journal of Hebei Institute of Architecture and Civil Engineering

摘　　要：足球比赛场景中球员居多、足球目标偏小且移动速度快,足球检测识别难度很大。为了解决这一问题,提出一种基于改进的YOLOv5的足球检测方法,增加使用了OTA(Optimal Transport Assignment)损失函数来优化模型提高对足球目标的识别精度,最后在Roboflow的足球数据集上进行训练,对足球比赛场景下的足球进行目标检测实现足球识别。根据实验可以得出结论:改进后的YOLOv5算法的足球识别不仅提高了足球的识别性能与精度,而且有效地提高了检测速度,具有更好的识别性能。In the soccer game scene,there are most players,the soccer target is small and moves fast.In order to solve this problem,this paper proposes a soccer detection method based on improved YOLOv5,and adds OTA(Optimal Transport Assignment)loss function to optimize the model to improve the recognition accuracy of the soccer target.Finally,it trained on the soccer data set of Roboflow,and detected the soccer object in the soccer game scene to realize soccer recognition.According to the experiments,it can be concluded that the improved YOLOv5 algorithm for football recognition not only improves the recognition performance and accuracy of football,but also effectively improves the detection speed.It has better recognition performance.

关 键 词：YOLOv5 足球检测 Optimal Transport Assignment 损失函数 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]