Current page: Members->List of Members->Rainer Malaka
 

Rainer Malaka
European Media Laboratory
Membership Number: 44
Address: Schloss-Wolfsbrunnenweg 33, 69256 Heidelberg, Germany
Email: rainer.malaka@eml.org
Phone: +49 6221 533 206
Fax: +49 6221 522 298
URL: http://www.eml.org

Biographical Sketch
Rainer Malaka

European Media Laboratory (EML) Personal Memory Group
EML, the European Media Laboratory GmbH, is a private research institute for Information Technology (IT) and its applications. EML’s center of interest is to develop and prototypically realize intelligent methods of information processing, which, in turn, hold promise for future utilization. EML’s goal is to transform research results into commercially viable product models or systems. The actual manufacture and marketing of such products are not among its purposes. Nevertheless, EML does work to ensure that its research and development efforts will find real-world commercial applications. It does this by granting licenses and by actively supporting the formation of new and spin-off ventures.

Projects of the EML Personal Memory Group: The Personal Memory Group addresses the major challenge of accessing the information content of heterogeneous and complex unstructured and weakly structured data in a dynamic and context-dependent manner. KTS Project: Deep Map Bmb+f Projects: SmartKom, EMBASSI, GEIST EU 5th: Crumpet, Invite

The subgroup for cognitive image understanding develops architectures for the automatic extraction of knowledge from visual information. Especially in urban environments this knowledge enables better services of mobile devices.

Possible advantages range from better awareness of the users’ locations as well as the ability to detect complex objects in the environment.

Image understanding for an analysis of the visual context has to be done based on single (monocular) picture streams as input. Which, moreover, stems from relative cheap and low-resolution cameras in the mobile devices. A main task of image understanding is of identify and classify the objects in those picture streams. This task is particularly difficult as many of the object features vary greatly, e.g., color, texture illumination and appearance.


Site generated on Friday, 06 January 2006