"Could Facial Recognition Technology Have Caught the Boston Bombers?"

In a word, no.

Although the wide array of surveillance photos in Copley Square played a role in apprehending Dzhokar and Tamerlan Tsarnaev, advanced technology was not the key to the manhunt, as a bombing victim identified the brothers from a series of photos, which were subsequently released to the public to provide tips (a.k.a. HUMINT) that led the investigators to the brothers.

As Sean Gallagher pointed out in an Ars Technica piece last month: “For people who understand how facial recognition works, this comes as no surprise. Despite advances in the technology, systems are only as good as the data they’re given to work with.” And despite what CSI or NCIS may have led us to believe, “Video from a gas station surveillance camera or a police CCTB camera on some lamppost cannot suddenly be turned into a high-resolution image of a suspect’s face that can then be thrown against a drivers’ license photo database to spit out an instant match.”

A more optimistic take on the topic comes from ForeignPolicy.com’s Joshua Keating, who while admitting that old fashioned detective work was the key to tracking the Tsarnaevs, “this is pretty new technology and . . . we may be getting closer to this kind of thing actually being useful.”

Keating’s optimism is based on a study by Michigan State computer scientists Joshua Klontz and Anil Jain testing whether existing facial recognition software could have identified the Tsarnaevs based on the security camera images taken just before the bombings by adding three headshots of the each brother to a database of more than a million mugshots. When the database was filtered to only look at Caucasian men in their 20s (for some reason, Tamerlan’s photo seemed to draw a lot of female matches . . . which is ironic given his history of domestic violence and fundamentalist beliefs about gender relations), the program produced one bull’s eye based on Dzhokar’s high school graduation photo.  

But this is a thin reed upon which to express hope that this technology will significantly alter the tactics of manhunts, whether domestic or strategic. First, Dzhokar’s school picture would not actually have been available to law enforcement, and the headshot from a driver’s license photo did not place higher than 19th in any of the searches. Conversely, although NeoFace 3.1 did produce one correct match, another young man’s mugshot not only produced a bull’s eye, but was also ranked the third closest match against the other two photos of Dzhokar. So the facial recognition technology did produce a good match, only it was of the wrong person. Finally, because Tamerlan was wearing sunglasses on April 15th, his own mugshot from his 2009 domestic violence arrest did not place higher than 116,342nd as a match.

In other words, no matter how well the technology is developed, simple countermeasures such as wearing sunglasses will likely impede its effectiveness.

Gallagher notes that under the best circumstances, facial recognition can be extremely accurate, but to do so "almost always requires some skilled guidance from humans." This calls to mind former Delta Force commander Pete Blaber’s admonition about relying on technology in manhunts: “The reality and complexity of life virtually guarantee there will never be” an all-purpose technological panacea for finding people. “Instead, these types of capabilities should be looked at as part of an overall system. A buffet of capabilities that could be used in combination with our guys working the situation on the ground to assist in the vexing challenge of locating a wanted man.”