News: Facial Recognition Technology to be used at Cardiff Airport

Frome the Register: Cardiff now joins Manchester Airport in allowing inbound passengers to have their passports checked automatically with facial recognition systems. The option is available for adults with chipped biometric passports, issued since 2006, which, amongst other information, have a picture of the holder encoded on them that can be compared to the subject checking through.

The article refers to the embarassing revelations last year that facial recognition technology in use at Manchester airport was unable to distinguish between pictures of Wynona Ryder and Osama Bin Laden. This happened because the machines initially gave far too many false negatives, and in reponse staff turned the settings so low they effectively 'switched them off'.

News: More on facial recognition technology

According to a report from USA Today, the FBI has begun trials of facial recognition technology in North Carolina. Reportedly, the trials helped to identify a suspect to a double homicide who had seemingly relocated to North Carolina from California and assumed a false name. FBI officers took a photo from the suspect’s California driver’s license and ran it against the photos contained in North Carolina’s Department of Motor Vehicles database. From “dozens” of potential matches, an FBI investigator zeroed in on one particular individual. That individual has now been placed under arrest. Marc Rotenberg of EPIC, however, questioned how effective the use of such technologies will be in counter-terrorist efforts since good photos of terrorists will rarely be available in DMV databases or elsewhere.

Comment: Lies and Faces

The ABA Journal has published an interesting article on ongoing efforts to develop a better lie detector. Particular focus is placed on the use of various technologies such as EEGs and MRI to obtain a picture of brain activity. But the article also discusses the use of technology to examine eye movement and detect minute changes in facial expression. A number of critics have questioned the reliability of such methods. Given that the article appears in a publication of a lawyers’ professional association, it is not surprising that it places particular emphasis on the potential use of these technologies to develop evidence to be used in criminal prosecution. But, are there other ways that these methods might be implemented in the counter-terrorist context? Many of us have been put through little interview sessions at the airport during “heightened threat levels” before being permitted to board a flight. The idea behind these interviews being that the observation of our behavior when answering these questions as well as the actual content of the answers themselves might give us away if we have anything sinister planned. Immigration officials also generally ask us some questions before allowing us to enter a different country – although these interviews have a broader purpose than merely trying to ferret out terrorists. Might some government decide these interviews could be made more effective if we were having our brain activity, eye or facial movements scanned while they were being carried out?

On the subject of facial recognition technology, this area seems to be getting more and more attention, particularly in the security sector. There’s been a project at the University of Zurich which examined facial expression and emotion as well as their relevance for facial recognition technology. In order to be effective, this kind of technology will have to match faces that change in all manner of ways in the course of everyday human activity to static ID-photos that have been recorded in a database.

There’s also been a thread of research that has aimed at developing technology that can detect "abnormal behavior" or emerging dangerous situations - see for instance John's earlier post on INDECT. I can see how these two threads could merge where facial recognition technology would be used not only for identification but also in threat detection: i.e., the attribution of certain emotions to facial images could be used to determine whether dangerous or abnormal behavior is present.