Faculty of Informatics Vienna University of Technology Institute of Computer Aided Automation PRIP Home PRIP Home
Personal tools
You are here: Home Teaching PRIP-Talks Andreas Zweng: Unexpected Human Behavior Recognition in Image Sequences
Navigation
 

Andreas Zweng: Unexpected Human Behavior Recognition in Image Sequences

Abschlussvortrag DA

What
  • Presentation
When Mar 04, 2010
from 02:15 pm to 02:15 pm
Where Sem 183/2
Add event to calendar vCal
iCal

Abstract:

This master thesis discusses image processing methods for detecting abnormal and unexpected behavior within human crowds and the drawbacks of existing methods. Currently abnormal behavior recognition is treated as a high-level image processing task relying on existing visual surveillance methods such as object-tracking or event detection. Therefore, the result of behavior recognition depends on the quality of the output of these methods and is affected by their weaknesses. This thesis discusses a new approach which is independent from high-level methods like event detection or object-tracking and is therefore more robust in complex scenarios. This approach uses features including the spatio-temporal movement of crowds, the pace of the included persons and the density of the crowds. Training sequences are used to detect values for these features under normal circumstances, which are then used to detect deviations during an unexpected event. The results of the proposed approach are discussed and compared with existing methods.

Document Actions