Spatio-temporal context for more accurate dense point trajectories estimation

Qingxuan Shi; Yao Lu; Tianfei Zhou

doi:10.1109/CIS.2014.137

Spatio-temporal context for more accurate dense point trajectories estimation

Qingxuan Shi, Yao Lu, Tianfei Zhou

School of Computer Science and Technology

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

2 Citations (Scopus)

Abstract

Dense point trajectories estimation is a challenging yet important problem due to its potential of supporting other fields, such as motion estimation, action recognition, etc. In previous work, dense motion trackers always estimate trajectories based on consecutive frames and ignore scene context prior, thereby suffering from inaccurate estimation. In this paper, we present a novel dense point trajectories estimation framework which integrates trajectories spatio-temporal context into the estimation process. The spatial context for a trajectory refers to the support from its neighbouring trajectories, while the temporal context indicates the temporal appearance consistency for each trajectory. To obtain accurate and compact trajectories, we formulate the problem as an inference process in a Markov Random Field(MRF).We measure the accuracy of the algorithms on MIT sequences. Experimental results demonstrate that our methods can give more accurate dense point trajectories efficiently.

Original language	English
Title of host publication	Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	256-259
Number of pages	4
ISBN (Electronic)	9781479974344
DOIs	https://doi.org/10.1109/CIS.2014.137
Publication status	Published - 20 Jan 2015
Event	10th International Conference on Computational Intelligence and Security, CIS 2014 - Kunming, Yunnan, China Duration: 15 Nov 2014 → 16 Nov 2014

Publication series

Name	Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014

Conference

Conference	10th International Conference on Computational Intelligence and Security, CIS 2014
Country/Territory	China
City	Kunming, Yunnan
Period	15/11/14 → 16/11/14

Keywords

Dense point trajectories estimation
Markov random fields
Spatio-temporal context

Access to Document

10.1109/CIS.2014.137

Cite this

Shi, Q., Lu, Y., & Zhou, T. (2015). Spatio-temporal context for more accurate dense point trajectories estimation. In Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014 (pp. 256-259). Article 7016895 (Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CIS.2014.137

Shi, Qingxuan ; Lu, Yao ; Zhou, Tianfei. / Spatio-temporal context for more accurate dense point trajectories estimation. Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014. Institute of Electrical and Electronics Engineers Inc., 2015. pp. 256-259 (Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014).

@inproceedings{ed2980e29aa647c894ac0a48a6f024a0,

title = "Spatio-temporal context for more accurate dense point trajectories estimation",

abstract = "Dense point trajectories estimation is a challenging yet important problem due to its potential of supporting other fields, such as motion estimation, action recognition, etc. In previous work, dense motion trackers always estimate trajectories based on consecutive frames and ignore scene context prior, thereby suffering from inaccurate estimation. In this paper, we present a novel dense point trajectories estimation framework which integrates trajectories spatio-temporal context into the estimation process. The spatial context for a trajectory refers to the support from its neighbouring trajectories, while the temporal context indicates the temporal appearance consistency for each trajectory. To obtain accurate and compact trajectories, we formulate the problem as an inference process in a Markov Random Field(MRF).We measure the accuracy of the algorithms on MIT sequences. Experimental results demonstrate that our methods can give more accurate dense point trajectories efficiently.",

keywords = "Dense point trajectories estimation, Markov random fields, Spatio-temporal context",

author = "Qingxuan Shi and Yao Lu and Tianfei Zhou",

note = "Publisher Copyright: {\textcopyright} 2014 IEEE.; 10th International Conference on Computational Intelligence and Security, CIS 2014 ; Conference date: 15-11-2014 Through 16-11-2014",

year = "2015",

month = jan,

day = "20",

doi = "10.1109/CIS.2014.137",

language = "English",

series = "Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "256--259",

booktitle = "Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014",

address = "United States",

}

Shi, Q, Lu, Y & Zhou, T 2015, Spatio-temporal context for more accurate dense point trajectories estimation. in Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014., 7016895, Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014, Institute of Electrical and Electronics Engineers Inc., pp. 256-259, 10th International Conference on Computational Intelligence and Security, CIS 2014, Kunming, Yunnan, China, 15/11/14. https://doi.org/10.1109/CIS.2014.137

Spatio-temporal context for more accurate dense point trajectories estimation. / Shi, Qingxuan; Lu, Yao; Zhou, Tianfei.
Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014. Institute of Electrical and Electronics Engineers Inc., 2015. p. 256-259 7016895 (Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014).

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

TY - GEN

T1 - Spatio-temporal context for more accurate dense point trajectories estimation

AU - Shi, Qingxuan

AU - Lu, Yao

AU - Zhou, Tianfei

PY - 2015/1/20

Y1 - 2015/1/20

N2 - Dense point trajectories estimation is a challenging yet important problem due to its potential of supporting other fields, such as motion estimation, action recognition, etc. In previous work, dense motion trackers always estimate trajectories based on consecutive frames and ignore scene context prior, thereby suffering from inaccurate estimation. In this paper, we present a novel dense point trajectories estimation framework which integrates trajectories spatio-temporal context into the estimation process. The spatial context for a trajectory refers to the support from its neighbouring trajectories, while the temporal context indicates the temporal appearance consistency for each trajectory. To obtain accurate and compact trajectories, we formulate the problem as an inference process in a Markov Random Field(MRF).We measure the accuracy of the algorithms on MIT sequences. Experimental results demonstrate that our methods can give more accurate dense point trajectories efficiently.

AB - Dense point trajectories estimation is a challenging yet important problem due to its potential of supporting other fields, such as motion estimation, action recognition, etc. In previous work, dense motion trackers always estimate trajectories based on consecutive frames and ignore scene context prior, thereby suffering from inaccurate estimation. In this paper, we present a novel dense point trajectories estimation framework which integrates trajectories spatio-temporal context into the estimation process. The spatial context for a trajectory refers to the support from its neighbouring trajectories, while the temporal context indicates the temporal appearance consistency for each trajectory. To obtain accurate and compact trajectories, we formulate the problem as an inference process in a Markov Random Field(MRF).We measure the accuracy of the algorithms on MIT sequences. Experimental results demonstrate that our methods can give more accurate dense point trajectories efficiently.

KW - Dense point trajectories estimation

KW - Markov random fields

KW - Spatio-temporal context

UR - http://www.scopus.com/inward/record.url?scp=84922874590&partnerID=8YFLogxK

U2 - 10.1109/CIS.2014.137

DO - 10.1109/CIS.2014.137

M3 - Conference contribution

AN - SCOPUS:84922874590

T3 - Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014

SP - 256

EP - 259

BT - Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 10th International Conference on Computational Intelligence and Security, CIS 2014

Y2 - 15 November 2014 through 16 November 2014

ER -

Shi Q, Lu Y, Zhou T. Spatio-temporal context for more accurate dense point trajectories estimation. In Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014. Institute of Electrical and Electronics Engineers Inc. 2015. p. 256-259. 7016895. (Proceedings - 2014 10th International Conference on Computational Intelligence and Security, CIS 2014). doi: 10.1109/CIS.2014.137

Spatio-temporal context for more accurate dense point trajectories estimation

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this