Differences

This shows you the differences between the selected revision and the current version of the page.

projects:gems 2012/03/30 05:09 projects:gems 2018/04/18 21:08 current
Line 1: Line 1:
 +
Line 4: Line 5:
Emergency Management relies on timely and sufficient information. During an emergency, the efficient display of large amounts of information in a central command room is a vital requirement The purpose of this project was to develop human-computer interaction methods for such environments. Emergency Management relies on timely and sufficient information. During an emergency, the efficient display of large amounts of information in a central command room is a vital requirement The purpose of this project was to develop human-computer interaction methods for such environments.
We have concentrated on three subjects in detail: We have concentrated on three subjects in detail:
- * Face analysis, +  * Face analysis, 
- * Body analysis, + * Body analysis, 
- * Hand gesture analysis.+ * Hand gesture analysis.
For face analysis, we have worked on both face recognition and facial expression analysis. For body gesture analysis, one needs to track and analyse body movements. For hand gesture recognition, we have used color and depth sensors to track human bodies in general environments. Hand gestures consist of two main components: the hand shape and its trajectory.  Therefore, the signal is inherently a multimodal time series signal and techniques used to represent time series are suited for this application. We have worked on shape classification and time series analysis for gesture classification. We have used probabilistic methods to a fit a 3D skeletal model of the body and hand. We have generated systems to recognize gestures in real time. For face analysis, we have worked on both face recognition and facial expression analysis. For body gesture analysis, one needs to track and analyse body movements. For hand gesture recognition, we have used color and depth sensors to track human bodies in general environments. Hand gestures consist of two main components: the hand shape and its trajectory.  Therefore, the signal is inherently a multimodal time series signal and techniques used to represent time series are suited for this application. We have worked on shape classification and time series analysis for gesture classification. We have used probabilistic methods to a fit a 3D skeletal model of the body and hand. We have generated systems to recognize gestures in real time.