Facial Recognition: What is Private?


Researching Tractica’s newly-published report on Facial Recognition turned up some interesting use cases.  Among biometric modalities, facial recognition is unique – it’s the most frequent way that we humans recognize each other.  Any technology that stores people’s faces is bound to generate discussion.  And become emotional.

Whether or not to remain private is often treated by legislation as a personal decision – an opt-in.  Even Europe, where citizens have long demanded personal privacy, has 7 of the top 10 selfie-publishing cities globally.  Still, laws in Europe tend to give each individual the right to be on the web or stay away.

Facial recognition biometrics could change all that.  It may not put people unknowingly on Facebook, but facial images can be stored without anyone’s knowledge, let alone agreement.  One study estimated that Great Britain, with a population of over 60 million, may have nearly 6 million closed circuit TV (CCTV) cameras deployed.  Few in Britain or anywhere realize yet that facial recognition vendors can build face-print databases by scanning the video from all those cameras.

A face-print database is similar in concept to a fingerprint database, with two big differences.  First, a person may have no idea that his facial image has been captured.  Facial image capture is passive and possibly stealthy.  Fingerprints require the subject to present their finger for printing, so presumably anyone fingerprinted is aware of it.  Second, as noted above, facial images are “us” – they are how we know each other.  The same is rarely true of fingerprints.

Facial recognition is not federally regulated in the United States.  The U.S. National Telecommunications and Information Administration (NTIA) has been leading a process to create a voluntary code of conduct.  Recent reports suggest that privacy advocates have withdrawn from the process over concerns that proposed regulation will not preserve personal privacy.  This may have something to do with proposed use cases.

One such use case is digital signage.  Retailers and advertisers can use biometrics to anonymously profile customers by their facial images to serve up specific ads for each person.  Biometrics can also profile what demographic of people spend the most time in each part of a retail store, useful for product placement or promotions.  Digital signage is perhaps still a bit in the future given current profiling capabilities, but it is coming:  the recent Digital Signage Expo in Las Vegas attracted well over 200 exhibitors.  Yet, few words generate more emotion than profiling.

Personalized hospitality and entertainment is another use case.  For example:  a guest enters a hotel room and a biometrics system triggers a thermostat setting, perhaps turns on a sports channel, maybe even orders an in-room meal.  Incredibly, some vendors have suggested that facial recognition could meet this use case.  Yes, a camera in the hotel room.  I think that voice and speech recognition are more likely candidates.  Just as with digital signage, vendors appear to be considering the art of the possible but maybe not thinking consumers’ responses all the way through.

To be sure, it is possible to anonymously capture images and profiles without tying them to a specific person.  That can be a tough sell on the general public, though.  How does a retailer or even a law enforcement agency prove that images are kept anonymously?

In the likely event that the last question cannot be answered to anyone’s satisfaction, it is possible that personal privacy issues could present a roadblock for facial recognition technologies.  Nearly every obstacle can be overcome.  But this one may require some planning and some transparency.

Comments are closed.