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 tracker of body parts in meeting videos. We have focused mainly on tracking heads in meeting video data from the AMI project. The AMI project is concerned with the development of meeting browsers and remote meeting assistants for instrumented meeting rooms. Our system is intended to process video data acquired from camera online. This sets bounds in the terms of computational cost of used algorithms. Therefore we decided first to create a fast solution and analyze its results in order to tell if further improvements are needed.
We have divided tracking task into two separate parts. First skin color model, background subtraction and topological feature based object classification is used to detect heads in video. The detected objects are consequently tracked throughout the video sequence by using KLT feature based tracker. Tracking results are stored in the XML format for evaluation.
@INPROCEEDINGS{FITPUB8427, author = "Michal Hradi\v{s} and Roman Jur\'{a}nek", title = "Real-time Tracking of Participants in Meeting Video", pages = 5, booktitle = "Proceedings of The 10th Central European Seminar on Computer Graphics", year = 2006, location = "Budm\v{e}\v{r}ice, SK", language = "english", url = "https://www.fit.vut.cz/research/publication/8427" }