The course of Advanced Pattern Recognition Systems aims at providing the students with practical software instruments for solving real recognition problems, as those coming from the surveillance/quality-control/automotive/entertainment scenarios. To this sake, the lessons are organized as practical problems, which will be faced by considering theory elements coming from the Pattern Recognition course, and embedding them into MATLAB or C software. It is clear that the Laboratory sessions in this course are fundamental, and will take most of the time.
PROBLEM: Detection and recognition of: people, faces, scenes and objects in general. Associated algorithms and techniques: Generative Learning, Discriminative Learning, Hybrid Learning, Boosting, GIST, SIFT (reminders of), covariances, SURF, SDALF, Bag of Words.
PROBLEM: Modeling of moving objects: tracking of single objects, groups of objects, action recognition, expression recognition. Associated algorithms and techniques: Monte Carlo Methods (Particle Filtering), Online Learning, Spline, Snakes
Project or Seminar