Wanted: algorithms that can identify people from security-camera footage.
The Facebook algorithm that auto-tags people in photographs might be slightly creepy, but also of interest to the intelligence community.
The IC's research and development unit is hosting a new contest in search of the best facial recognition algorithms that can identify individuals in images taken from the “wild,” for example, sources such as security footage.
The Intelligence Advanced Research Projects Activity's "Face Recognition Prize Challenge" seeks algorithms that can accurately and quickly match a photo found in passive footage to another of the same individual from a gallery, as well as systems that can verify, or match, two images of the same person while rejecting photos of other individuals. The most accurate search algorithm wins $25,000. The fastest wins $5,000, and the most accurate verification algorithm wins $20,000.
The technology could be helpful in "preventing the next random act of violence or catching a child predator," and may become essential for public safety professionals, IARPA Program Manager Chris Boehnen said in a statement. An abundance of training data from real-world scenarios has made more sophisticated facial recognition technology possible, he said.
The challenge is just one of several biometric-themed projects IARPA has launched recently. 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.
The contest runs until June 15 and winners will be announced in October.
NEXT STORY: Stinger Missiles Can Now Shoot Down Small Drones