Auto-Context and Its Applications

  • Zhuowen Tu | Laboratory of Neuro Imaging

The notion of using context information for solving the high-level vision and medical imaging problems has been increasingly realized in the field.
The current literature using Markov Random Fields (MRFs) and Conditional Random Fields (CRFs) often involves specific algorithm design, in which the modeling and computing stages are studied in isolation. In this talk, I will present an auto-context algorithm. Auto-context learns an integrated low-level and context model, and is very general and easy to implement. It selects and fuses a large number of low-level appearance features, with implicit context and shape information, through discriminative models.
Under nearly an identical parameter setting in training, we apply the algorithm on three challenging vision applications: object segmentation, human body configuration, and scene region labeling. Moreover, context also plays a very important role in medical/brain images where the anatomical structures are mostly positioned and constrained.

In the second part of my talk, I will present various new learning approaches for modeling, shape matching, and object detection.

Speaker Details

Zhuowen Tu is an assistant professor in the laboratory of neuro imaging (LONI), Department of Neurology, and also an assistant professor in the Department of Computer Science, UCLA. He received his PhD from the Ohio State University and his M.E. from Tsinghua University. Before joining the LONI, Zhuowen Tu was a member of technical staff at Siemens Corporate Research and a Post-doctoral fellow in the Department of Statistics at UCLA. Zhuowen Tu was awarded with the David Marr prize in 2003. He was recently given the first prize on a grand challenge for brain image segmentation in 2007.

Watch Next