The U.S. spy world’s high tech research agency, the Intelligence Advanced Research Projects Activity or IARPA, has launched a new face-recognition challenge, as Atlantic Media’s government tech news site NextGov reports this week.
Unlike the procession of so-called “artificial intelligence” algorithms that have emerged in recent years for the purpose of identifying faces in photos or surveillance footage, but which have in fact been so crude that they can’t even compensate for the racial biases of their creators, however, IARPA is looking for something decidedly more advanced. According to the announcement of the Face Recognition Prize Challenge, the goal is to develop algorithms for identifying so-called “unconstrained images,” which are also tellingly described as “non-cooperative” images or “images in the ‘wild.'”
“The goal of the Face Recognition Prize Challenge is to improve core face recognition accuracy and expand the breadth of capture conditions and environments suitable for successful face recognition,” the announcement states. This is a two-part Challenge. “Face identification involves executing one-to-many search to return the correct entry from a gallery, if any,” while “face verification,” the second step, “requires the algorithm to match two faces of the same person while correctly rejecting faces of different persons,” the announcement continues. “Both tasks involve ‘non-cooperative’ images where subjects were unaware of the camera or, at least, did not engage with, or pose for, the camera.”
While privacy and civil liberties concerns may not be high on their list of priorities, IARPA acknowledges, at least tacitly, that there have been some very troubling failures in past government attempts at developing face-recognition technology that is even reliable enough to use to begin with.
In a sub-section of the announcement with a header that purports to answer the tantalizing question of “Why We’re Doing This,” IARPA notes that “Face recognition is hard. Algorithms are known to commit both false negative and false positive errors, especially when factors such as head pose, illumination, and facial expression depart from formal portrait photograph standards. IARPA is also aware that enormous research has been conducted in recent years with the advent of various deep neural network technologies. IARPA is interested to know whether this rich vein of research has produced advancements in face recognition accuracy.”
The Face Recognition Prize Challenge is accepting submissions through June 15, and winners of a total of $50,000 in prize money will be announced on Halloween (Oct. 31) of this year. The technology developed could ultimately be helpful in “preventing the next random act of violence or catching a child predator,” and could become essential for public safety employees, IARPA Program Manager Chris Boehnen reportedly said.
Whether the surveillance apparatus the government hopes to build with this technology will actually stops crimes from happening before they happen is something we can’t possibly know — at least not yet, although the movie Minority Report is looking more and more like a documentary these days.
“The challenge is just one of several biometric-themed projects IARPA has launched recently,” writes. “Such projects may attract mainstream attention as President Donald Trump directs the Homeland Security Department to invest in biometric tracking that could follow travelers entering and exiting the United States. IARPA’s Odin project, for instance, awards funds to companies developing technology that can detect when people are trying to disguise their fingerprints or iris scans. Another, called Janus, is aimed at improving face recognition in videos.”
As The Intercept reported in January, Trump’s transition team included several people with ties to the biometric surveillance industry and specifically to makers of face-recognition products. As I’ve previously noted, the U.S. government is not only looking into beefing up security on this side of the border, but at potentially funding a biometric surveillance system for Mexico.
IARPA’s Face Recognition Prize Challenge is just the latest development of an ongoing trend towards what often seems like the government intends to ultimately become a system of total, constant and completely inescapable surveillance. It would not be the least bit surprising if Trump’s “law and order” agenda eventually involved a push for greater use of government video cameras and biometric monitoring technologies like “non-cooperative” face recognition — not only at airports and borders but in all of the areas of public and private life that state surveillance increasingly intrudes on.