Crowd saliency detection via global similarity structure

Mei Kuan Lim, Kok Ven Jyn , Chen Change Loy, Chee Seng Chan

Research output: Chapter in Book/Report/Conference proceedingConference contribution

18 Citations (Scopus)

Abstract

It is common for CCTV operators to overlook interesting events taking place within the crowd due to large number of people in the crowded scene (i.e. marathon, rally). Thus, there is a dire need to automate the detection of salient crowd regions acquiring immediate attention for a more effective and proactive surveillance. This paper proposes a novel framework to identify and localize salient regions in a crowd scene, by transforming low-level features extracted from crowd motion field into a global similarity structure. The global similarity structure representation allows the discovery of the intrinsic manifold of the motion dynamics, which could not be captured by the low-level representation. Ranking is then performed on the global similarity structure to identify a set of extrem a. The proposed approach is unsupervised so learning stage is eliminated. Experimental results on public datasets demonstrates the effectiveness of exploiting such extrem a in identifying salient regions in various crowd scenarios that exhibit crowding, local irregular motion, and unique motion areas such as sources and sinks.

Original languageEnglish
Title of host publicationProceedings - International Conference on Pattern Recognition
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3957-3962
Number of pages6
ISBN (Electronic)9781479952083
DOIs
Publication statusPublished - 1 Jan 2014
Externally publishedYes
Event22nd International Conference on Pattern Recognition, ICPR 2014 - Stockholm, Sweden
Duration: 24 Aug 201428 Aug 2014

Other

Other22nd International Conference on Pattern Recognition, ICPR 2014
CountrySweden
CityStockholm
Period24/8/1428/8/14

Fingerprint

Closed circuit television systems
Unsupervised learning

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition

Cite this

Lim, M. K., Ven Jyn , K., Loy, C. C., & Chan, C. S. (2014). Crowd saliency detection via global similarity structure. In Proceedings - International Conference on Pattern Recognition (pp. 3957-3962). [6977391] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICPR.2014.678

Crowd saliency detection via global similarity structure. / Lim, Mei Kuan; Ven Jyn , Kok; Loy, Chen Change; Chan, Chee Seng.

Proceedings - International Conference on Pattern Recognition. Institute of Electrical and Electronics Engineers Inc., 2014. p. 3957-3962 6977391.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Lim, MK, Ven Jyn , K, Loy, CC & Chan, CS 2014, Crowd saliency detection via global similarity structure. in Proceedings - International Conference on Pattern Recognition., 6977391, Institute of Electrical and Electronics Engineers Inc., pp. 3957-3962, 22nd International Conference on Pattern Recognition, ICPR 2014, Stockholm, Sweden, 24/8/14. https://doi.org/10.1109/ICPR.2014.678
Lim MK, Ven Jyn  K, Loy CC, Chan CS. Crowd saliency detection via global similarity structure. In Proceedings - International Conference on Pattern Recognition. Institute of Electrical and Electronics Engineers Inc. 2014. p. 3957-3962. 6977391 https://doi.org/10.1109/ICPR.2014.678
Lim, Mei Kuan ; Ven Jyn , Kok ; Loy, Chen Change ; Chan, Chee Seng. / Crowd saliency detection via global similarity structure. Proceedings - International Conference on Pattern Recognition. Institute of Electrical and Electronics Engineers Inc., 2014. pp. 3957-3962
@inproceedings{499fe580900c4b6f8f189c595bae6ff0,
title = "Crowd saliency detection via global similarity structure",
abstract = "It is common for CCTV operators to overlook interesting events taking place within the crowd due to large number of people in the crowded scene (i.e. marathon, rally). Thus, there is a dire need to automate the detection of salient crowd regions acquiring immediate attention for a more effective and proactive surveillance. This paper proposes a novel framework to identify and localize salient regions in a crowd scene, by transforming low-level features extracted from crowd motion field into a global similarity structure. The global similarity structure representation allows the discovery of the intrinsic manifold of the motion dynamics, which could not be captured by the low-level representation. Ranking is then performed on the global similarity structure to identify a set of extrem a. The proposed approach is unsupervised so learning stage is eliminated. Experimental results on public datasets demonstrates the effectiveness of exploiting such extrem a in identifying salient regions in various crowd scenarios that exhibit crowding, local irregular motion, and unique motion areas such as sources and sinks.",
author = "Lim, {Mei Kuan} and {Ven Jyn }, Kok and Loy, {Chen Change} and Chan, {Chee Seng}",
year = "2014",
month = "1",
day = "1",
doi = "10.1109/ICPR.2014.678",
language = "English",
pages = "3957--3962",
booktitle = "Proceedings - International Conference on Pattern Recognition",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Crowd saliency detection via global similarity structure

AU - Lim, Mei Kuan

AU - Ven Jyn , Kok

AU - Loy, Chen Change

AU - Chan, Chee Seng

PY - 2014/1/1

Y1 - 2014/1/1

N2 - It is common for CCTV operators to overlook interesting events taking place within the crowd due to large number of people in the crowded scene (i.e. marathon, rally). Thus, there is a dire need to automate the detection of salient crowd regions acquiring immediate attention for a more effective and proactive surveillance. This paper proposes a novel framework to identify and localize salient regions in a crowd scene, by transforming low-level features extracted from crowd motion field into a global similarity structure. The global similarity structure representation allows the discovery of the intrinsic manifold of the motion dynamics, which could not be captured by the low-level representation. Ranking is then performed on the global similarity structure to identify a set of extrem a. The proposed approach is unsupervised so learning stage is eliminated. Experimental results on public datasets demonstrates the effectiveness of exploiting such extrem a in identifying salient regions in various crowd scenarios that exhibit crowding, local irregular motion, and unique motion areas such as sources and sinks.

AB - It is common for CCTV operators to overlook interesting events taking place within the crowd due to large number of people in the crowded scene (i.e. marathon, rally). Thus, there is a dire need to automate the detection of salient crowd regions acquiring immediate attention for a more effective and proactive surveillance. This paper proposes a novel framework to identify and localize salient regions in a crowd scene, by transforming low-level features extracted from crowd motion field into a global similarity structure. The global similarity structure representation allows the discovery of the intrinsic manifold of the motion dynamics, which could not be captured by the low-level representation. Ranking is then performed on the global similarity structure to identify a set of extrem a. The proposed approach is unsupervised so learning stage is eliminated. Experimental results on public datasets demonstrates the effectiveness of exploiting such extrem a in identifying salient regions in various crowd scenarios that exhibit crowding, local irregular motion, and unique motion areas such as sources and sinks.

UR - http://www.scopus.com/inward/record.url?scp=84919904395&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84919904395&partnerID=8YFLogxK

U2 - 10.1109/ICPR.2014.678

DO - 10.1109/ICPR.2014.678

M3 - Conference contribution

AN - SCOPUS:84919904395

SP - 3957

EP - 3962

BT - Proceedings - International Conference on Pattern Recognition

PB - Institute of Electrical and Electronics Engineers Inc.

ER -