Event Location: TBA
Bio: Aswin Sankaranarayanan is an Assistant Professor in the ECE Department at Carnegie Mellon University, Pittsburgh, PA. His research interests lie in the areas of computer vision, signal processing, and image and video acquisition. Prof. Sankaranarayanan received his B.Tech in Electrical Engineering from the Indian Institute of Technology, Madras in 2003 and MSc and PhD degrees from the Department of Electrical and Computer Engineering at the University of Maryland, College Park in 2007 and 2009, respectively. He was awarded the Distinguished Dissertation Fellowship by the Dept. of Electrical and Computer Engineering at the University of Maryland in 2009. He was a post-doctoral researcher at Rice University from October 2009 to December 2012.

Abstract: Our fascination with detail is never ending. We have built cameras that capture images with billions of pixels and videos at millions of frames per second. The key enabling technology behind these cameras is the role of Silicon as the sensing material of choice in visible spectra of light. In other modalities, where sensing is inherently costly, sensing at high spatial and temporal resolutions often comes with steep constraints. Two classic examples are sensing in infrared, where sensor materials are expensive, and MRI, where capture time is costly. In such cases, traditional methods for sensing, which advocate sampling faster and with higher resolution, do not scale well.

In this talk, I will outline an emerging method for sensing images and videos that (a) exploits redundancies in real-world images and videos, to (b) sense with far-fewer measurements as compared to a traditional sensor. I will present multiple examples from both my research and others’ that provide insight into how such a sensing strategy works. Special emphasis would be on tradeoffs and sensor modifications required for such designs to work.