Several theoretical methods have been developed in the past years to evaluate the generalization ability of a classifier: they provide extremely useful insights on the learning ph...
In this paper we demonstrate that the support vector tracking (SVT) framework first proposed by Avidan is equivalent to the canonical Lucas-Kanade (LK) algorithm with a weighted E...
We present a component-based method and two global methods for face recognition and evaluate them with respect to robustness against pose changes. In the component system we first...
: An explicit formula is presented for reconstructing a This article derives an explicit formula for the integer values finite-support object defined on a lattice of points and tak...
Learning Bayesian Belief Networks (BBN) from corpora and incorporating the extracted inferring knowledge with a Support Vector Machines (SVM) classifier has been applied to charac...