Speaker: Achuta Kadambi
Affiliation: Research Scientist, Harvard University
Abstract: Understanding how light travels through scenes can transform autonomous driving, medical imaging and consumer photography. Unfortunately, this understanding has proved difficult to achieve: trillions of possible light paths are sensed by just millions of pixels. This inherent curse of dimensionality has imposed limits on existing imaging systems. I aim to surmount this barrier by expanding a theoretical framework known as computational light transport, which models complex interactions between light and matter in a tractable space.
In this talk, I will underscore the practical implications of computational light transport. First, I will describe a solution to a previously under-constrained problem in polarimetric vision that enables 3D imaging at unprecedented quality. I will then touch upon advances in time of flight imaging, to image objects outside the line of sight or behind scattering media. Looking ahead, this computational approach to imaging is the start of a broader trend where images will be computed rather than captured.
Bio: Achuta Kadambi received his PhD from the Massachusetts Institute of Technology in December 2017 and joined the research faculty at Harvard University. He studies imaging from the perspective of optics and computer science. Achuta’s publications have been presented at ICCV, CVPR, SIGGRAPH, SIGGRAPH Asia, and ICCP. His doctoral work has resulted in over a dozen (co-authored) patent applications, a (co-authored) textbook in MIT Press, the Lemelson-MIT Student Prize, the Qualcomm Innovation Fellowship, and the Draper Fellowship.
For further information, please contact Prof. Abeer Alwan ()
Date(s) - Feb 13, 2018
11:00 am - 12:30 pm
EE-IV Shannon Room #54-134
420 Westwood Plaza - 5th Flr., Los Angeles CA 90095