The scenario has been played out in countless action movies: Computer geek sits in a darkened room staring at a screen. He gets an alert that a camera nearby has matched a face in the crowd with some miscreant in his database. Next scene: FBI agent accosts the bad guy. Is it any wonder we think technology can protect us?
Sadly, Hollywood's imagination is way ahead of the technology. While some airports and communities, such as Virginia Beach, Va., and several British cities are deploying face-recognition technology, public tests have been disappointing. In the first half of an eight-week test in Palm Beach International Airport, a system using software from Identix Inc. failed 53% of the time. In a 90-day test at Boston's Logan International Airport, consulting firm Counter Technology Inc. found that systems from Identix and others produced too many alerts--many of them false alarms. The problem: Differences in lighting and angles between the video image of the real person and photos stored in a database.
Identix blames its low scores on faulty installations. If the system is correctly calibrated, it is 98% accurate, the company claims. Vendors also stress that the technology has some uses that few experts criticize. It works well when people stay still, or small groups are herded through control points, says Michael Thieme, director of special projects at International Biometric Group, which completed an extensive study for the Transportation Security Administration in June. This approach may be suited to casinos, but isn't practical for high-traffic airports or street corners. For such fluid environments, the technology "is so far away from working, it's hard to say when it will be widely deployed," Thieme says. But Hollywood can always dream.
By Heather Green