• Image parsing, segmentation, and region classification

• Context representations in image analysis 

• Structured prediction models

• Learning in undirected graphical models,  including feature induction and structure learning

• Motion analysis and synthesis
  

Research Projects

• Image Labeling based on Conditional Random Fields

• Combining image segmentation with region classification;

• Utilizing context information from multiple levels.

• Appearance based Localization

  • Learning prototypical features for localization;

  • Using side information to learn the image manifold.

Publications

Xuming He, and Richard S. Zemel
Latent Topic Random Fields: Learning using a taxonomy of labels. (pdf)
To appear in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2008

Liam Stewart, Xuming He, and Richard S. Zemel
Learning Flexible Features for Conditional Random Fields. (pdf)
To appear in IEEE Transactions on PAMI, 2008

Xuming He
Learning Structured Prediction Models for Image Labeling. (pdf)
PhD thesis. DCS, University of Toronto, 2007

Xuming He, Richard Zemel, and Volodymyr Mnih
Topological Map Learning from Outdoor Image Sequences. (Tech-Report version: pdf)
In Journal of Field Robotics, Volume 23, Issue 11-12 , P. 1091 - 1104, 2007

Xuming He, Richard Zemel, and Deb Ray
Learning and incorporating top-down cues in image segmentation. (pdf)
In the 9th European Conference on Computer Vision (ECCV), 2006.

Xuming He, Richard Zemel, and Volodymyr Mnih
Learning Landmarks for Localization via Manifolds.
In NIPS Workshop on Machine Learning Based Robotics in Unstructured Environments, 2005

Xuming He, Richard Zemel, and Miguel Carreira-Perpinan
Multiscale conditional random fields for image labelling. (pdf)
In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2004

Last updated by Feb, 2008