Spatial-temporal motion information integration for action detection and recognition in non-static background

Dianting Liu, Mei-Ling Shyu, Guiru Zhao

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

7 Citations (Scopus)

Abstract

Various motion detection methods have been proposed in the past decade, but there are seldom attempts to investigate the advantages and disadvantages of different detection mechanisms so that they can complement each other to achieve a better performance. Toward such a demand, this paper proposes a human action detection and recognition framework to bridge the semantic gap between low-level pixel intensity change and the high-level understanding of the meaning of an action. To achieve a robust estimation of the region of action with the complexities of an uncontrolled background, we propose the combination of the optical flow field and Harris3D corner detector to obtain a new spatial-temporal estimation in the video sequences. The action detection method, considering the integrated motion information, works well with the dynamic background and camera motion, and demonstrates the advantage of the proposed method of integrating multiple spatial-temporal cues. Then the local features (SIFT and STIP) extracted from the estimated region of action are used to learn the Universal Background Model (UBM) for the action recognition task. The experimental results on KTH and UCF YouTube Action (UCF11) data sets show that the proposed action detection and recognition framework can not only better estimate the region of action but also achieve better recognition accuracy comparing with the peer work.

Original languageEnglish
Title of host publicationProceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013
PublisherIEEE Computer Society
Pages626-633
Number of pages8
ISBN (Print)9781479910502
DOIs
StatePublished - Jan 1 2013
Event2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013 - San Francisco, CA, United States
Duration: Aug 14 2013Aug 16 2013

Other

Other2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013
CountryUnited States
CitySan Francisco, CA
Period8/14/138/16/13

Fingerprint

Optical flows
Flow fields
Pixels
Semantics
Cameras
Detectors

Keywords

  • Action Detection
  • Action Recognition
  • Gaussian Mixture Models (GMM)
  • GMM Supervector
  • Spatio-temporal Motion Information Integration
  • Universal Background Model (UBM)

ASJC Scopus subject areas

  • Information Systems

Cite this

Liu, D., Shyu, M-L., & Zhao, G. (2013). Spatial-temporal motion information integration for action detection and recognition in non-static background. In Proceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013 (pp. 626-633). [6642527] IEEE Computer Society. https://doi.org/10.1109/IRI.2013.6642527

Spatial-temporal motion information integration for action detection and recognition in non-static background. / Liu, Dianting; Shyu, Mei-Ling; Zhao, Guiru.

Proceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013. IEEE Computer Society, 2013. p. 626-633 6642527.

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

Liu, D, Shyu, M-L & Zhao, G 2013, Spatial-temporal motion information integration for action detection and recognition in non-static background. in Proceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013., 6642527, IEEE Computer Society, pp. 626-633, 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013, San Francisco, CA, United States, 8/14/13. https://doi.org/10.1109/IRI.2013.6642527
Liu D, Shyu M-L, Zhao G. Spatial-temporal motion information integration for action detection and recognition in non-static background. In Proceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013. IEEE Computer Society. 2013. p. 626-633. 6642527 https://doi.org/10.1109/IRI.2013.6642527
Liu, Dianting ; Shyu, Mei-Ling ; Zhao, Guiru. / Spatial-temporal motion information integration for action detection and recognition in non-static background. Proceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013. IEEE Computer Society, 2013. pp. 626-633
@inproceedings{328f7b509ab0455191f68611e624fe98,
title = "Spatial-temporal motion information integration for action detection and recognition in non-static background",
abstract = "Various motion detection methods have been proposed in the past decade, but there are seldom attempts to investigate the advantages and disadvantages of different detection mechanisms so that they can complement each other to achieve a better performance. Toward such a demand, this paper proposes a human action detection and recognition framework to bridge the semantic gap between low-level pixel intensity change and the high-level understanding of the meaning of an action. To achieve a robust estimation of the region of action with the complexities of an uncontrolled background, we propose the combination of the optical flow field and Harris3D corner detector to obtain a new spatial-temporal estimation in the video sequences. The action detection method, considering the integrated motion information, works well with the dynamic background and camera motion, and demonstrates the advantage of the proposed method of integrating multiple spatial-temporal cues. Then the local features (SIFT and STIP) extracted from the estimated region of action are used to learn the Universal Background Model (UBM) for the action recognition task. The experimental results on KTH and UCF YouTube Action (UCF11) data sets show that the proposed action detection and recognition framework can not only better estimate the region of action but also achieve better recognition accuracy comparing with the peer work.",
keywords = "Action Detection, Action Recognition, Gaussian Mixture Models (GMM), GMM Supervector, Spatio-temporal Motion Information Integration, Universal Background Model (UBM)",
author = "Dianting Liu and Mei-Ling Shyu and Guiru Zhao",
year = "2013",
month = "1",
day = "1",
doi = "10.1109/IRI.2013.6642527",
language = "English",
isbn = "9781479910502",
pages = "626--633",
booktitle = "Proceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013",
publisher = "IEEE Computer Society",

}

TY - GEN

T1 - Spatial-temporal motion information integration for action detection and recognition in non-static background

AU - Liu, Dianting

AU - Shyu, Mei-Ling

AU - Zhao, Guiru

PY - 2013/1/1

Y1 - 2013/1/1

N2 - Various motion detection methods have been proposed in the past decade, but there are seldom attempts to investigate the advantages and disadvantages of different detection mechanisms so that they can complement each other to achieve a better performance. Toward such a demand, this paper proposes a human action detection and recognition framework to bridge the semantic gap between low-level pixel intensity change and the high-level understanding of the meaning of an action. To achieve a robust estimation of the region of action with the complexities of an uncontrolled background, we propose the combination of the optical flow field and Harris3D corner detector to obtain a new spatial-temporal estimation in the video sequences. The action detection method, considering the integrated motion information, works well with the dynamic background and camera motion, and demonstrates the advantage of the proposed method of integrating multiple spatial-temporal cues. Then the local features (SIFT and STIP) extracted from the estimated region of action are used to learn the Universal Background Model (UBM) for the action recognition task. The experimental results on KTH and UCF YouTube Action (UCF11) data sets show that the proposed action detection and recognition framework can not only better estimate the region of action but also achieve better recognition accuracy comparing with the peer work.

AB - Various motion detection methods have been proposed in the past decade, but there are seldom attempts to investigate the advantages and disadvantages of different detection mechanisms so that they can complement each other to achieve a better performance. Toward such a demand, this paper proposes a human action detection and recognition framework to bridge the semantic gap between low-level pixel intensity change and the high-level understanding of the meaning of an action. To achieve a robust estimation of the region of action with the complexities of an uncontrolled background, we propose the combination of the optical flow field and Harris3D corner detector to obtain a new spatial-temporal estimation in the video sequences. The action detection method, considering the integrated motion information, works well with the dynamic background and camera motion, and demonstrates the advantage of the proposed method of integrating multiple spatial-temporal cues. Then the local features (SIFT and STIP) extracted from the estimated region of action are used to learn the Universal Background Model (UBM) for the action recognition task. The experimental results on KTH and UCF YouTube Action (UCF11) data sets show that the proposed action detection and recognition framework can not only better estimate the region of action but also achieve better recognition accuracy comparing with the peer work.

KW - Action Detection

KW - Action Recognition

KW - Gaussian Mixture Models (GMM)

KW - GMM Supervector

KW - Spatio-temporal Motion Information Integration

KW - Universal Background Model (UBM)

UR - http://www.scopus.com/inward/record.url?scp=84891125039&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84891125039&partnerID=8YFLogxK

U2 - 10.1109/IRI.2013.6642527

DO - 10.1109/IRI.2013.6642527

M3 - Conference contribution

SN - 9781479910502

SP - 626

EP - 633

BT - Proceedings of the 2013 IEEE 14th International Conference on Information Reuse and Integration, IEEE IRI 2013

PB - IEEE Computer Society

ER -