International Journal of Computer Vision

, Volume 88, Issue 2, pp 169–188

Gaussian Processes for Object Categorization

Authors

    • Microsoft Research
  • Kristen Grauman
    • University of Texas at Austin
  • Raquel Urtasun
    • UC Berkeley EECS & ICSI
  • Trevor Darrell
    • UC Berkeley EECS & ICSI
Open AccessArticle

DOI: 10.1007/s11263-009-0268-3

Cite this article as:
Kapoor, A., Grauman, K., Urtasun, R. et al. Int J Comput Vis (2010) 88: 169. doi:10.1007/s11263-009-0268-3

Abstract

Discriminative methods for visual object category recognition are typically non-probabilistic, predicting class labels but not directly providing an estimate of uncertainty. Gaussian Processes (GPs) provide a framework for deriving regression techniques with explicit uncertainty models; we show here how Gaussian Processes with covariance functions defined based on a Pyramid Match Kernel (PMK) can be used for probabilistic object category recognition. Our probabilistic formulation provides a principled way to learn hyperparameters, which we utilize to learn an optimal combination of multiple covariance functions. It also offers confidence estimates at test points, and naturally allows for an active learning paradigm in which points are optimally selected for interactive labeling. We show that with an appropriate combination of kernels a significant boost in classification performance is possible. Further, our experiments indicate the utility of active learning with probabilistic predictive models, especially when the amount of training data labels that may be sought for a category is ultimately very small.

Keywords

Object recognition Gaussian process Kernel combination Active learning

Copyright information

© The Author(s) 2009