Apr 17, 2013 - CVPR Poster
Jason Chang and Donglai Wei's work on "A Video Representation Using Temporal Superpixels" is accepted to the annual Computer Vision and Pattern Recognition conference.
Our group focuses on the analysis of complex, high-dimensional data. We
combine elements of Bayesian inference, information theory,
optimization, and physical sensor models to develop scalable algorithms
with theoretical performance guarantees. Application areas include
multi-modal data fusion, distributed inference under resource
constraints, structural inference, resource management in sensor
networks, and analysis of video, seismic volumes, and radar images.