Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Pottics - the potts topic model for semantic image segmentation

: Dann, Christoph


Pinz, Axel (Ed.):
Pattern recognition. Joint 34th DAGM and 36th OAGM symposium 2012 : Graz, Austria, August 28 - 31, 2012; proceedings
Berlin: Springer, 2012 (Lecture Notes in Computer Science 7476)
ISBN: 978-3-642-32716-2
ISBN: 3-642-32716-8
ISBN: 978-3-642-32717-9
ISSN: 0302-9743
German Association for Pattern Recognition (DAGM Symposium) <34, 2012, Graz>
Austrian Association for Pattern Recognition (OAGM Symposium) <36, 2012, Graz>
Fraunhofer IGD ()
image processing; image segmentation; semantic labeling; graph representation; Forschungsgruppe Visual Inference (VINF)

We present a novel conditional random field (CRF) for semantic segmentation that extends the common Potts model of spatial coherency with latent topics, which capture higher-order spatial relations of segment labels. Specifically, we show how recent approaches for producing sets of figure-ground segmentations can be leveraged to construct a suitable graph representation for this task. The CRF model incorporates such proposal segmentations as topics, modelling the joint occurrence or absence of object classes. The resulting model is trained using a structured large margin approach with latent variables. Experimental results on the challenging VOC'10 dataset demonstrate significant performance improvements over simpler models with less spatial structure.