Learning a knowledge base of ontological concepts for high-level scene interpretation

Ontological concept descriptions of scene objects and aggregates play an essential role in model-based scene interpretation. An aggregate specifies a set of objects with certain properties and relations which together constitute a meaningful scene entity. In this paper we show how ontological concept descriptions for spatially related objects and aggregates can be learnt from positive and negative examples. Our approach features a rich representation language encompassing quantitative and qualitative attributes and relations. Using examples from the buildings domain, we show that learnt aggregate concepts for window arrays, balconies and other structures can be successfully employed in the conceptual knowledge base of a scene interpretation system. Furthermore we argue that our approach can be extended to cover ontological concepts of any kind, with very few restrictions.