Research Overview

My research interests lie in the areas of computer vision as well as machine learning and pattern recognition. I am particularly interested in statistical models and inference algorithms that relate to problems of visual inference.

My recent research has focused on probabilistic models for low-level vision, particularly on higher-order Markov random fields. The Fields of Experts model that I developed can be used for various image reconstruction tasks such as image denoising and image inpainting. Beyond this, I am interested in how related models can be used as priors for optical flow and scene depth. Aside from applications of this model, I am working on better and more efficient inference techniques for such models.

In the past I have also been working with Intel Research, particularly with the Computational Nanovision research group. My work focused on improving image denoising techniques that are related to non-linear diffusion filtering.

Publications

Dissertation

High-Order Markov Random Fields for Low-Level Vision. Ph.D. Dissertation, Brown University, May 2007. [pdf, 14.42MB], [ps, 26.16MB], [ps.gz, 13.76MB], [abstract]

Journal Papers

Stefan Roth and Michael J. Black: On the spatial statistics of optical flow. International Journal of Computer Vision (IJCV), 74(1):33–50, August 2007. [Publisher pdf], [Preprint pdf, 1.8MB], [Preprint ps, 9.2MB], [Preprint ps.gz, 3.5MB], [abstract]

Christian Schellewald, Stefan Roth, and Christoph Schnörr: Evaluation of a convex relaxation approach to the quadratic assignment of relational object views. Image and Vision Computing (IVC), 25(8):1301-1314, August 2007. To appear. [Publisher pdf], [abstract]

Conference Papers

Stefan Roth and Michael J. Black: Steerable Random Fields . In Proc. of the IEEE International Conference on Computer Vision (ICCV), Oct. 2007. To appear. [pdf, 1.95MB], [ps, 6.03MB], [ps.gz, 1.28MB], [abstract]

Simon Baker, Daniel Scharstein, J.P. Lewis, Stefan Roth, Michael J. Black, and Richard Szeliski: A Database and Evaluation Methodology for Optical Flow. In Proc. of the IEEE International Conference on Computer Vision (ICCV), Oct. 2007. To appear. [pdf, 1.70MB], [abstract], [benchmark site]

Teodor M. Moldovan, Stefan Roth, and Michael J. Black. Denoising archival films using a learned Bayesian model. In Proc. of IEEE International Conference on Image Processing (ICIP), pp. 2641-2644, Oct. 2006. [pdf, 761kB], [ps, 2.0MB], [ps.gz, 996kB], [abstract]

Stefan Roth and Michael J. Black. Specular Flow and the Recovery of Surface Structure. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), vol. 2, pp. 1869-1876, June 2006. [pdf, 681kB], [ps, 1.96MB], [ps.gz, 722kB], [abstract]

Xiangyang Lan, Stefan Roth, Daniel P. Huttenlocher, and Michael J. Black. Efficient belief propagation with learned higher-order Markov random fields. In A. Leonardis, H. Bischof, and A. Prinz, eds., Proc. of the European Conference on Computer Vision (ECCV), Volume 2, LNCS 3952, pp. 269-282, Springer Verlag, 2006. [pdf, 704kB], [ps, 1.74MB], [ps.gz, 1.01MB], [abstract]

Frank Wood, Stefan Roth, and Michael J. Black: Modeling Neural Population Spiking Activity with Gibbs Distributions. In Y. Weiss and B. Schölkopf and J. Platt, eds., Advances in Neural Information Processing Systems 18, pp. 1539-1546, 2006. [pdf, 104kB], [ps, 978kB], [ps.gz, 469kB], [abstract]

Stefan Roth and Michael J. Black: On the Spatial Statistics of Optical Flow. In Proc. of the IEEE International Conference on Computer Vision (ICCV), vol. 1, pp. 42-49, Oct. 2005. (Honorable Mention for the Marr Prize). [pdf, 1.40MB], [ps, 5.76MB], [ps.gz, 1.71MB], [abstract]

Stefan Roth and Michael J. Black: Fields of Experts: A Framework for Learning Image Priors. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), vol. 2, pp. 860-867, June 2005. [pdf, 493kB], [ps, 4.88MB], [ps.gz, 3.12MB], [abstract]

Stefan Roth, Leonid Sigal, and Michael J. Black: Gibbs Likelihoods for Bayesian Tracking. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), vol. 1, pp. 886-893, June 2004. [pdf, 327kB], [ps, 2.46MB], [ps.gz, 1.38MB], [abstract]

Leonid Sigal, Sidharth Bhatia, Stefan Roth, Michael J. Black, and Michael Isard: Tracking Loose-limbed People. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), vol. 1, pp. 421-428, June 2004. [pdf, 716kB], [abstract]

Christian Schellewald, Stefan Roth, and Christoph Schnörr: Evaluation of Convex Optimization Techniques for the Weighted Graph-Matching Problem in Computer Vision. In B. Radig, S. Florczyk, eds., Proc. of the 23rd DAGM-Symposium, LNCS vol. 2191, pp. 361-368, Springer, 2001. [pdf, 185kB], [ps, 505kB], [ps.gz, 130kB], [abstract], [bibTeX]

H.-J. Bender, R. Männer, C. Poliwoda, S. Roth, and M. Walz: Reconstruction of 3D Catheter Paths from 2D X-Ray Projections. In C. Taylor and A. Colchester, eds., Proc. MICCAI, LNCS vol. 1679, pp. 981–989, Springer, 1999.

Journal Papers in Submission or in Preparation

Stefan Roth and Michael J. Black: Fields of experts (working title). IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI). In preparation.

Undergraduate Thesis

Stefan Roth: Analysis of a Deterministic Annealing Method for Graph Matching and Quadratic Assignment Problems in Computer Vision. Diploma Thesis, University of Mannheim, May 2001.

Technical Reports

Christian Schellewald, Stefan Roth, and Christoph Schnörr: Performance Evaluation of a Convex Relaxation Approach to the Quadratic Assignment of Relational Object Views. Technical Report 2/2002, University of Mannheim, Computer Science Series, February 2002. [abstract]

Refereed Abstracts

Michael J. Black and Stefan Roth: On the Receptive Fields of Markov Random Fields. Cosyne 2005. [abstract]

Stefan Roth, Fulvio Domini, and Michael J. Black: Specular Flow and the Perception of Surface Reflectance. Journal of Vision, 3(9): 413a, 2003. Presented at VisionSciences 2003. [ppt, 1.43MB], [abstract]

Copyright Notice

All materials are presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors and other holders (publishers). These works may not be reposted without the explicit permission of the copyright holder.