Facial recognition technology is old news. It’s widespread enough to be in every iPhone 10 and it tags our pictures on Facebook. Most facial recognition tech works from a fairly narrow range of options, though – limited possibilities of who might be in a picture or who a device belongs to. What’s new in facial recognition is high speed processing and the ability to use AI to identify individuals from an enormous pool of possibilities.
The Recognition Revolution
China has been in the news recently for the country’s use of facial recognition technology in Xinjiang province. Located near the border and home to a large number of Uighur minorities, the technology aims to identify people on state watch lists who might be roaming the city. Such surveillance is unfortunately unsurprising in China, where the government restricts internet access and public discourse to an extreme degree, but it also suggests exactly what’s possible – and it could be coming your way.
Several different technological improvements serve as the foundation for AI-driven facial recognition, the first of which is higher quality video. Unsurprisingly, old video is notoriously hard to analyze since it’s pixelated and poorly defined. Modern CCTV cameras, though, provide a minimum 1080P HD video for surveillance footage that looks as good as your favorite television show.
Kapil Dhingra, founder of AI platform Pixmettle said “The other major technological improvement supporting facial recognition in surveillance is AI and machine learning. At this point, facial recognition tech doesn’t just scan through a database for potential matches, but can actually learn how to group and efficiently eliminate whole swaths of faces from consideration. It can identify gender and racial markers, color, and more. The problem is, they still aren’t precise – facial recognition isn’t DNA or fingerprinting – and it comes with certain risks.”
Of all the potential real time applications of facial recognition technology, one of the leading contenders in the US right now is as part of police body cameras. Advocates for this technology suggest it could help officers identify suspects more easily based on database information, and consider it a natural extension of the push for body cameras by antiviolence activists. Opponents, on the other hand, suggest that such technology could be used to justify poor judgment and increase violence towards minorities.
A less dramatic use of facial recognition technology, and one more in line with current capacities, recommends employing it as part of business identification and verification strategies. The designers of TrueFace.ai, for example, hope that their software will make it easier for fintech and e-learning groups to verify users, adding to a multi-key process by combining facial data with passwords and challenge questions. As multifactor authorization becomes the norm, why not use something as unchanging as a user’s face?
The more cameras populate our normal environments, the more likely big data companies will draw on them for data – to identify not just how many people walk by an advertisement, for example, but who those individuals are. Companies can identify how often someone comes to a physical storefront, even if they pay in cash. Video data becomes actionable in new and increasingly personal ways.
So, do businesses or the government know where you are? Maybe not – but don’t discount the possibility. Computers could be watching you soon.