Multi-instance multi-label learning (MIML) is a framework for supervised classification where the objects to be classified are bags of instances associated with multiple labels....
Many semi-supervised learning algorithms only
deal with binary classification. Their extension to the
multi-class problem is usually obtained by repeatedly
solving a set of bina...
In the last decade, graph-cut optimization has been popular for a variety of pixel labeling problems. Typically graph-cut methods are used to incorporate a smoothness prior on a l...
Deformable model fitting has been actively pursued in the computer vision
community for over a decade. As a result, numerous approaches have
been proposed with varying degrees of...
Deformable model fitting has been actively pursued in the computer vision community for over a decade. As a result, numerous approaches have been proposed with varying degrees of ...