GridTrack: Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids

Özgür Erkent; David Sierra Gonzalez; Anshul Paigwar; Christian Laugier

doi:10.1007/978-3-030-87156-7_15

GridTrack: Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids

Özgür Erkent
, David Sierra Gonzalez
, Anshul Paigwar
, Christian Laugier

Institut national de recherche en informatique et en automatique

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Multiple Object Tracking is an important task for autonomous vehicles. However, it gets difficult to track objects when it is hard to detect them due to occlusion or distance to the sensors. We propose a method, “GridTrack”, to overcome this difficulty. We fuse a dynamic occupancy grid map (DOGMa) with an object detector. DOGMa is obtained by applying a Bayesian filter on raw sensor data. This improves the tracking of the partially observed/unobserved objects with the help of the Bayesian filter on raw data, which has a powerful prediction capability. We develop a network to track the objects on the grid and fuse information from previous detections in this network. The experiments show that the multi-object tracking accuracy is high with the usage of the proposed method.

Original language	English
Title of host publication	Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings
Editors	Markus Vincze, Timothy Patten, Henrik I Christensen, Lazaros Nalpantidis, Ming Liu
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	180-194
Number of pages	15
ISBN (Print)	9783030871550
DOIs	https://doi.org/10.1007/978-3-030-87156-7_15
Publication status	Published - 2021
Event	13th International Conference on Computer Vision Systems, ICVS 2021 - Virtual, Online Duration: 22 Sept 2021 → 24 Sept 2021

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12899 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	13th International Conference on Computer Vision Systems, ICVS 2021
City	Virtual, Online
Period	22/09/21 → 24/09/21

Keywords

Autonomous vehicles
Occupancy grids
Tracking

Access to Document

10.1007/978-3-030-87156-7_15

Cite this

Erkent, Ö., Gonzalez, D. S., Paigwar, A., & Laugier, C. (2021). GridTrack: Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids. In M. Vincze, T. Patten, H. I. Christensen, L. Nalpantidis, & M. Liu (Eds.), Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings (pp. 180-194). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12899 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-030-87156-7_15

Erkent, Özgür ; Gonzalez, David Sierra ; Paigwar, Anshul et al. / GridTrack : Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids. Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings. editor / Markus Vincze ; Timothy Patten ; Henrik I Christensen ; Lazaros Nalpantidis ; Ming Liu. Springer Science and Business Media Deutschland GmbH, 2021. pp. 180-194 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{8e880296864a4b20a883a17e30185b15,

title = "GridTrack: Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids",

abstract = "Multiple Object Tracking is an important task for autonomous vehicles. However, it gets difficult to track objects when it is hard to detect them due to occlusion or distance to the sensors. We propose a method, “GridTrack”, to overcome this difficulty. We fuse a dynamic occupancy grid map (DOGMa) with an object detector. DOGMa is obtained by applying a Bayesian filter on raw sensor data. This improves the tracking of the partially observed/unobserved objects with the help of the Bayesian filter on raw data, which has a powerful prediction capability. We develop a network to track the objects on the grid and fuse information from previous detections in this network. The experiments show that the multi-object tracking accuracy is high with the usage of the proposed method.",

keywords = "Autonomous vehicles, Occupancy grids, Tracking",

author = "{\"O}zg{\"u}r Erkent and Gonzalez, \{David Sierra\} and Anshul Paigwar and Christian Laugier",

note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Switzerland AG.; 13th International Conference on Computer Vision Systems, ICVS 2021 ; Conference date: 22-09-2021 Through 24-09-2021",

year = "2021",

doi = "10.1007/978-3-030-87156-7\_15",

language = "English",

isbn = "9783030871550",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "180--194",

editor = "Markus Vincze and Timothy Patten and Christensen, \{Henrik I\} and Lazaros Nalpantidis and Ming Liu",

booktitle = "Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings",

address = "Germany",

}

Erkent, Ö, Gonzalez, DS, Paigwar, A & Laugier, C 2021, GridTrack: Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids. in M Vincze, T Patten, HI Christensen, L Nalpantidis & M Liu (eds), Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12899 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 180-194, 13th International Conference on Computer Vision Systems, ICVS 2021, Virtual, Online, 22/09/21. https://doi.org/10.1007/978-3-030-87156-7_15

GridTrack: Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids. / Erkent, Özgür; Gonzalez, David Sierra; Paigwar, Anshul et al.
Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings. ed. / Markus Vincze; Timothy Patten; Henrik I Christensen; Lazaros Nalpantidis; Ming Liu. Springer Science and Business Media Deutschland GmbH, 2021. p. 180-194 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12899 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - GridTrack

T2 - 13th International Conference on Computer Vision Systems, ICVS 2021

AU - Erkent, Özgür

AU - Gonzalez, David Sierra

AU - Paigwar, Anshul

AU - Laugier, Christian

PY - 2021

Y1 - 2021

N2 - Multiple Object Tracking is an important task for autonomous vehicles. However, it gets difficult to track objects when it is hard to detect them due to occlusion or distance to the sensors. We propose a method, “GridTrack”, to overcome this difficulty. We fuse a dynamic occupancy grid map (DOGMa) with an object detector. DOGMa is obtained by applying a Bayesian filter on raw sensor data. This improves the tracking of the partially observed/unobserved objects with the help of the Bayesian filter on raw data, which has a powerful prediction capability. We develop a network to track the objects on the grid and fuse information from previous detections in this network. The experiments show that the multi-object tracking accuracy is high with the usage of the proposed method.

AB - Multiple Object Tracking is an important task for autonomous vehicles. However, it gets difficult to track objects when it is hard to detect them due to occlusion or distance to the sensors. We propose a method, “GridTrack”, to overcome this difficulty. We fuse a dynamic occupancy grid map (DOGMa) with an object detector. DOGMa is obtained by applying a Bayesian filter on raw sensor data. This improves the tracking of the partially observed/unobserved objects with the help of the Bayesian filter on raw data, which has a powerful prediction capability. We develop a network to track the objects on the grid and fuse information from previous detections in this network. The experiments show that the multi-object tracking accuracy is high with the usage of the proposed method.

KW - Autonomous vehicles

KW - Occupancy grids

KW - Tracking

UR - https://www.scopus.com/pages/publications/85115865117

U2 - 10.1007/978-3-030-87156-7_15

DO - 10.1007/978-3-030-87156-7_15

M3 - Conference contribution

AN - SCOPUS:85115865117

SN - 9783030871550

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 180

EP - 194

BT - Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings

A2 - Vincze, Markus

A2 - Patten, Timothy

A2 - Christensen, Henrik I

A2 - Nalpantidis, Lazaros

A2 - Liu, Ming

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 22 September 2021 through 24 September 2021

ER -

Erkent Ö, Gonzalez DS, Paigwar A, Laugier C. GridTrack: Detection and Tracking of Multiple Objects in Dynamic Occupancy Grids. In Vincze M, Patten T, Christensen HI, Nalpantidis L, Liu M, editors, Computer Vision Systems - 13th International Conference, ICVS 2021, Proceedings. Springer Science and Business Media Deutschland GmbH. 2021. p. 180-194. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-87156-7_15