Recognising the problem

[Dave Birch] An interesting series of talks at Biometrics 2010 reminded me how quickly face recognition software is improving. The current state of the art can be illustrated with some of the examples given by NIST in their presentation on testing.

  • A 1:1.6m search on 16-core 192Gb blade (about $40k machine) takes less than one second, and the speed of search continues to improve. So if you have a database of a million people, and you’re checking a picture against that database, you can do it in less than second.
  • The false non-match rate (in other words, what proportion of searches return the wrong picture) best performance is accelerating: in 2002 it was 20%, by 2006 it was 3% and by 2010 it had fallen to 0.3%. This is an order of magnitude fall every four years and there’s no reason to suspect that it will not continue.
  • The results seem to degrade by the log of population size (so that a 10 times bigger database delivers only twice the miss rate). Rather fascinatingly, no-one seems to know why, but I suppose it must be some inherent property of the algorithms used.

We’re still some way from Hollywood-style biometrics where the FBI security camera can spot the assassin in the Superbowl crowd.

What is often overlooked is that biometric systems used to regulate access of one form or another do not provide binary yes/no answers like conventional data systems. Instead, by their very nature, they generate results that are “probabilistic”. That is what makes them inherently fallible. The chance of producing an error can be made small but never eliminated. Therefore, confidence in the results has to be tempered by a proper appreciation of the uncertainties in the system.

[From Biometrics: The Difference Engine: Dubious security | The Economist]

So when you put all of this together, you can see that we are heading into some new territory. Even consumer software such as iPhoto has this stuff built in to it.

face-rec

It’s not perfect, but it’s pretty good. Consumers (and suppliers) do, though, have an unrealistic idea about what biometrics can do as components of a bigger system.

But Microsoft’s new gaming weapon uses “facial and biometric recognition” that creates a 3D model of a player. “It recognises a 3D model that has walked into the room and automatically logs that player in,” Mr Hinton said… “It knows when they are sneakily trying to log into their older brother’s account and trying to cheat the system… You can’t do it. Your face is the ultimate detection for the device.”

[From Game console ‘rejects’ under-age players | Herald Sun]

This sounds sort of fun. Why doesn’t my bank build this into its branches so that when I walk in?


Subscribe to our newsletter

You have successfully subscribed to the newsletter

There was an error while trying to send your request. Please try again.

By accepting the Terms, you consent to Consult Hyperion communicating with you regarding our events, reports and services through our regular newsletter. You can unsubscribe anytime through our newsletters or by emailing us.