Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

A shape-based object class model for knowledge transfer

: Stark, Michael; Goesele, Michael; Schiele, Bernt


Institute of Electrical and Electronics Engineers -IEEE-:
IEEE 12th International Conference on Computer Vision, ICCV 2009. DVD : September 27 - October 4, 2009, Kyoto, Japan
Los Alamitos/Calif.: IEEE Computer Society, 2009
ISBN: 978-1-4244-4419-9
International Conference on Computer Vision (ICCV) <12, 2009, Kyoto>
Conference Paper
Fraunhofer IGD ()
object class detection; Markov Chain; Monte Carlo method

Object class models trained on hundreds or thousands of images have shown to enable robust detection. Transferring knowledge from such models to new object classes trained from a few or even as little as one training instance however is still in its infancy. This paper designs a shape-based model that allows to easily and explicitly transfer knowledge on three different levels: transfer of individual parts' shape and appearance information, transfer of local symmetry between parts, and transfer of part topology. Due to the factorized form of the model, knowledge can either be transferred for the complete model or just partial knowledge corresponding to certain aspects of the model. The experiments clearly demonstrate that the proposed model is competitive with the state-of-the-art and enables both full and partial knowledge transfer.