Background Image

Natasha Banerjee

In this Section

Assistant Professor
Computer Science
357 Science Center


Clarkson University
Box 5815
Potsdam, NY 13699-5815
Phone: 315/268-3831
E-mail: nbanerje@clarkson.edu
Website:

http://www.natasha-banerjee.com/

RESEARCH INTERESTS

My research lies at the intersection of computer graphics, computer vision, and big data. In particular, I use large datasets of images, videos, and three-dimensional information to solve underconstrained problems in computer vision and graphics. Here are some of the areas I am looking at:

IMAGE MANIPULATIONS
Photo-manipulation in 3D has allowed everyday users to perform drastic transforms to their images. Yet, the task of providing interactive full-range control in images while representing the diversity of the real world---the glisten of dew drops on a rose, the dirt and rust on a fifteen year-old bike---is by no means complete. I am interested in leveraging large datasets of 3D models and environment illuminations with fast cloud-based computations to provide full-range near real-time 3D control over objects in images. On the flip side, as we get closer to representing the real world in photographs, people find the task of perceptually distinguishing real and altered images challenging. I am interested in performing research on using physical inaccuracies in altered images or statistical differences between real and altered images to tell them apart.

UNDERSTANDING HUMAN-OBJECT INTERACTIONS
Humans have a lifeftime of interacting with objects, which allows them to intuit the shape, appearance, and behaviors of objects, as well as interactions with objects without having complete sensory information about the objects. I am interested in building and analyzing datasets of 3D models on human-object interactions. I am also interested in leveraging the analysis in augmented reality and in gesture-controlled object motions for assisted living.

USING IMAGE INTERACTIONS TO OBTAIN QUALITY CROWD-SOURCED DATA
Data is valuable, however collecting data can often be challenging. I would like to know how we may leverage image interactions to engage people in submitting usable data for a variety of applications. Applications that we are currently looking at include software and environmental engineering. We foresee that image-based data collection approaches can be applied in fields such as biometrics and forensics as well.

Banerjee

Please visit Natasha's
Website for more information.