My research is in the area of computer vision and computer graphics, especially at the intersection of the two. I am particularly interested in using data-driven techniques to tackle problems which are very hard to model parametrically but where large quantities of data are readily available. The ultimate goal is to use the ever-growing amount of stored visual information (digital photo albums, webcams, movies, etc.) to learn, understand, and resynthesize the visual world around us.
In very broad strokes, here are the main current themes of my research:
Qualitative 3D Reasoning for Image Interpretation: The ability to see and understand the three-dimensional world behind a two-dimensional image goes to the very heart of the computer vision problem. The overall objective of this research effort is, given a single image, to automatically produce a coherent interpretation of the depicted scene. On one level, such interpretation should include opportunistically recognizing known objects (e.g. people, houses, cars, trees) and known materials (e.g. grass, sand, rock, foliage) as well as their rough positions and orientations within the scene. But more than that, the goal is to capture the overall qualitative sense of the scene.
"Brute-forcing" Vision: What could you do with a billion images? Taking inspiration from Google -- the A.I. for the post-modern world -- we want to utilize the huge amount of existing visual data to "look-up" similar images as a cue to interpreting a previously unseen photograph. That is, we would like to sample from the entire space of scenes as a way of exhaustively modeling our visual world? If this works, it might allow us to "brute force" many currently unsolvable vision and graphics problems!
Understanding (and Faking) Visual Realism: Why is it that most of computer-generated imagery doesn't look very realistic? What is it that the Renaissance artists knew that we don't? Which bits of the visual experience is it important to "get right", and which could be safely faked without anyone noticing? The ultimate goal is to make synthesized images appear as real and convincing as regular photographs.