现代信息科技2025，Vol.9Issue(14)：32-36,5.DOI:10.19850/j.cnki.2096-4706.2025.14.007
基于ByteTrack的林下复杂环境中树干多目标跟踪研究
Research on Multi-target Trunk Tracking in Complex Understory Environments Based on ByteTrack
王耀宗 1黄建平1
作者信息
- 1. 东北林业大学 计算机与控制工程学院,黑龙江 哈尔滨 150040
- 折叠
摘要
Abstract
In the complex forest environment,the target tracking task faces severe challenges,especially in tasks such as trunk tracking,the phenomena of occlusion and dense targets are particularly prominent.Aiming at the problem of target loss and ID switches in the trunk tracking task caused by complex occlusion and dense tracking target environment in the forest,this paper improves the ByteTrack algorithm.In the matching process,a combination of appearance model and motion model is introduced,which effectively reduces the number of target ID switches.At the same time,the IoU measurement method which is more suitable for the occlusion environment is replaced.Furthermore,it improves the adaptability of the algorithm to occlusion and the tracking performance of dense targets in complex understory environments.关键词
目标跟踪/算法优化/深度学习/外观模型匹配Key words
target tracking/algorithm optimization/Deep Learning/appearance model matching分类
信息技术与安全科学引用本文复制引用
王耀宗,黄建平..基于ByteTrack的林下复杂环境中树干多目标跟踪研究[J].现代信息科技,2025,9(14):32-36,5.
