Mark Sheinin

I'm a Post-doctoral Research Associate at Carnegie Mellon University's Robotic Institute at the Illumination and Imaging Laboratory

Prior to that, I completed my Ph.D. (direct track) at the Technion -- Israeli Institute of Technology under the supervision of Prof. Yoav Schechner. I completed my M.Sc. at Technion's Viterbi Faculty of Electrical Engineering.

 

Email:       marksheinin [at] gmail [dot] com

Office:      115 Smith Hall, The Robotics Institute

 

 

 

Informal research statement

My research interests lie at the intersection of computer vision and computational photography. Computer vision relies on images and video captured by cameras. Since the discovery of the camera obscura in ancient times and up to the modern electronic devices, the cameras we use have not changed much. Sure, we digitized the image and improved the optics but the underlying principle behind the modern camera remains the same: it gathers light from scene points and focuses them on a 2D sensing plane.

 

But why should settle for that? Yes, we can train on millions of color images to infer scene depth, or we can just build a camera that senses color and depth simultaneously. And why stop at depth? One can imagine a whole slew of hidden scene properties to add to our "cameras" including material properties, various light components (direct and indirect), temperature, tiny motion variations, hyperspectral information, very fast motion, and more. As one of my mentors told me once "the goal of computational imaging is to sense the invisible" (beyond human vision).

If "photography" (φωτο-γραφία) can be translated from Greek as "drawing with light", then computational photography should then be "photo-codikopoisi" (φωτο-κωδικοποίηση) or "coding with light", or perhaps "photo-logismos" (φωτο-λογισμός) which means calculating or reasoning with light*.

 

 

 

* Thanks to Kyros Kutulakos for advice with the proper Greek wordplay on the phrase "coding with light".