TITLE: Visual Learning of Categories using Epitome

PRESENTER: Katarina Mele

AFFILIATION: School of Computer Science & Engineering and National ICT Australia

DATE: Friday 15th October 2004

TIME: 12 noon - 1 pm

PLACE: CSE K17 1st Floor Seminar Room

ABSTRACT:

In the presentation the problem of appearance based classification methods will be addressed. The proposed presentation of a class is inspired by a generative statistical model called "epitome". The epitome of an image is its miniature, condensed version containing the essence of the textural and shape properties of an image. In order to obtain a model for classification geometrical information is added into the epitome. The applicability of such presentation to object classification will be discussed.

BIOGRAPHY OF SPEAKER:

Katarina Mele received her Master's degree in computer science from the University of Ljubljana, Faculty of Computer and Information Science in 2003. The title of her M.Sc. is "Objects Recognition on Cluttered Backgrounds". Currently she is a visiting student at the UNSW where she is going to continue with the work on her PhD thesis. Her main research includes appearance based vision, object recognition and classification, pattern recognition and machine learning.

HOST

Associate Professor Sowmya Arcot
School of Computer Science and Engineering
University of New South Wales
somya"at"cse.unsw.edu.au
+61 2 9385 6933 (Internal: x56933)

SEMINART CONVENER:

Bernhard Hengst
National ICT Australia, Symbolic Machine Learning & Knowledge Acquisition
Tel: +61-2-9385-6940