《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (3): 661-673.DOI: 10.11772/j.issn.1001-9081.2022010150

• 人工智能 •    下一篇

基于孪生网络的单目标跟踪算法综述

王梦亭, 杨文忠(), 武雍智   

  1. 新疆大学 信息科学与工程学院,乌鲁木齐 830046
  • 收稿日期:2022-02-11 修回日期:2022-04-28 接受日期:2022-05-05 发布日期:2022-05-24 出版日期:2023-03-10
  • 通讯作者: 杨文忠
  • 作者简介:王梦亭(1995—),女,河南周口人,硕士研究生,主要研究方向:单目标跟踪、计算机视觉
    杨文忠(1971—),男,河南南阳人,副教授,博士,CCF会员,主要研究方向:图像处理
    武雍智(1995—),男,甘肃张掖人,硕士研究生,主要研究方向:行人重识别、计算机视觉。
  • 基金资助:
    新疆维吾尔自治区科技重大专项(2020A02001-1);新疆维吾尔自治区科技计划项目(202104120007);江西省自然科学基金资助项目(20202BAB202023)

Survey of single target tracking algorithms based on Siamese network

Mengting WANG, Wenzhong YANG(), Yongzhi WU   

  1. School of Information Science and Engineering,Xinjiang University,Urumqi Xinjiang 830046,China
  • Received:2022-02-11 Revised:2022-04-28 Accepted:2022-05-05 Online:2022-05-24 Published:2023-03-10
  • Contact: Wenzhong YANG
  • About author:WANG Mengting, born in 1995, M. S. candidate. Her research interests include single object tracking, computer vision.
    YANG Wenzhong, born in 1971, Ph. D., associate professor. His research interests include image processing.
    WU Yongzhi, born in 1995, M. S. candidate. His research interests include person re-identification, computer vision.
  • Supported by:
    Major Project of Science and Technology of Xinjiang Uygur Autonomous Region(2020A02001-1);Science and Technology Program of Xinjiang Uygur Autonomous Region(202104120007);Natural Science Foundation of Jiangxi Province(20202BAB202023)

摘要:

单目标跟踪是计算机视觉领域的一个重要研究方向,在视频监控、自动驾驶等领域应用广泛。对于单目标跟踪算法,尽管已有大量总结研究,但大多基于相关滤波或深度学习。近年来,基于孪生网络的跟踪算法因在精度和速度之间取得的平衡受到研究者们的广泛关注,然而目前对该类型算法的总结分析相对较少,并且对这些算法的架构层面缺少系统分析。为深入了解基于孪生网络的单目标跟踪算法,对大量相关文献进行了总结与分析。首先阐述孪生网络的结构和应用,并根据孪生跟踪算法架构组成的分类介绍了各跟踪算法;然后列举单目标跟踪领域常用的数据集和评价指标,对25个主流跟踪算法在OTB2015数据集上分别进行整体和各属性的性能比较与分析,并列出23个孪生跟踪算法在LaSOT和GOT-10K测试集上的性能以及推理时的速度;最后对基于孪生网络的目标跟踪算法的研究进行总结,并对未来的发展方向进行展望。

关键词: 孪生网络, 单目标跟踪, 计算机视觉, 互相关, 无锚框

Abstract:

Single object tracking is an important research direction in the field of computer vision, and has a wide range of applications in video surveillance, autonomous driving and other fields. For single object tracking algorithms, although a large number of summaries have been conducted, most of them are based on correlation filter or deep learning. In recent years, Siamese network-based tracking algorithms have received extensive attention from researchers for their balance between accuracy and speed, but there are relatively few summaries of this type of algorithms and it lacks systematic analysis of the algorithms at the architectural level. In order to deeply understand the single object tracking algorithms based on Siamese network, a large number of related literatures were organized and analyzed. Firstly, the structures and applications of the Siamese network were expounded, and each tracking algorithm was introduced according to the composition classification of the Siamese tracking algorithm architectures. Then, the commonly used datasets and evaluation metrics in the field of single object tracking were listed, the overall and each attribute performance of 25 mainstream tracking algorithms was compared and analyzed on OTB 2015 (Object Tracking Benchmark) dataset, and the performance and the reasoning speed of 23 Siamese network-based tracking algorithms on LaSOT (Large-scale Single Object Tracking) and GOT-10K (Generic Object Tracking) test sets were listed. Finally, the research on Siamese network-based tracking algorithms was summarized, and the possible future research directions of this type of algorithms were prospected.

Key words: Siamese network, single target tracking, computer vision, cross-correlation, anchor-free

中图分类号: