ISWC 2006 tutorial, 14 October 2006, 9:00am - 4:30pm

Hands-on Activity Context Recognition
Oliver Amft1, David Bannach2, Kai Kunze2, Paul Lukowicz2
1 Wearable Computing Lab., ETH Zurich, Switzerland
2 Embedded Systems Group, University of Passau, Germany

Quick links: Abstract  Material  Links

Abstract

This workshop will discuss examples and practical methods used for context recognition applications. Participants will have the opportunity to see and experiment with an activity event recognition application.

In the introductory part (morning) industry-relevant application examples out of EU projects are presented. After discussion of some basic concepts, including sensing modalities, recording and annotation we briefly cover practical approaches for sensor data segmentation, event spotting and classification, all relevant for wearable systems.

In the practical part (afternoon) participants will use the Context Recognition Network Toolbox to perform data annotation and classifier training using their own notebooks. The algorithms can be evaluated in an on-line gesture-controlled game challenge. The participant scores are discussed at the end of the workshop.



Material used during the tutorial

The CRN Toolbox is available under GNU LGPL license and can be downloaded/ checked out as indicated in the Wiki. The toolbox configuration used for the gesture recognition is under development and not included yet. Please contact me if you are interested.


Links