Publication Details

Methods for Trajectory Outlier Detection in Surveillance Video

PEŠEK Martin. Methods for Trajectory Outlier Detection in Surveillance Video. In: Proceedings of the 18th Conference STUDENT EEICT 2012. Volume 3. Brno: Brno University of Technology, 2012, pp. 474-478. ISBN 978-80-214-4462-1. Available from: http://www.feec.vutbr.cz/EEICT/2012/sbornik/03doktorskeprojekty/11inteligentnisystemy/03-xpesek07.pdf
Czech title
Metody pro detekci odlehlých trajektorií v dohledovém videu
Type
conference paper
Language
english
Authors
Pešek Martin, Ing. (DIFS FIT BUT)
URL
Keywords

data mining, outlier detection, moving objects data, trajectory, surveillance video, TOP-EYE, GMM

Abstract

Outlier detection in trajectory data extracted from surveillance video is a useful data analysis task, which can reveal anomalous or suspicious behaviour of moving objects. This paper deals with the description, evaluation and comparision of two methods for the trajectory outlier detection. Firstly, TOP-EYE algorithm, which gradually computes and accumulates outlying score of a trajectory. Secondly, a model-based method which uses the Gaussian mixture model (GMM) and identifies trajectories not following the learned model as outliers.

Published
2012
Pages
474-478
Proceedings
Proceedings of the 18th Conference STUDENT EEICT 2012
Series
Volume 3
Conference
Student EEICT 2012, Brno, CZ
ISBN
978-80-214-4462-1
Publisher
Brno University of Technology
Place
Brno, CZ
BibTeX 
@INPROCEEDINGS{FITPUB9930,
   author = "Martin Pe\v{s}ek",
   title = "Methods for Trajectory Outlier Detection in Surveillance Video",
   pages = "474--478",
   booktitle = "Proceedings of the 18th Conference STUDENT EEICT 2012",
   series = "Volume 3",
   year = 2012,
   location = "Brno, CZ",
   publisher = "Brno University of Technology",
   ISBN = "978-80-214-4462-1",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/9930"
}
Back to top