Results for CS 331B: Representation Learning in Computer Vision

This course surveys recent developments in representation learning that are relevant to visual recognition and understanding tasks. In particular we will examine: 1) why representations matter; 2) classical and moderns methods of forming and learning representations in 2D and 3D computer vision; 3) how to close the loop between sensing and action for perception robotics; 4) how to connect visual-based representations with language; 5) methods for analyzing and visualizing representations. In addition to regular lectures and talks by invited speakers, we will read advanced papers on this topic, and carry out in-depth discussions of these papers as well as the students' own research projects.