Jeremy Klotz

jklotz [at] cs.columbia.edu

I'm a fourth-year Ph.D. student at Columbia University. I am advised by Shree Nayar, and my work is generously funded by the NDSEG fellowship. Before my Ph.D., I worked with Aswin Sankaranarayanan at CMU.

My research is in computational imaging. Specifically, I work on visual sensing methods that capture the least information neccessary to solve a task. My work demonstrates that a minimalist approach to visual sensing brings a variety of benefits over conventional cameras, such as preserving privacy and enabling self-powered cameras. Recently, my research has taken a foray into imaging and vision techniques for solar energy. Before coming to Columbia, I worked on structured light systems for 3D scanning and predicting fruit freshness.

Resume / Google Scholar / LinkedIn

Research

We introduce a minimal sensing method to iteratively orient a solar panel toward the optimal orientation in any lighting environment, regardless of the complexity of the illumination.

UrbanSky consists of 1,067 outdoor HDR lighting environments in New York City. At each location, we capture the illumination using a 360° camera, we measure the global horizontal irradiance, and we record the capture date, time of day, location, and current weather conditions.

A minimalist camera captures the smallest number of measurements needed to solve a task. Rather than using square pixels, a minimalist camera uses freeform pixels whose shape are automatically learned for the task at-hand.

Cricket is a novel light sensor that can measure light without the use of a power supply or a battery.

We introduce a low-bandwidth method for 3D scanning with position sensing diodes that is robust to global illumination.

We show subsurface scattering measurements are useful for predicting fruit freshness.

Website Template