SFO testing video analytics

Published 24 July 2006

Speaking of rail security, Congress may want to consider the use of video analytics — now under testing at several of the nation’s airports — in mitigating suspicious activity at rail stations

The bombings of trains in Mumbai, Madrid, and London have really focused the spotlight on the vulnerabilities of U.S. railways. The government has taken notice, and is making efforts to minimize risk from terrorist attack. We reported last week about DHS testing new antiterror detection technology at the Exchange Place PATH station in New Jersey. This step in the right direction is good news, but the bad news is that there are still thousands of miles of railway in the United States left vulnerable. One way to manage this problem may come in the form of video analytics. We have reported on this video-capturing analytics solution before, but now it seems to be picking up some steam. The technology uses software algorithms to scan surveillance video gathered by closed circuit television cameras to search for specific visual patterns and behaviors. Though the industry is small, twelve firms sold $65 million worth of video analytics products last year according to Frost & Sullivan.

The rail industry — and Congress — should take a page out of airports’ security approaches. At San Francisco International Airport (SFO) Sunnyvale, California-based Vidient Systems video analytics software is used. SFO is part of a DHS $30 million trial project with several other airports testing video analytics. Paul Foster, aviation security manager at SFO, said his office began working with Vidient about three years ago, using a grant from the Transportation Security Administration (TSA) to help install, test and refine successive upgrades of the software to detect an array of visual occurrences — such as a vehicle going up an exit ramp, a person coming over a fence, or the behavior that airport officials call “human tailgating” — two workers coming through a security door at once. When the software detects something out of the ordinary or any suspicious behavior, an alert will notify security officials who are then able to focus in on specific video of the incident. This type of image sifting is what makes video analytics an exciting technology. No longer will there be the need to rely on actual people — who often times get tired and daydream — to sieve through hours of video surveillance. Civil libertarian groups, however, are not keen on this type of surveillance invasion. “This is the latest in a line of post 9/11 data-mining technologies,” said Nicole Ozer with the American Civil Liberties Union of Northern California. “The actions of everyone will be captured and monitored and sifted by algorithms, and if they look troubling, you may get flagged.” Gary Resnick, vice president of Sunnyvale’s Vidient Systems argues that the new technology will not only be more affective in security, but also reduce the amount of random searches.