Publication Details
Real-time Tracking of Participants in Meeting Video
Tracking, Head detection, Skin color model, KLT tracker
The aim of our work was to develop a robust real-time trackerof body parts in meeting videos. We have focused mainly on tracking heads inmeeting video data from the AMI project. The AMI project is concerned with thedevelopment of meeting browsers and remote meeting assistants for instrumentedmeeting rooms. Our system is intended to process video data acquired fromcamera online. This sets bounds in the terms of computational cost of usedalgorithms. Therefore we decided first to create a fast solution and analyzeits results in order to tell if further improvements are needed.
We havedivided tracking task into two separate parts. First skin color model, backgroundsubtraction and topological feature based object classification is used todetect heads in video. The detected objects are consequently tracked throughoutthe video sequence by using KLT feature based tracker. Tracking results arestored in the XML format for evaluation.
@inproceedings{BUT28831,
author="Michal {Hradiš} and Roman {Juránek}",
title="Real-time Tracking of Participants in Meeting Video",
booktitle="Proceedings of The 10th Central European Seminar on Computer Graphics",
year="2006",
pages="89--93",
address="Budměřice",
url="http://www.fit.vutbr.cz/research/groups/graph/publi/2006/2006-Hradis-CESCG-HeadTracking.pdf"
}