Artificial Intelligence For Security Camera Systems

In years of economic recession, the level of character crime tends to grow quickly. If one has concerns about safeguarding the well-being of your employees and protecting your character, then installing systems to enhance security can be a wise move. Theft and fraud cost American business more than $40 billion a year.

Security surveillance use has become extensive as more people want to protect themselves from harm. An ABC News/Washington Post poll in 2007 found that 71 percent of Americans favor increased video surveillance. Already in the United States, there are more than 30 million surveillance cameras shooting in excess of 4 billion hours of video recording every week.

When planning a security camera system, one needs to ask: what is the objective? Do we want a video record so that it can be reviewed in the event of an anomaly, or do we want the ability to react real-time (to an event in progress)? Either objective can be achieved, but the system will need to be more complex if the objective is to react real-time.

Mounting video cameras is cheap, but funding human resources to observe the output is expensive. Like so many problems, the solution can be computer technology. Computers cannot do the job by themselves, but they can analyze the imagery and automatically alert a human operator to any suspicious event.

“It’s impossible for insignificant mortals with eyeballs and brains to course of action all the information we’re gathering,” says Stephen Russell, CEO of 3VR.

Progress with artificial intelligence (AI) techniques has largely been pushed by our efforts to combat terrorism. This technology is called “Video Analytics”. Facial recognition software is already in use hotels, edges, airports, etc. This software is far from perfect, but it is a “force-multiplier”, that is, instead of a 100 humans monitoring 20 video screens each, the computer software screens for suspicious behavior, and a single human can monitor 2000 cameras. The system can be integrated with the Internet, so that the human need not be on-site.

The technology used by Homeland Security is trickling down to systems being employed to monitor businesses. It is a much simpler task for the software to identify suspicious activity in a loading dock or warehouse during a period when it should be idle, than it is to determine suspicious activity in a busy ecosystem.

Formerly dominated by burglar alarm companies and camera experts, the video surveillance field is rapidly becoming an industry of information technology integrators, who can offer a more complex approach. Revenue generated from surveillance software is expected to increase from $245 million last year (2008) to more than $900 million in 2013, according to a report by ABI Research.

It is important that we are realistic with our expectations of video analytics. Computers have limited visual intelligence and can perform some video analytic roles dependably, but only by constraining the application. Computers are good at locating possible events, and humans are good at validating those events.

Leave a Reply