New Security Camera Tells Suspicious People from Lost Ones

The new “smart” surveillance system, designed by James W. Davis, an associate professor of computer science and engineering at the Ohio State University, and doctoral student Karthik Sankaranarayanan, surpasses regular surveillance cameras by a great deal, as it is no longer limited to the tilt-pan-rotate trio that characterizes regular cams. The new designs have the ability to trace a single person in the crowd, and figure out if that someone is lost or is acting suspiciously.

"In our research, we care what you do, not who you are. We aim to analyze and model the behavior patterns of people and vehicles moving through the scene, rather than attempting to determine the identity of people. We are trying to automatically learn what typical activity patterns exist in the monitored area, and then have the system look for atypical patterns that may signal a person of interest – perhaps someone engaging in nefarious behavior or a person in need of help," Davis says.

"The first piece of software expands the small field of view that traditional pan-tilt-zoom security cameras offer." It is able to create a 360 degree-wide view of an area, which is very similar to that obtained through a frog's eye. Though it takes a few minutes for the camera to photograph its surroundings and create the panoramic view, the system is still effective in that it can survey a larger area than regular ones, and can then be oriented by a single click, to follow a specific person.

Most of the workload is processed by intelligent computer software, which will soon be endowed with the ability to "hand off" the chase, meaning that, if more smart cams are linked together in a unified system, once the tracked person moves out of the sight of one camera, he or she would be followed via another one.

"That's the advantage of linking all the cameras together in one system – you could follow a person's trajectory seamlessly," Davis says. "Humans can pick out a lost person really well. I believe you could build an algorithm that would also be able to do it."