TU Dortmund → Department of Computer Science → LS XII → Pattern Recognition Group → Publications → Publication Details
C. Kleine-Cosack, M. H. Hennecke, S. Vajda and G. A. Fink Ambient Intelligence, pages 157-166, 2010.
Smart environments rely on context classification in order to be able to support users in their daily lives. Therefore, measurements provided by sensors distributed throughout the environment are analyzed. A main drawback of the solutions proposed so far is that the type of sensors and their placement often needs to be specifically adjusted to the problem addressed. Instead, we propose to perform context classification based on the analysis of acoustic events, which can be observed using arrays of microphones. Consequently, the sensor setup can be kept rather general and a wide range of contexts can be discriminated. In an experimental evaluation within a smart conference room we demonstrate the advantages of our new approach.