This paper describes a system that can build appearance models of animals automatically from a video sequence of the relevant animal with no explicit supervisory information. The ...
We propose a new technique for fusing multiple cues to robustly segment an object from its background in video sequences that suffer from abrupt changes of both illumination and p...
Francesc Moreno-Noguer, Alberto Sanfeliu, Dimitris...
We propose an information fusion approach to tracking objects from different viewpoints that can detect and recover from tracking failures. We introduce a reliability measure that...
Broadcasters often reuse video clips while reporting news stories. New clips are added to the story as it develops replacing old clips. A subset of these video clips may be replay...
Jedrzej Z. Miadowicz, John M. Gauch, Abhishek Shiv...
We propose a segmentation-based method of object tracking using image warping and Kalman filtering. The object region is defined to include a group of patches, which are obtained ...