Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians

Robert T. Collins; Peter Carr

doi:10.1007/978-3-319-10605-2_20

Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians

Robert T. Collins, Peter Carr

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

12 Scopus citations

Abstract

Although 'tracking-by-detection' is a popular approach when reliable object detectors are available, missed detections remain a difficult hurdle to overcome. We present a hybrid stochastic/deterministic optimization scheme that uses RJMCMC to perform stochastic search over the space of detection configurations, interleaved with deterministic computation of the optimal multi-frame data association for each proposed detection hypothesis. Since object trajectories do not need to be estimated directly by the sampler, our approach is more efficient than traditional MCMCDA techniques. Moreover, our holistic formulation is able to generate longer, more reliable trajectories than baseline tracking-by-detection approaches in challenging multi-target scenarios.

Original language	English (US)
Title of host publication	Computer Vision, ECCV 2014 - 13th European Conference, Proceedings
Publisher	Springer Verlag
Pages	298-313
Number of pages	16
Edition	PART 2
ISBN (Print)	9783319106045
DOIs	https://doi.org/10.1007/978-3-319-10605-2_20
State	Published - 2014
Event	13th European Conference on Computer Vision, ECCV 2014 - Zurich, Switzerland Duration: Sep 6 2014 → Sep 12 2014

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Number	PART 2
Volume	8690 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	13th European Conference on Computer Vision, ECCV 2014
Country/Territory	Switzerland
City	Zurich
Period	9/6/14 → 9/12/14

All Science Journal Classification (ASJC) codes

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-319-10605-2_20

Cite this

Collins, R. T., & Carr, P. (2014). Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians. In Computer Vision, ECCV 2014 - 13th European Conference, Proceedings (PART 2 ed., pp. 298-313). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8690 LNCS, No. PART 2). Springer Verlag. https://doi.org/10.1007/978-3-319-10605-2_20

Collins, Robert T. ; Carr, Peter. / Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians. Computer Vision, ECCV 2014 - 13th European Conference, Proceedings. PART 2. ed. Springer Verlag, 2014. pp. 298-313 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).

@inproceedings{487bc667d0b24a10b3e07a21be02741f,

title = "Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians",

abstract = "Although 'tracking-by-detection' is a popular approach when reliable object detectors are available, missed detections remain a difficult hurdle to overcome. We present a hybrid stochastic/deterministic optimization scheme that uses RJMCMC to perform stochastic search over the space of detection configurations, interleaved with deterministic computation of the optimal multi-frame data association for each proposed detection hypothesis. Since object trajectories do not need to be estimated directly by the sampler, our approach is more efficient than traditional MCMCDA techniques. Moreover, our holistic formulation is able to generate longer, more reliable trajectories than baseline tracking-by-detection approaches in challenging multi-target scenarios.",

author = "Collins, {Robert T.} and Peter Carr",

year = "2014",

doi = "10.1007/978-3-319-10605-2_20",

language = "English (US)",

isbn = "9783319106045",

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

publisher = "Springer Verlag",

number = "PART 2",

pages = "298--313",

booktitle = "Computer Vision, ECCV 2014 - 13th European Conference, Proceedings",

address = "Germany",

edition = "PART 2",

note = "13th European Conference on Computer Vision, ECCV 2014 ; Conference date: 06-09-2014 Through 12-09-2014",

}

Collins, RT & Carr, P 2014, Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians. in Computer Vision, ECCV 2014 - 13th European Conference, Proceedings. PART 2 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 8690 LNCS, Springer Verlag, pp. 298-313, 13th European Conference on Computer Vision, ECCV 2014, Zurich, Switzerland, 9/6/14. https://doi.org/10.1007/978-3-319-10605-2_20

Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians. / Collins, Robert T.; Carr, Peter.
Computer Vision, ECCV 2014 - 13th European Conference, Proceedings. PART 2. ed. Springer Verlag, 2014. p. 298-313 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8690 LNCS, No. PART 2).

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

TY - GEN

T1 - Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians

AU - Collins, Robert T.

AU - Carr, Peter

PY - 2014

Y1 - 2014

N2 - Although 'tracking-by-detection' is a popular approach when reliable object detectors are available, missed detections remain a difficult hurdle to overcome. We present a hybrid stochastic/deterministic optimization scheme that uses RJMCMC to perform stochastic search over the space of detection configurations, interleaved with deterministic computation of the optimal multi-frame data association for each proposed detection hypothesis. Since object trajectories do not need to be estimated directly by the sampler, our approach is more efficient than traditional MCMCDA techniques. Moreover, our holistic formulation is able to generate longer, more reliable trajectories than baseline tracking-by-detection approaches in challenging multi-target scenarios.

AB - Although 'tracking-by-detection' is a popular approach when reliable object detectors are available, missed detections remain a difficult hurdle to overcome. We present a hybrid stochastic/deterministic optimization scheme that uses RJMCMC to perform stochastic search over the space of detection configurations, interleaved with deterministic computation of the optimal multi-frame data association for each proposed detection hypothesis. Since object trajectories do not need to be estimated directly by the sampler, our approach is more efficient than traditional MCMCDA techniques. Moreover, our holistic formulation is able to generate longer, more reliable trajectories than baseline tracking-by-detection approaches in challenging multi-target scenarios.

UR - http://www.scopus.com/inward/record.url?scp=84906518711&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84906518711&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-10605-2_20

DO - 10.1007/978-3-319-10605-2_20

M3 - Conference contribution

AN - SCOPUS:84906518711

SN - 9783319106045

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

SP - 298

EP - 313

BT - Computer Vision, ECCV 2014 - 13th European Conference, Proceedings

PB - Springer Verlag

T2 - 13th European Conference on Computer Vision, ECCV 2014

Y2 - 6 September 2014 through 12 September 2014

ER -

Collins RT, Carr P. Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians. In Computer Vision, ECCV 2014 - 13th European Conference, Proceedings. PART 2 ed. Springer Verlag. 2014. p. 298-313. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). doi: 10.1007/978-3-319-10605-2_20

Hybrid stochastic/deterministic optimization for tracking sports players and pedestrians

Abstract

Publication series

Other

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this