Dynamic Vision for Control
Abstract
We have developed a comprehensive set of analytical and computational tools to exploit visual data for the purpose of control and interaction with complex, dynamic and uncertain environments. The accomplishment of the goals set forth in the original proposal was articulated into three parallel research tracks. (1) Tracking; focused on the establishment of correspondence of low-level statistics across temporal samples, including the development of representations that are invariant to local illumination changes, co-variant with respect to finite-dimensional group transformations, and insensitive to non-invertible transformations due to non-group deformations, partial occlusions etc.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 18, 2012
- Accession Number
- ADA567121
Entities
People
- Stefano Soatto
Organizations
- University of California, Los Angeles