My research interests lie at the intersection of visual recognition and robotics and how interaction with the physical world can aid in understanding visual input. To that end, I'm interested in designing recognition algorithms that are (1) minimally supervised by human-generated labels, (2) flexible enough to learn from a few or many examples, and (3) work well in an online setting. Previously, I have worked on a method for semantic segmentation in video using dyanmic CNN architectures as well as domain adaptation for segmentation and detection models. In undergrad, I contributed to a project applying semi-supervised learning techniques to historical photographs to discover trends in fashion and hairstyle over the past century.
Clockwork Convnets for Video Semantic Segmentation
Evan Shelhamer*, Kate Rakelly*, Judy Hoffman*, Trevor Darrell
Video Semantic Segmentation Workshop at European Conference in Computer Vision (ECCV), 2016
We propose a video recognition framework that relies on
two key observations: 1) while pixels may change rapidly from frame to frame,
the semantic content of a scene evolves more slowly, and 2) execution can be
viewed as an aspect of architecture, yielding purpose-fit computation schedules
for networks. We define a novel family of "clockwork" convnets driven by fixed
or adaptive clock signals that schedule the processing of different layers at different
update rates according to their semantic stability.
CS70 - Summer 2014 (Teaching Assistant)
Discrete Mathematics for Computer Science covers proof techniques, modular arithmetic, polynomials, and probability.
EE40 - Summer 2013 (Teaching Assistant)
Introduction to Circuits covers analyzing, designing, and building electronic circuits using op amps and passive components. (Note this class along with EE20 have been replaced by the EE16A/B series as of Fall 2015.)