Skip to main content
Home Asia Europe U.S. World Business Tech Science Entertainment Sport Travel Weather Specials Video I-Reports
Science and Space News
In association with:

Smart cameras spot shady behavior

By Dean Irvine for CNN
Adjust font size:
Decrease fontDecrease font
Enlarge fontEnlarge font

LONDON, England (CNN) -- Ever felt like you were being watched? With the rise in the number of surveillance cameras on our streets and in our offices, more often than not you are.

But rather than just watching you, the next generation of surveillance cameras will be able to tell if you're up to no good and, developers hope, spot crimes and misdemeanors before they happen.

Normally it is down to the judgment of a security guard watching a bank of monitors to decide if something looks suspicious on one of their screens.

However, with an average of over 100 monitors for each security guard to keep an eye on, it is practically an impossible task. There is also the loss in attention that naturally occurs when people are sitting watching screens nonstop.

Industry experts suggest that after 12 minutes of continuous video monitoring an operator will miss up to 45 percent of screen activity. That rises to up to 95 percent after 22 minutes.

Too many screens, not enough eyeballs

Trying to solve the problem of too many monitors and not enough pairs of eyes is Dr. Sergio Velastin of Kingston University in the UK.

"Humans will always be better than machines at spotting real behavior, but most security guards have an almost impossible task to watch so many screens all the time that they can't be used practically," Velastin told CNN.

Velastin has been developing the software with Ipsotek, a UK based company, pioneering the new visual imaging devices.

"Most people do things in a fairly straight forward way and we're able to gain statistical knowledge of what they do. From that it follows that you can raise an alarm if something is deemed 'infrequent', which usually means abnormal or suspicious," he said.

It sounds simple enough, but the task of creating a computer program that can filter out all the normal background goings on of a situation, be it on a train station platform or high street, has proved to be extremely complex.

Rather than going down the route of face recognition technology, the program tries to keep it simple by comparing any number of situations and actions that would be captured by a camera with an empty background.

By using an algorithm to tell the normal from the abnormal, the software is able to alert a security guard if it deems something to be out of the ordinary.

A person standing close to the edge of a train platform might not necessarily be an immediate suicide risk, but the program is able to track that person and evaluate the amount of time they've been there and so constantly monitor the situation.

"We remove the background furniture of a picture and just highlight the new people and objects. This is moving on from things like a medium motion sensor. That's fine for things like a prison fence, but not for the London Underground or a busy shopping street where you constantly have movement," said Velastin.

Watching out for civil liberties

In the UK there is approximately one surveillance camera for every 14 people and issues of invading civil liberties surround ever new development in our surveillance society.

One advantage of the new technology is that it is less invasive than face recognition software. Civil liberties groups have criticized technology that searches for particular people as it brings with it fears of profiling and of being watched regardless of whether anything illegal was being committed or not.

There is still some way to go before a "smart camera" can tell the difference between details such as a handshake and a punch, but Velastin believes they are not too far away.

"At the moment you can't get a camera that can do that, not in a meaningful way, but this is something we are working towards. In three to five years we hope to have a program that would identify from your walk whether or not you are carrying a gun," he told CNN.

High-definition smart cameras are also being developed in the U.S. by Interact Public Safety Systems. Their system is digital, which means that any dubious-looking activity can be watched remotely from hand-held devices or via the Internet.

The company has trialed its technology at a U.S. high school and can be tailored to look for certain abnormal behavior.

With the application of these smart cameras, ranging from spotting crimes on the high street or traffic violations to keeping children safe at school, the issue or who decides what is deemed suspicious comes into focus.

"Prejudices could be built into rule-based programs that have instructions of what should be spotted. There are elements of profiling in this and the possibility for abuse," says Velastin.

"But I'm into this because I'm socially minded and believe that what we're doing is creating technology that gives us a better chance to police properly. It's hard to do an audit trail of what someone does in a control room and with this technology there is less chance for abuse," he told CNN.

"In society we have to make a judgment of how far we want to be watched and how much we want to feel protected," said Velastin.

Moving from public spaces to our homes, Velastin believes the technology could provide a solution to caring for the elderly. "I wouldn't want someone watching me through a video camera 24 hours a day, but I might feel OK if a computer or robot was doing it. Given the choice what would you rather?"


With so many monitors to watch, spotting suspicious activity can be a near impossible task.


Would you feel safer with more surveillance cameras on the streets?
or View Results

Your comments


E-mail address

Your location


All fields are required
Comment Policy
CNN TV How To Get CNN Partner Hotels Contact Us Ad Info About Us Preferences
© 2007 Cable News Network.
A Time Warner Company. All Rights Reserved.
Terms under which this service is provided to you.
Read our privacy guidelines. Contact us. Site Map.
SERVICES » E-mail RSSRSS Feed PodcastsRadio News Icon CNN Mobile CNN Pipeline
Offsite Icon External sites open in new window; not endorsed by
Pipeline Icon Pay service with live and archived video. Learn more