《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (5): 1372-1377.DOI: 10.11772/j.issn.1001-9081.2022030377

• 第九届中国数据挖掘会议 • 上一篇    

融合人体全身表观特征的行人头部跟踪模型

张广耀1,2, 宋纯锋1,2()   

  1. 1.中国科学院大学 人工智能学院,北京 100049
    2.中国科学院自动化研究所 智能感知与计算研究中心,北京 100190
  • 收稿日期:2022-03-28 修回日期:2022-05-11 接受日期:2022-05-19 发布日期:2023-05-08 出版日期:2023-05-10
  • 通讯作者: 宋纯锋
  • 作者简介:张广耀(1996—),男,山东省济南人,硕士研究生,主要研究方向:目标检测、多目标跟踪
    宋纯锋(1989—),男,山东泰安人,助理研究员,博士,CCF会员,主要研究方向:模式识别、计算机视觉、深度学习。chunfeng.song@nlpr.ia.ac.cn
  • 基金资助:
    国家自然科学基金资助项目(62006231)

Pedestrian head tracking model based on full-body appearance features

Guangyao ZHANG1,2, Chunfeng SONG1,2()   

  1. 1.School of Artificial Intelligence,University of Chinese Academy of Sciences,Beijing 100049,China
    2.Center for Research on Intelligent Perception and Computing,Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China
  • Received:2022-03-28 Revised:2022-05-11 Accepted:2022-05-19 Online:2023-05-08 Published:2023-05-10
  • Contact: Chunfeng SONG
  • About author:ZHANG Guangyao, born in 1996, M. S. candidate. His research interests include object detection, multi-object tracking.
    SONG Chunfeng, born in 1989, Ph. D., assistant research fellow. His research interests include pattern recognition, computer vision, deep learning.
  • Supported by:
    National Natural Science Foundation of China(62006231)

摘要:

现有的行人多目标跟踪模型在密集场景下存在行人无法检出以及帧间关联混淆的问题。为了提高密集场景下行人跟踪的精确率,提出一种融合全身表观特征的行人头部跟踪模型HT-FF (Head Tracking with Full-body Features)。首先,使用行人头部检测器替代全身检测器,提高密集场景下行人的检出率;其次,利用人体姿态估计的信息为引导,获得去噪声的全身表观特征作为跟踪线索,大幅减少多帧之间关联时发生的混淆。HT-FF模型在密集场景下行人跟踪的基准数据集Head Tracking 21 (HT21)上的MOTA (Multiple Object Tracking Accuracy)和IDF1 (ID F1 Score)等多个指标上取得了最优的结果。HT-FF模型能有效缓解密集场景下行人跟踪丢失和混淆的问题,所提出的融合多线索的跟踪模型是行人跟踪任务的新范式。

关键词: 多目标跟踪, 运动模型, 动态模型, 特征匹配, 行人头部跟踪, 行人重识别, 人体姿态估计, 表观特征

Abstract:

The existing pedestrian multi-object tracking algorithms have the problems of undetectable pedestrians and inter-frame association confusion in dense scenes. In order to improve the precision of pedestrian tracking in dense scenes, a head tracking model based on full-body appearance features was proposed, namely HT-FF (Head Tracking with Full-body Features). Firstly, the head detector was used to replace the full-body detector to improve the detection rate of pedestrians in dense scenes. Secondly, using the information of human posture estimation as a guide, the noise-removed full-body appearance features were obtained as tracking clues, which greatly reduced the confusion in the association among multiple frames. HT-FF model achieves the best results on multiple indicators such as MOTA (Multiple Object Tracking Accuracy) and IDF1 (ID F1 Score) on benchmark dataset of pedestrian tracking in dense scenes — Head Tracking 21 (HT21). The HT-FF model can effectively alleviate the problem of lost and confused pedestrian tracking in dense scenes, and the proposed tracking model combining multiple clues is a new paradigm of pedestrian tracking model.

Key words: multi-object tracking, motion model, dynamic model, feature matching, pedestrian head tracking, pedestrian re-identification, human pose estimation, appearance feature

中图分类号: