The presentation is focusing on the research conducted on semantic information extraction from events. The research focuses on exploring techniques used for human behavior recognition from video sequences and creating complete and solid systems for anomaly detection observed in the video, human behavior recognition from video sequences and creating complete and solid systems for anomaly detection observed in the video, human behavior recognition and video indexing. Behavior understanding from events can be considered as a typical classification problem using predefined classes and modified classifiers for the recognition step. In the current presentation we present methods to solve this classification problem from a bottom up point of view using motion analysis for behavior understanding. The first part of our presentation includes preprocessing steps such as background subtraction for foreground pixels detection, tracking, etc. After the theoretical analysis of the problem and the presentation of our early conclusions, novel methods are proposed for behavior classification and semantic representation from events. Furthermore, a bottom-up approach for human behavior understanding is presented, using a multi camera system for anomaly detection. The experimental results of our research are also presented and future work is discussed.
Talk slides in pdf [~7,2MB]http://www.iit.demokritos.gr/docs/seminars/iptseminars-2010-antonakaki.pdf