LetterstotheEditor
Discuss This Article

Letters to the Editor

To the Editor:

I read with interest the article concerning Viisage Technology and its chairman, Denis Berube '65 ["More Than a Face in the Crowd," Spring 2002]. I have no doubt that Berube's management style, dedication and technological savvy all deserve close scrutiny and admiration. Where the article falls short, however, is in substantive evaluation of the product claims.

A quick search with Google brings up some interesting hits, including a July 17, 2002, story in the Boston Globe ("'Face Testing' at Logan is found lacking"); a May 29, 2002, article on TheStreet.com ("Glow Fades From Face-Recognition"); and a May 30, 2002, article in the Boston Business Journal ("Fresno Airport drops Viisage facial ID system, officials say.").

I also found the test reports at the Face Recognition Vendor Test site (www.frvt.org/frvt2000/) of particular interest. The test reports are too long to describe here in detail, but the results are not as reassuring as they could be. Of particular concern is the disturbingly low probability of correct identification under varying light conditions and distance from the camera and the results of testing at various poses (angle between subject face and camera).

I have no doubt that under tightly controlled conditions (facial recognition as a security authorization method in restricted facilities, a substitute for ATM identification, etc.), computerized facial recognition can be quite useful. But its ability to discern subjects under random conditions (crowds, varying lighting, varying angles, etc.) is still seriously lacking and does not merit its use as a significant tool in law enforcement.

When reporting on such technologies, you owe your readers much more in terms of understanding the technology at hand, its suitability to task as claimed by its vendors, and its social ramifications.

Alon Harpaz '00 (MBA)
Ashland, Mass.

To the Editor:

Re: the sidebar entitled "Security vs. Privacy" that accompanied the article on Viisage Technology in the Spring 2002 issue, I was taken aback by Denis Berube's Pollyannaish sidestepping of the privacy concerns raised by his company's fascinating and valuable technology. In asserting that "the system only looks for known threats to society," he ignores the system's ability to create databases of individuals on the fly and its owner's license to decide the nature of threats.

Imagine that J. Edgar Hoover, Joe McCarthy or the KGB had had access to this technology. A misguided government official or organization clandestinely plants a tiny $100 digital video camera at the site of a legal protest rally. The system scans the crowd, eliminating those faces that it knows belong to its own operatives, and stores all the other faces in a database of "undesirables." Rather than risk a riot by arresting members of the group at the time of the protest, the resulting database is transmitted to patrol cars, security checkpoints and supermarkets throughout the city. The protestors are quietly rounded up in the coming days, when they're all alone with no one to defend them.

This chillingly Orwellian picture must be the concern of all citizens, but especially of those, like Mr. Berube, who know this technology best. I am a big fan of this technology, but I believe it's critical that those closest to it lead the charge to prevent and counter its misuse. Mr. Berube shirks his responsibility and insults the intelligence of the WPI community by failing to embrace and address these legitimate concerns.

Marc C. Trudeau '81
Endicott, N.Y.

Denis Berube responds:

With regard to privacy concerns surrounding face-recognition technology, it is important to understand that Viisage's system does not store any images or know anything about the individuals being scanned. The system simply takes an image of the individual passing the camera and compares that image to images provided by law enforcement to determine if there is a match.

In a thoughtful article in the Sept. 1, 2002, Boston Globe, Alan Dershowitz, a constitutional scholar known to be an avid defender of our personal and constitutionally protected privacy, addressed the subject of privacy in the age of a terrorist war. Noting the common misunderstanding about the difference between privacy and anonymity, he said no one is granted anonymity, which is contradictory to the obligation of government to provide services and security to individuals. He called a national ID using face-recognition technology acceptable.

In regards to the issue of performance, face-recognition technology works well in appropriate applications. Applications for which it is well suited include visa identification, border crossings, driver's licenses, airport screening and police booking systems.

There has been a considerable amount of inaccurate information reported and repeated from nontechnical sources with regard to the performance of Viisage technology at Boston's Logan International Airport and the airport in Fresno. Logan conducted the most comprehensive test of face-recognition technology anywhere in the nation. Viisage achieved 90 percent performance rates during that test--incredible by any standards, and certainly sufficient to help to deter a terrorist from joining one of us on an airplane. Face recognition technology can play an important role in making our airports safer at a nominal cost. (The Fresno Airport, by the way, issued a denial of the story mentioned in the letter shortly after it appeared.)

There are 14 companies selling face-recognition technology. Prior to 9/11, two face-recognition companies were acknowledged as competent by the U.S. Department of Defense; Viisage was one of them. Often, a report by the news media about a failure of face-recognition technology is, in fact, about the failure of a particular company's technology. With so many newcomers to this arena, there are plenty of opportunities to find negative news. But generalizing from these individual failures is a disservice to the competent companies that provide a good product and deploy it appropriately.

transformations@wpi.edu
Maintained by: webmaster@wpi.edu
Last modified: Jan 13, 2003, 11:18 EST
[WPI] [Alumni] [Home]