Could police officers someday identify criminals just by looking at them?
That's the vision being touted by Taser International, which holds a monopoly on "conducted electrical weapons" for law enforcement and is aiming to build one for police body cameras, according to a Businessweek cover story.
In the story, Lieutenant Dan Zehnder of the Las Vegas Police Department imagined himself patrolling the Las Vegas Strip with his Axon model Taser body camera streaming back to headquarters. The footage gets "real-time analysis, and then in my earpiece there is, 'Hey, that guy you just passed 20 feet ago has an outstanding warrant,'" he told the magazine. "Wow."
Taser's CEO, Rick Smith, told Businessweek that the company plans to begin live-streaming body camera footage to the cloud by 2017, and that facial recognition could arrive soon after that.
Facial recognition has, in fact, long been part of Taser's plan. It's been mentioned in Taser press releases as far back as 2009. In 2010, a Taser spokesman told GQ that Axon would turn "every cop [into] RoboCop."
That Taser spokesman didn't respond to a request to speak about the company's facial recognition plans. But Taser is not alone.
"All of a sudden, the simple act of walking past a police officer becomes a law enforcement interaction."
"You've already got the ability to use cameras to tap into databases to find the license plates of stolen vehicles and overdue parking tickets," said Stan Ross, CEO of Digital Ally, one of a growing number of companies fighting for market share in the fast-growing body camera industry. The business case for facial recognition is obvious, he told me. Cops and police chiefs who are aware of facial recognition "are really excited to try it."
"Why wouldn't we be pushing to bring that technology to the next level?" he said.
Robert Vanman of WatchGuard—another body camera competitor—had similar thoughts. "In regards to facial recognition, WatchGuard will certainly be deploying that technology in the future," he said. "We are the clear technology leader in hardware, and we plan to keep it that way."
But Vanman brought the discussion down to earth.
"Facial recognition will require enough pixel resolution to be effective (to get good recognition results the image needs to contain about 50 pixels between the eyes)," he wrote. "To run facial recognition algorithms in real time will require substantial processing power and an on-camera database (which will require frequent updating). Those elements work against the battery life needs."
So there are practical challenges—video resolution that isn't yet crisp enough; and battery life that isn't yet long enough. Not to mention that some police departments can't even get decent enough internet speeds to download their body cam footage to in-house servers, let alone livestream them to the cloud.
But one can imagine a future when those hurdles are cleared.
Clare Garvie, a legal fellow who studies facial recognition technology at Georgetown, emphasized that real-time biometric searches can be done right now.
Mobile fingerprint scanners used in some jurisdictions make it possible for a cop to identify someone immediately during a traffic stop, for example. And since resolution capabilities are much higher with photos than they are with video (and since posing for an officer during a traffic stop would create a photo similar to a mugshot), facial recognition is much more feasible if police departments use mobile mugshot databases like the ones offered by companies such as DataWorks.
But both of these options involve police identifying someone who has already been stopped.
"With body cameras, that entire structure is completely gone," Garvie said. "Instead of having a particularized interaction—again, in the future, assuming the technological capabilities are there—this is not particularized at all. It's essentially running a search on everybody walking past a given officer on his patrol. There's no notice given to these people, and there's certainly no consent. And there's no police interaction even in place. No probable cause for a search."
She continued: "All of a sudden, the simple act of walking past a police officer becomes a law enforcement interaction."
Most police departments haven't even begun to think this through. Have a look at the Body Worn Camera Scorecard and you'll notice that Baltimore is the only major police department in the country that's addressed concerns about facial recognition. Oregon, too, has passed rules prohibiting "the use of facial recognition or other biometric matching technology to analyze recordings obtained through the use of [body worn cameras]." But there are about 18,000 active police departments in the US, and Baltimore and Oregon don't cover many of them.
Which is to say there are more questions than answers about police using facial recognition tech. One thing's for sure, however, according to Andrew G. Ferguson—a law professor at the University of the District of Columbia: "Body worn cameras with facial recognition technology will alter police tactics," he said. "It, like much of the impact of big data policing, will encourage more aggressive conduct with suspects with prior records or histories of violence, and likely encourage less aggressive interactions with people without prior criminal conduct."
At the same time, Ferguson has written in the past that there's a possible upside as well. "Police perceive ambiguous actions as suspicious because of subtle cues or instincts," Ferguson wrote in a 2014 paper. "These judgments also unfortunately include explicit and implicit biases, policing traditions, and the frailties of human perception.Replacing those generalized intuitions with precise detail about actual people"—such as whether their face is connected to an active warrant—"should result in a more accurate policing strategy."
But that's only if the technology gets there. As of now? It's more dream than reality.