Video cameras in transit systems are all well and good, but unless someone—or better yet, a computer—is watching all those cams, the technology won’t help fight crime or terrorism.
An 85-page report from the University of Southern Florida Center for Urban Transportation Research has taken a look at how video analytics can help that mission, explaining challenges, describing the technology and its history, making recommendations and (maybe most helpfully) comparing the capabilities of various video analytics tools on the market.
The report, “Evaluation of Smart Video for Transit Event Detection,” was prepared for the Florida Department of Transportation Research Center and completed in June.
It cites other studies noting that in just 20 minutes of watching security monitors, the boredom reduces the attention of most individuals to an unacceptable level.
An analytic system can increase situation awareness, provide early detection of potential threats and provide forensic evidence after an incident, the report notes.
But it also identified drawbacks with video analytic systems—environmental variables such as lighting and weather; false positives; the need for re-calibration of analytic algorithms; and a lack of standard evaluation methods for behavior. There’s no reliable independent verifiaction of the claims of analytic software providers. Also, analytic systems typically rely on human analysis as well to assess threats and prompt further action.
“Improved core technology algorithms are needed to increase the reliability of human behavior recognition,” the report says in its Executive Summary.
According to the report, while CCTV use is nearly ubiquitous in large transit systems, most don’t make effective use of analytics.
In Florida, for example only 20 percent of agencies use any form of video analytics (forensic or real-time) for surveillance purposes, and it was unclear to the report’s authors how much, if at all, the agencies were using the tools effectively. At the time of the survey, no agency in Florida was considering evaluating or deploying analytic systems, reporting that budget constraint was a limiting factor.
New Jersey Transit, the largest statewide transit agency in the United States, uses real-time video analytics in conjunction with its 1,400-camera CCTV system to detect unattended packages in its facilities.
In New York, the report cites news reports that the Metropolitan Transit Authority’s $300 million surveillance system was not living up to its promise. “This situation is likely to recur since there is no formal independent evaluation of such systems,” the report says. “In simple terms, no studies corroborate the vendor’s performance claims or indicate a relative performance comparison across different available products.”
View the complete report here.