Multiview Vehicle Tracking by Graph Matching Model

Minye Wu, Guli Zhang, Ning Bi, Ling Xie, Yuanquan Hu, Zhiru Shi; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2019, pp. 29-36

Abstract


Using multiple visual cameras to sensing traffic, especially tracking of vehicles, is a challenging task because of the large number of vehicle models, non-overlapping views, occlusion, view change and time-consuming algorithms. All of them remain obstacles in real world deployment. In this work, we propose a novel and flexible vehicle tracking framework, which formulates matching problem as a graph matching problem and solve it from the bottom up. In our framework, many restrictions can be added into the graph uniformly and simply. Moreover, we introduced an iterative Graph Matching Solver algorithm which can divide and reduce the graph matching problem's scale efficiently. Additionally, We also take the advantage of geographic information and make a combination with deep ReID features, motion and temporal information. The result shows that our algorithm achieves a 9th place at the AI City Challenge 2019.

Related Material


[pdf]
[bibtex]
@InProceedings{Wu_2019_CVPR_Workshops,
author = {Wu, Minye and Zhang, Guli and Bi, Ning and Xie, Ling and Hu, Yuanquan and Shi, Zhiru},
title = {Multiview Vehicle Tracking by Graph Matching Model},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2019}
}