Local appearance models in the neighborhood of salient image features, together with local and/or global geometric constraints, serve as the basis for several recent and effective...
Data is often collected over a distributed network, but in many cases, is so voluminous that it is impractical and undesirable to collect it in a central location. Instead, we mus...
This paper p'VlJOses a /lew meThod for incoherent lIlotion recognitioll from video sequences. We use time-series ~pariQ-lempora/illlensity gradients within a space-rime patch...
This paper introduces a novel approach for the recognition of a wide vocabulary of Arabic handwritten words. Note that there is an essential difference between the global and anal...
In this paper we propose a novel method for generic object localization. The method is based on modeling the object as a graph at two levels: a local substructural representation ...