Publication Details
Brno University of Technology at TRECVid 2013: Interactive Surveillance Event Detection
Mlích Jozef, Ing. (DCGM FIT BUT)
Pešek Martin, Ing. (DIFS FIT BUT)
Volf Tomáš, Ing. (DIFS FIT BUT)
Zemčík Pavel, prof. Dr. Ing. (DCGM FIT BUT)
Zendulka Jaroslav, doc. Ing., CSc. (DIFS FIT BUT)
In the paper, we describe our experiments in the interactive surveillance event detection pilot (SED) of the 2009 TRECVid evaluation. Our approach inherits functionality of the Surveillance Network Augmented by Retrieval (SUNAR) system, which is an information retrieval based wide area (video) surveillance system being developed at Faculty of IT, Brno University of Technology. It contains both standard and experimental techniques evaluated at the AVSS 2009/10 Multi-Camera Tracking Challenge. We have deployed active learning functionality (Bayesian, SVM and HMM) based on moving objects' trajectory statistics and shape classification using Video Terror Application Programming Interface (VTApi), which was created to unify and accelerate the intelligent vision applications development.
@INPROCEEDINGS{FITPUB10485,
author = "Petr Chmela\v{r} and Jozef Ml\'{i}ch and Martin Pe\v{s}ek and Tom\'{a}\v{s} Volf and Pavel Zem\v{c}\'{i}k and Jaroslav Zendulka",
title = "Brno University of Technology at TRECVid 2013: Interactive Surveillance Event Detection",
pages = "1--9",
booktitle = "2013 TREC Video Retrieval Evaluation Notebook Papers and Slides",
year = 2013,
location = "Gaithersburg, US",
language = "english",
url = "https://www.fit.vut.cz/research/publication/10485"
}