Video event detection: From subvolume localization to spatiotemporal path search

Du Tran, Junsong Yuan, David Alexander Forsyth

Research output: Contribution to journalArticle

Abstract

Although sliding window-based approaches have been quite successful in detecting objects in images, it is not a trivial problem to extend them to detecting events in videos. We propose to search for spatiotemporal paths for video event detection. This new formulation can accurately detect and locate video events in cluttered and crowded scenes, and is robust to camera motions. It can also well handle the scale, shape, and intraclass variations of the event. Compared to event detection using spatiotemporal sliding windows, the spatiotemporal paths correspond to the event trajectories in the video space, thus can better handle events composed by moving objects. We prove that the proposed search algorithm can achieve the global optimal solution with the lowest complexity. Experiments are conducted on realistic video data sets with different event detection tasks, such as anomaly event detection, walking person detection, and running detection. Our proposed method is compatible with different types of video features or object detectors and robust to false and missed local detections. It significantly improves the overall detection and localization accuracy over the state-of-the-art methods.

Original languageEnglish (US)
Article number6567857
Pages (from-to)404-416
Number of pages13
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume36
Issue number2
DOIs
StatePublished - Feb 1 2014

Fingerprint

Event Detection
Cameras
Trajectories
Detectors
Path
Sliding Window
Experiments
Anomaly Detection
Moving Objects
Low Complexity
Search Algorithm
Person
Trivial
Optimal Solution
Camera
Detector
Trajectory
Motion
Formulation
Experiment

Keywords

  • Event detection
  • action detection
  • dynamic programming
  • max-path search
  • multiple event detection

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition
  • Computational Theory and Mathematics
  • Artificial Intelligence
  • Applied Mathematics

Cite this

Video event detection : From subvolume localization to spatiotemporal path search. / Tran, Du; Yuan, Junsong; Forsyth, David Alexander.

In: IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 36, No. 2, 6567857, 01.02.2014, p. 404-416.

Research output: Contribution to journalArticle

@article{0ff66d6bdb644171aef7a2a9dc905ea5,
title = "Video event detection: From subvolume localization to spatiotemporal path search",
abstract = "Although sliding window-based approaches have been quite successful in detecting objects in images, it is not a trivial problem to extend them to detecting events in videos. We propose to search for spatiotemporal paths for video event detection. This new formulation can accurately detect and locate video events in cluttered and crowded scenes, and is robust to camera motions. It can also well handle the scale, shape, and intraclass variations of the event. Compared to event detection using spatiotemporal sliding windows, the spatiotemporal paths correspond to the event trajectories in the video space, thus can better handle events composed by moving objects. We prove that the proposed search algorithm can achieve the global optimal solution with the lowest complexity. Experiments are conducted on realistic video data sets with different event detection tasks, such as anomaly event detection, walking person detection, and running detection. Our proposed method is compatible with different types of video features or object detectors and robust to false and missed local detections. It significantly improves the overall detection and localization accuracy over the state-of-the-art methods.",
keywords = "Event detection, action detection, dynamic programming, max-path search, multiple event detection",
author = "Du Tran and Junsong Yuan and Forsyth, {David Alexander}",
year = "2014",
month = "2",
day = "1",
doi = "10.1109/TPAMI.2013.137",
language = "English (US)",
volume = "36",
pages = "404--416",
journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",
issn = "0162-8828",
publisher = "IEEE Computer Society",
number = "2",

}

TY - JOUR

T1 - Video event detection

T2 - From subvolume localization to spatiotemporal path search

AU - Tran, Du

AU - Yuan, Junsong

AU - Forsyth, David Alexander

PY - 2014/2/1

Y1 - 2014/2/1

N2 - Although sliding window-based approaches have been quite successful in detecting objects in images, it is not a trivial problem to extend them to detecting events in videos. We propose to search for spatiotemporal paths for video event detection. This new formulation can accurately detect and locate video events in cluttered and crowded scenes, and is robust to camera motions. It can also well handle the scale, shape, and intraclass variations of the event. Compared to event detection using spatiotemporal sliding windows, the spatiotemporal paths correspond to the event trajectories in the video space, thus can better handle events composed by moving objects. We prove that the proposed search algorithm can achieve the global optimal solution with the lowest complexity. Experiments are conducted on realistic video data sets with different event detection tasks, such as anomaly event detection, walking person detection, and running detection. Our proposed method is compatible with different types of video features or object detectors and robust to false and missed local detections. It significantly improves the overall detection and localization accuracy over the state-of-the-art methods.

AB - Although sliding window-based approaches have been quite successful in detecting objects in images, it is not a trivial problem to extend them to detecting events in videos. We propose to search for spatiotemporal paths for video event detection. This new formulation can accurately detect and locate video events in cluttered and crowded scenes, and is robust to camera motions. It can also well handle the scale, shape, and intraclass variations of the event. Compared to event detection using spatiotemporal sliding windows, the spatiotemporal paths correspond to the event trajectories in the video space, thus can better handle events composed by moving objects. We prove that the proposed search algorithm can achieve the global optimal solution with the lowest complexity. Experiments are conducted on realistic video data sets with different event detection tasks, such as anomaly event detection, walking person detection, and running detection. Our proposed method is compatible with different types of video features or object detectors and robust to false and missed local detections. It significantly improves the overall detection and localization accuracy over the state-of-the-art methods.

KW - Event detection

KW - action detection

KW - dynamic programming

KW - max-path search

KW - multiple event detection

UR - http://www.scopus.com/inward/record.url?scp=84891607575&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84891607575&partnerID=8YFLogxK

U2 - 10.1109/TPAMI.2013.137

DO - 10.1109/TPAMI.2013.137

M3 - Article

C2 - 24356358

AN - SCOPUS:84891607575

VL - 36

SP - 404

EP - 416

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

SN - 0162-8828

IS - 2

M1 - 6567857

ER -