邹北骥, 李伯洲, 刘姝. 基于中心点检测和重识别的多行人跟踪算法[J]. 武汉大学学报 ( 信息科学版), 2021, 46(9): 1345-1353. DOI: 10.13203/j.whugis20210328
引用本文: 邹北骥, 李伯洲, 刘姝. 基于中心点检测和重识别的多行人跟踪算法[J]. 武汉大学学报 ( 信息科学版), 2021, 46(9): 1345-1353. DOI: 10.13203/j.whugis20210328
ZOU Beiji, LI Bozhou, LIU Shu. A Multi-Pedestrian Tracking Algorithm Based on Center Point Detection and Person Re-identification[J]. Geomatics and Information Science of Wuhan University, 2021, 46(9): 1345-1353. DOI: 10.13203/j.whugis20210328
Citation: ZOU Beiji, LI Bozhou, LIU Shu. A Multi-Pedestrian Tracking Algorithm Based on Center Point Detection and Person Re-identification[J]. Geomatics and Information Science of Wuhan University, 2021, 46(9): 1345-1353. DOI: 10.13203/j.whugis20210328

基于中心点检测和重识别的多行人跟踪算法

A Multi-Pedestrian Tracking Algorithm Based on Center Point Detection and Person Re-identification

  • 摘要: 在基于视频的多目标运动跟踪中,目标检测和重识别具有很强的相关性。目前常将目标检测和重识别网络分别进行训练和使用,因此实时跟踪速度不能达到要求。针对多目标跟踪(multiple object tracking,MOT)中行人身份切换和跟踪丢失问题,将行人重识别模块进行遮挡优化并嵌入行人检测网络,由此提出了一种基于中心点检测和重识别的多行人跟踪算法。首先建立了行人运动模型,通过中心点检测得到行人最优状态估计;然后根据深层特征融合的行人重识别模型,利用马氏距离和余弦距离增强行人身份辨别能力;最后利用匈牙利算法进行在线数据关联,同时利用卡尔曼滤波剔除不准确的结果,对未关联的丢失目标做运动预测。利用所提算法和其他跟踪算法分别在MOT15、MOT16、MOT17数据集上进行多行人跟踪对比实验,结果表明,所提算法的多目标跟踪精度(multiple object tracking accuracy,MOTA)分别为63.5、72.4、70.9, 正确识别的检测和计算的检测数的比值(identity F1?measure,IDF1)最优, 且保证了实时跟踪速率, 验证了所提跟踪算法的有效性。

     

    Abstract:
      Objectives  In video-based multiple object tracking, the object detection and re-identification have a strong correlation. The existing methods generally train the object detection and re-identification networks separately, which makes the tracking speed fail to meet the real-time requirements. In this paper, we integrate the detection and re-identification into one network to accelerate the tracking process, and also solve the problems of identity switching and failure tracking.
      Methods  This paper develops a pedestrian motion model and obtains the optimal state estimation of pedestrians using center point detection. The person re-identification model with deep layer features uses the Mahalanobis distance and cosine distance to enhance the ability of person identification. And the Hungary algorithm is used for data online association, where the state estimation results become more accurate using Kalman filtering, and the unrelated lost objects are predicted by motion.
      Results  Experiments are conducted on MOT15, MOT16 and MOT17 datasets using the proposed algorithm and other multi-pedestrian tracking algorithms, and the multiple object tracking accuracy of tracking results using our proposed algorithm is 63.5, 72.4 and 70.9, respectively, and the identity F1-measure is optimal, with the real-time rate.
      Conclusions  The proposed algorithm can accelerate the tracking speed by network parameter sharing, and improve the recognition accuracy by person re-identification training.

     

/

返回文章
返回