Loading [MathJax]/extensions/tex2jax.js

Lifted Disjoint Paths with Application in Multiple Object Tracking

Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung

Autorschaft

Organisationseinheiten

Externe Organisationen

  • Max-Planck-Institut für Informatik

Details

OriginalspracheEnglisch
Titel des SammelwerksProceedings of the 37th International Conference on Machine Learning
PublikationsstatusVeröffentlicht - 2020

Publikationsreihe

NameProceedings of Machine Learning Research
Band119
ISSN (elektronisch)2640-3498

Abstract

We present an extension to the disjoint paths problem in which additional \emph{lifted} edges are introduced to provide path connectivity priors. We call the resulting optimization problem the lifted disjoint paths problem. We show that this problem is NP-hard by reduction from integer multicommodity flow and 3-SAT. To enable practical global optimization, we propose several classes of linear inequalities that produce a high-quality LP-relaxation. Additionally, we propose efficient cutting plane algorithms for separating the proposed linear inequalities. The lifted disjoint path problem is a natural model for multiple object tracking and allows an elegant mathematical formulation for long range temporal interactions. Lifted edges help to prevent id switches and to re-identify persons. Our lifted disjoint paths tracker achieves nearly optimal assignments with respect to input detections. As a consequence, it leads on all three main benchmarks of the MOT challenge, improving significantly over state-of-the-art.

Zitieren

Lifted Disjoint Paths with Application in Multiple Object Tracking. / Hornakova, Andrea; Henschel, Roberto; Rosenhahn, Bodo et al.
Proceedings of the 37th International Conference on Machine Learning. 2020. (Proceedings of Machine Learning Research; Band 119).

Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung

Hornakova, A, Henschel, R, Rosenhahn, B & Swoboda, P 2020, Lifted Disjoint Paths with Application in Multiple Object Tracking. in Proceedings of the 37th International Conference on Machine Learning. Proceedings of Machine Learning Research, Bd. 119. <https://arxiv.org/abs/2006.14550>
Hornakova, A., Henschel, R., Rosenhahn, B., & Swoboda, P. (2020). Lifted Disjoint Paths with Application in Multiple Object Tracking. In Proceedings of the 37th International Conference on Machine Learning (Proceedings of Machine Learning Research; Band 119). https://arxiv.org/abs/2006.14550
Hornakova A, Henschel R, Rosenhahn B, Swoboda P. Lifted Disjoint Paths with Application in Multiple Object Tracking. in Proceedings of the 37th International Conference on Machine Learning. 2020. (Proceedings of Machine Learning Research).
Hornakova, Andrea ; Henschel, Roberto ; Rosenhahn, Bodo et al. / Lifted Disjoint Paths with Application in Multiple Object Tracking. Proceedings of the 37th International Conference on Machine Learning. 2020. (Proceedings of Machine Learning Research).
Download
@inproceedings{31c5b8b6254e4d1baeb532938c280363,
title = "Lifted Disjoint Paths with Application in Multiple Object Tracking",
abstract = "We present an extension to the disjoint paths problem in which additional \emph{lifted} edges are introduced to provide path connectivity priors. We call the resulting optimization problem the lifted disjoint paths problem. We show that this problem is NP-hard by reduction from integer multicommodity flow and 3-SAT. To enable practical global optimization, we propose several classes of linear inequalities that produce a high-quality LP-relaxation. Additionally, we propose efficient cutting plane algorithms for separating the proposed linear inequalities. The lifted disjoint path problem is a natural model for multiple object tracking and allows an elegant mathematical formulation for long range temporal interactions. Lifted edges help to prevent id switches and to re-identify persons. Our lifted disjoint paths tracker achieves nearly optimal assignments with respect to input detections. As a consequence, it leads on all three main benchmarks of the MOT challenge, improving significantly over state-of-the-art. ",
keywords = "cs.CV, cs.DM",
author = "Andrea Hornakova and Roberto Henschel and Bodo Rosenhahn and Paul Swoboda",
note = "Acknowledgement This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany{\textquoteright}s Excellence Strategy within the Cluster of Excellence PhoenixD (EXC 2122). We thank Laura Leal-Taix{\'e} for initiating the collaboration. We thank all reviewers for their valuable comments.",
year = "2020",
language = "English",
series = "Proceedings of Machine Learning Research",
booktitle = "Proceedings of the 37th International Conference on Machine Learning",

}

Download

TY - GEN

T1 - Lifted Disjoint Paths with Application in Multiple Object Tracking

AU - Hornakova, Andrea

AU - Henschel, Roberto

AU - Rosenhahn, Bodo

AU - Swoboda, Paul

N1 - Acknowledgement This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy within the Cluster of Excellence PhoenixD (EXC 2122). We thank Laura Leal-Taixé for initiating the collaboration. We thank all reviewers for their valuable comments.

PY - 2020

Y1 - 2020

N2 - We present an extension to the disjoint paths problem in which additional \emph{lifted} edges are introduced to provide path connectivity priors. We call the resulting optimization problem the lifted disjoint paths problem. We show that this problem is NP-hard by reduction from integer multicommodity flow and 3-SAT. To enable practical global optimization, we propose several classes of linear inequalities that produce a high-quality LP-relaxation. Additionally, we propose efficient cutting plane algorithms for separating the proposed linear inequalities. The lifted disjoint path problem is a natural model for multiple object tracking and allows an elegant mathematical formulation for long range temporal interactions. Lifted edges help to prevent id switches and to re-identify persons. Our lifted disjoint paths tracker achieves nearly optimal assignments with respect to input detections. As a consequence, it leads on all three main benchmarks of the MOT challenge, improving significantly over state-of-the-art.

AB - We present an extension to the disjoint paths problem in which additional \emph{lifted} edges are introduced to provide path connectivity priors. We call the resulting optimization problem the lifted disjoint paths problem. We show that this problem is NP-hard by reduction from integer multicommodity flow and 3-SAT. To enable practical global optimization, we propose several classes of linear inequalities that produce a high-quality LP-relaxation. Additionally, we propose efficient cutting plane algorithms for separating the proposed linear inequalities. The lifted disjoint path problem is a natural model for multiple object tracking and allows an elegant mathematical formulation for long range temporal interactions. Lifted edges help to prevent id switches and to re-identify persons. Our lifted disjoint paths tracker achieves nearly optimal assignments with respect to input detections. As a consequence, it leads on all three main benchmarks of the MOT challenge, improving significantly over state-of-the-art.

KW - cs.CV

KW - cs.DM

UR - http://www.scopus.com/inward/record.url?scp=85105252245&partnerID=8YFLogxK

M3 - Conference contribution

T3 - Proceedings of Machine Learning Research

BT - Proceedings of the 37th International Conference on Machine Learning

ER -

Von denselben Autoren