Slashdot Mirror
Smart Cameras To Predict Crimes
hairybacchus writes: "The Independent News is reporting that scientists at Kingston University in London have developed video processing software that is able to predict behavior patterns of the people on-screen. They say it will be used to alleviate congestion in the London Underground or alert police to potential muggings. I wonder how long it will be before this is combined with face-recognition technology? It's spooky." I can't wait. "We searched you because the computer told us to." Trust the Computer.
Robot cameras 'will predict crimes before they happen'
CCTV: By learning behaviour patterns, computers could soon alert police when an unmanned camera sees 'suspicious' activity
By Andrew Johnson
21 April 2002
Computers and CCTV cameras could be used to predict and prevent crime before it happens.
Scientists at Kingston University in London have developed software able to anticipate if someone is about to mug an old lady or plant a bomb at an airport.
It works by examining images coming in from close circuit television cameras (CCTV) and comparing them to behaviour patterns that have already programmed into its memory.
The software, called Cromatica, can then mathematically work out what is likely to happen next. And if it is likely to be a crime it can send a warning signal to a security guard or police officer.
The system was developed by Dr Sergio Velastin, of Kingston University's Digital Imaging Research Centre, to improve public transport.
By predicting crowd flow, congestion patterns and potential suicides on the London Underground, the aim was to increase the efficiency and safety of transport systems.
The software has already been tested at London's Liverpool Street Station.
Dr Velastin explained that not feeling safe was a major reason why some people did not use public transport. "In some ways, women and the elderly are effectively excluded from the public transport system," he said.
CCTV cameras help improve security, he said, but they are monitored by humans who can lose concentration or miss things. It is especially difficult for the person watching CCTV to remain vigilant if nothing happens for a long period of time, he said.
"Our technology excels at carrying out the boring, repetitive tasks and highlighting potential situations that could otherwise go unnoticed," he added.
While recent studies have shown that cameras tend to move crime on elsewhere rather than prevent it completely, in certain environments, such as train stations, they are still useful.
And Dr Velastin believes his creation has a much wider social use than just improving transport.
His team of European researchers are improving the software so that eventually it will be capable of spotting unattended luggage in an airport. And it will be able to tell who left it there and where that person has gone.
However, the computer is not yet set to replace the human being altogether.
"The idea is that the computer detects a potential event and shows it to the operator, who then decides what to do - so we are still a long way off from machines replacing humans," Dr Velastin says.
"Well kids, you tried your best, and you failed. The lesson is, never try." -Homer Simpson
http://www.cordis.lu/telematics/tap_transport/rese arch/projects/cromatica.html
Their other projects are also interesting as well
"It is a greater offense to steal men's labor, than their clothes"
There is nothing scary about this; in fact, humans already do it on a regular basis. A department store security guard scopes out a crowd of shoppers for potential shoplifters. An airport security guard scans a terminal for suspicious activity. A cop checks out a crowded street looking for potential muggers and pickpockets.
The trouble is, humans are inefficient and expensive, and their "gut instincts" may be fallible. The mall security guard may be the only guy watching a dozen closed-circuit monitors, and he may even be dozing off from the monotony of his job. The airport guard might be a minimum wage high-school dropout with barely any training. The cop's instincts are pretty good, but as objective as he tries to be, he unconsciously tends to target members of a particular race instead of going by solid scientific indicators.
This technology (if it works) will be a Good Thing because:
1. It improves upon an existing system that helps keep us safe.
2. It could be more effective and consistent.
3. It could apply rules objectively, and could be designed to flag activities that truly are suspicious (e.g. "casing" a department store) rather than those that merely look suspicious to biased humans (e.g. a young black man in a record store). This means that it could help protect our rights more than the current system.
Cheers,
IT
Power corrupts. PowerPoint corrupts absolutely.
News on philipKdick.com
Greg
(Inside a nuclear plant)
Aaaarrrggh! Run! The canary has mutated!
This article is much more in-depth and does a better job of representing the technology. The article posted to Slashdot implies that Cromatica can predict a mugging. Cromatica identifies congestion and predicts suicide attempts. And it does this with pretty simple algorithms.
Briefly: Cromatica views crowds as changing colors against a background. When the colors stop, this is congestion. Likewise, suicide attempts are indicated by lingering for 10 minutes or more. It's pretty easy to identify a single person against an empty backdrop.
Of course, people are working on predicting muggings, and the article goes into that as well.
The article also has links to the research itself.