Human Detection via Classification on Riemannian Manifolds

Oncel Tuzel (1,3) Fatih Porikli (3) and Peter Meer (1,2)

(1)Department of Computer Science
(2)Department of Electrical and Computer Engineering
Rutgers University, Piscataway, NJ 08854, USA

(3) Mitsubishi Electric Research Laboratories
Cambridge, MA 02139

We present a new algorithm to detect humans in still images utilizing covariance matrices as object descriptors. Since these descriptors do not lie on a vector space, well known machine learning techniques are not adequate to learn the classifiers. The space of d-dimensional nonsingular covariance matrices can be represented as a connected Riemannian manifold. We present a novel approach for classifying points lying on a Riemannian manifold by incorporating the a priori information about the geometry of the space. The algorithm is tested on INRIA human database where superior detection rates are observed over the previous approaches.

2007 Computer Vision and Pattern Recognition Conference, Minneapolis, Minnesota, June 2007.
Return to Research: Content-based retrieval        Return to List of Publications
Download the paper