The two faces of face-recognition technology
The two faces of face-recognition technology

Oct 1, 2001 12:00 PM

Face-recognition technology emerged in the public consciousness during the Super Bowl at Raymond James Stadium in Tampa last January, when the technology was used to identify possible criminals. In the last few weeks, the computer-based identification of people based on facial characteristics is higher-profile than ever. Its use has been suggested as an investigation tool following the terrorist attacks, and there is talk about its possible use as a component in the needed improvements to the nation's airport security.

Overnight, face recognition moved from the cloistered narrow technical and scientific community, to gain international attention that has generated enormous public awareness and interest. The level of media coverage has been extraordinary, and the airwaves have been filled with rhetoric that ranges from, "it is okay to spy on criminals" to "chilling" and "scary," and, of course, "Orwellian" and "Big Brother," just to pile on.

However, there has been little thoughtful reflection on the appropriate ways to deploy this new technology to best protect the American right to privacy and the presumption of innocence, while responding to new realities that require new law enforcement tools to combat 21st century crimes, such as terrorism, identity theft and Internet fraud.

Face-recognition technology has emerged since the early 1990s, primarily from research originally developed at MIT's renowned Media Labs. In a short period, face recognition has gained acceptance because it is convenient, non-intrusive, intuitive and cost-effective. However, these attributes heighten fears that inappropriate proliferation and potential abuse will result because face recognition can be so inconspicuously and unobtrusively implemented.

Conducting surveillance and spying on pedestrians as they walk city streets and stroll their own neighborhoods is vastly different than providing responsible security in enclosed public places such as airports, casinos, stadiums, embassies and other high-risk targets. Certainly, no reasonable person is troubled by a baggage search at the airport, but most Americans would consider it unreasonable to have their bag searched as they walk down a city street or enter their own neighborhood. This analogy applies to reasonable and unreasonable implementations of face-recognition technology. No one should support the unfettered, indiscriminate use of face recognition, or any other technology, for general face matching surveillance on commercial streets or in residential neighborhoods. This type of ubiquitous and arbitrary surveillance submits ordinary citizens to an unprecedented real-time continuous stream of digital lineups without any probable cause and without a compelling security threat. Such technological tyranny can create a siege mentality, violate our right to privacy and could impinge on our ability to travel freely.

It is unrealistic to expect any degree of successful results when scanning thousands of moving people in a crowd, under various lighting conditions, in challenging environmental surroundings and with no ability to use choke-point mechanisms such as turnstiles. In fact, in an uncontrolled physical environment where the cameras are searching for a single person in a crowd, false matches will exceed positive matches by an order of magnitude.

To be effective and successful, the state of the technology requires the ability to have some control over environmental factors, such as weather conditions, sunlight and nightlight as well as the ability to clearly segregate a single head from hundreds of bobbing heads in a moving crowd.

From the perspective of law enforcement effectiveness, the cost of the technology is trivial relative to the level of effort required to monitor all the cameras on a 247 basis. This level of surveillance also requires maintaining large numbers of human resources so that officers can be dispatched to the exact scene, within the few short seconds that it takes the suspect to cross the camera's field of view. This process is perhaps akin to trying to shoot a moving fly with a shotgun and then trying to determine where it landed. We no doubt could find Whitey Bulger, one of the FBI's Ten Most Wanted, if we put a camera on every street corner in America. However, the cost of the technology, equipment and law enforcement resources is prohibitive and impractical. Worse, undoubtedly many innocent people would be unjustifiably interrupted and interrogated first.

Once they understand the technology, the public will not only accept reasonable implementations of face recognition applications but would also readily and enthusiastically expect them.

There are a number of real world security and law enforcement investigative needs that would benefit from face-recognition technology. Identity theft is real and it is the fastest growing crime in America today, claiming 750,000 victims in 1998. The average loss is $30,000 per crime and it takes more than two years for a victim to restore actual identity. Another $10 billion a year is lost to check fraud alone in the U.S. More personal and heart-wrenching are the stories like the one reported on the May 2001 front page of USA TODAY. It was a story of a grieving widow who, four weeks after the death of her husband, found herself charged with a $160,000 bill that was incurred by an imposter with information gleaned from the obituary pages.

Some industry figures claim that more than 5 percent of State-issued drivers' licenses are duplicates. When, as a public safety official states, "one in every five fatal crashes involves a driver who does not have a valid license or whose license is a mystery to law enforcement," we need to be concerned. We have the technology to do better and we can. In fact, face-recognition technology has identified real criminals, including one who had obtained 17 fraudulent drivers' licenses. In a criminal's hands, a fraudulent driver's license enables a variety of crimes to happen from the voting booth, to forgery, to falsely gaining duplicate eligibility to entitlement programs and to Internet fraud.

Internet security is another serious and well-documented problem. Credit card fraud runs 10 times the fraud rate of in-person retail transactions. School-age children can commit Internet crimes unimagined and unparalleled in scope by our most hardened career criminals and, as the Social Security Administration's inspector general said to Congress, "It's risk free". Face recognition can well serve as a "digital signature" for authenticating and verifying Internet business transactions. Face-recognition technology can have a major impact in the fight against one of the cruelest invasions and violations of our privacy identity theft.

Access control/keyless entry and ATMs are additional examples where face recognition is being used positively for consumer convenience, to increase personal security and to improve individual privacy. Unauthorized property entry and security are huge problems for public schools, college campuses as well as offices, factories and government buildings.

Certainly, the benefit of surveillance technology is not to indiscriminately capture millions of individual images in the hopes that a wanted pickpocket is among them. Rather, much like a fingerprint that is left at a crime scene, so are "face prints" left on a surveillance camera. Like a fingerprint, a "face print" can be captured, searched and analyzed so an investigator can match a known perpetrator to an appropriate database file. Using "face prints" in the same investigative way that investigators have used fingerprint technology for the past 100 years is a reasonable way to use face recognition in conjunction with CCTV surveillance cameras. This is the opportunity of the technology provide law enforcement with an additional tool to apprehend real criminals. This is the opportunity that we should not let the wrong face of face recognition obscure.

Face recognition and biometric technologies are here, they work and they are beneficial. It is vital that policy makers, public safety officials and the industry win public confidence and protect their right to privacy by not supporting intrusive, inappropriate and ineffective biometric applications. Instead, the focus should be on using this new powerful technology to help achieve our common objectives improve personal convenience, improve public safety and improve individual privacy.

Tom Colatosti is president and CEO of Viisage Technology Inc., Littleton, Mass., a provider of face recognition technology.

Viisage Technology 29

Viisage Technology Inc. has offered the FBI free use of Viisage's face-recognition technology to aid in the apprehension or identification of the persons responsible for the terrorism in New York City and Washington.

"It is clear that the use of face-recognition technology can make a real difference in identifying terrorists," said President and CEO Tom Colatosti.

  • ASIS Expo in Review, Dallas, Sept. 14-17, 1998
  • Portable digital video recorder - Micro DVR
  • Sept. 11 panel questions federal airport security
  • 12 Hour Line Activated Telephone Recorder
  • Military launches cyber-security campaign
  • 4 Channel Digital CCTV
  • P5080 - 5+ Hour Telephone with Extended Recording Time
  • Nothing But the Best
  • EXCLUSIVE - New WebBrick Launched
  • Color Outdoor CCTV Vandal Resistant Dome Video Security Camera
  • Security Camera News