Enhanced DeepSORT and StrongSORT for Multicattle Tracking With Optimized Detection and Re-Identification

IF 3.4 3区计算机科学 Q2 COMPUTER SCIENCE, INFORMATION SYSTEMS

IEEE Access Pub Date : 2025-01-27 DOI:10.1109/ACCESS.2025.3535092

Hyeon-Seok Sim;Hyun-Chong Cho

{"title":"Enhanced DeepSORT and StrongSORT for Multicattle Tracking With Optimized Detection and Re-Identification","authors":"Hyeon-Seok Sim;Hyun-Chong Cho","doi":"10.1109/ACCESS.2025.3535092","DOIUrl":null,"url":null,"abstract":"Recently, labor shortages in the farming industry have increased the demand for automation. Object tracking technology has emerged as a critical tool for monitoring livestock through automated systems. This study focuses on tracking individual cattle using object detection and tracking algorithms. Data were collected noninvasively using cameras, and a tracking-by-detection (TBD) approach was adopted. The proposed framework introduces multiple enhancements optimized for cattle tracking. These enhancements include a comparison of five different bounding box regression losses to improve detection accuracy, modifications to the Kalman filter state vector for more accurate bounding box predictions, and adjustments to the feature vector distance metric in the re-identification algorithm. YOLOv9-t was used as the detector, whereas DeepSORT and StrongSORT served as trackers. Compared with the baseline, which uses DeepSORT, the proposed method achieved significant improvements in higher-order tracking accuracy (HOTA) by 4.1%, multiple object tracking accuracy (MOTA) by 1.08%, and identification F1 score (IDF1) by 5.12%, reaching values of 78.64%, 90.29%, and 91.41%, respectively, while reducing the number of ID switches (IDSW).","PeriodicalId":13079,"journal":{"name":"IEEE Access","volume":"13 ","pages":"19353-19364"},"PeriodicalIF":3.4000,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10855392","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"IEEE Access","FirstCategoryId":"94","ListUrlMain":"https://ieeexplore.ieee.org/document/10855392/","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}

引用次数: 0

Abstract

Recently, labor shortages in the farming industry have increased the demand for automation. Object tracking technology has emerged as a critical tool for monitoring livestock through automated systems. This study focuses on tracking individual cattle using object detection and tracking algorithms. Data were collected noninvasively using cameras, and a tracking-by-detection (TBD) approach was adopted. The proposed framework introduces multiple enhancements optimized for cattle tracking. These enhancements include a comparison of five different bounding box regression losses to improve detection accuracy, modifications to the Kalman filter state vector for more accurate bounding box predictions, and adjustments to the feature vector distance metric in the re-identification algorithm. YOLOv9-t was used as the detector, whereas DeepSORT and StrongSORT served as trackers. Compared with the baseline, which uses DeepSORT, the proposed method achieved significant improvements in higher-order tracking accuracy (HOTA) by 4.1%, multiple object tracking accuracy (MOTA) by 1.08%, and identification F1 score (IDF1) by 5.12%, reaching values of 78.64%, 90.29%, and 91.41%, respectively, while reducing the number of ID switches (IDSW).

查看原文本刊更多论文

近来，由于养殖业劳动力短缺，对自动化的需求不断增加。物体跟踪技术已成为通过自动化系统监控牲畜的重要工具。本研究的重点是利用物体检测和跟踪算法跟踪牛群个体。使用摄像头以非侵入式方式收集数据，并采用检测跟踪 (TBD) 方法。所提出的框架引入了多种针对牛跟踪进行优化的增强功能。这些改进包括：比较五种不同的边界框回归损失以提高检测准确性；修改卡尔曼滤波器状态向量以获得更准确的边界框预测；以及调整重新识别算法中的特征向量距离度量。YOLOv9-t 被用作检测器，而 DeepSORT 和 StrongSORT 被用作跟踪器。与使用 DeepSORT 的基线相比，所提出的方法在高阶跟踪精度（HOTA）、多目标跟踪精度（MOTA）和识别 F1 分数（IDF1）方面均有显著提高，分别提高了 4.1%、1.08% 和 5.12%，达到 78.64%、90.29% 和 91.41%，同时减少了 ID 切换次数（IDSW）。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

IEEE Access COMPUTER SCIENCE, INFORMATION SYSTEMSENGIN-ENGINEERING, ELECTRICAL & ELECTRONIC

CiteScore

9.80

自引率

7.70%

发文量

6673

审稿时长

6 weeks

期刊介绍： IEEE Access® is a multidisciplinary, open access (OA), applications-oriented, all-electronic archival journal that continuously presents the results of original research or development across all of IEEE''s fields of interest. IEEE Access will publish articles that are of high interest to readers, original, technically correct, and clearly presented. Supported by author publication charges (APC), its hallmarks are a rapid peer review and publication process with open access to all readers. Unlike IEEE''s traditional Transactions or Journals, reviews are "binary", in that reviewers will either Accept or Reject an article in the form it is submitted in order to achieve rapid turnaround. Especially encouraged are submissions on: Multidisciplinary topics, or applications-oriented articles and negative results that do not fit within the scope of IEEE''s traditional journals. Practical articles discussing new experiments or measurement techniques, interesting solutions to engineering. Development of new or improved fabrication or manufacturing techniques. Reviews or survey articles of new or evolving fields oriented to assist others in understanding the new area.