TR2007-046

Human Detection via Classification on Riemannian Manifolds


Abstract:

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.

 

  • Related News & Events

    •  NEWS    CVPR 2007: 3 publications by Oncel Tuzel, Amit Agrawal and Ramesh Raskar
      Date: June 17, 2007
      Where: IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
      Research Area: Computer Vision
      Brief
      • The papers "Resolving Objects at Higher Resolution from a Single Motion-blurred Image" by Agrawal, A. and Raskar, R., "Human Detection via Classification on Riemannian Manifolds" by Tuzel, O., Porikli, F. and Meer, P. and "Statistics of Infrared Images" by Morris, N., Avidan, S., Matusik, W. and Pfister, H. were presented at the IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
    •  
    •  AWARD    CVPR 2007 Best Paper Runner Up of 1,300 submitted
      Date: January 1, 2007
      Awarded to: Oncel Tuzel, Fatih Porikli and Peter Meer
      Awarded for: "Human Detection via Classification of Riemannian Manifolds"
      Awarded by: IEEE Computer Vision and Pattern Recognition (CVPR)
      Research Area: Machine Learning
    •