@InProceedings{Sixta-CVWW13, IS = { zkontrolovano 23 Jan 2014 }, UPDATE = { 2013-01-25 }, author = {Sixta, Tom{\'a}{\v s}}, title = {Star Convex Object Detection by the Infinite Shape Mixture Model}, year = {2013}, pages = {2-8}, booktitle = {CVWW 2013: Proceedings of the 18th Computer Vision Winter Workshop}, publisher = {Vienna University of Technology}, address = {Karlsplatz 13, Vienna, Austria}, editor = {Kropatsch, Walter G. and Ramachandran, Geetha and Torres, Fuensanta}, book_pages = {7}, isbn = {978-3-200-02943-9}, month = {February}, day = {4-6}, venue = {Hernstein, Austria}, annote = {Shape is an important feature of many object categories. In this paper we propose a Bayesian framework for detection of unknown number of objects based on their shape. The task is formulated as a minimization of Bayesian risk. The loss function is designed in such a way that the number of objects need not to be known or even bounded. We introduce a probability distribution over object states (number of objects and their poses) called Infinite Shape Mixture Model which is a modification of Rasmussen's Infinite Gaussian Mixture Model. Conditional posterior distributions are derived for all parameters of the model in order to make the inference feasible. Performance of the model is tested on two brief experiments.}, keywords = {Object detection, Bayesian inference}, project = {GACR P202/12/2071}, }