Professor Jake Aggarwal's research is in computer vision, specifically in smart surveillance systems that identify suspicious activity and alert humans to follow-up. His approach has evolved from the basic problems of determining movement of a 3-dimensional object from 2-dimensional images to a sophisticated system that can recognize aggressive activity even with multiple players and a cluttered background.
"Human sight is highly selective. We focus on what is important and filter the rest. It is unbelievably difficult to replicate the judgment that happens automatically and effortlessly for humans with cameras and computers," says Aggarwal. "My group is currently studying ways to identify potential threats from suspicious driving patterns, human-vehicle interactions, and human activities. Preliminary results are very promising."