ITEM: As various companies try to find a good business case for AR beyond hunting Pokemons, China’s police force in Zhengzhou has already found one: facial recognition AR glasses to spot criminals – or any individual, really – in a crowd.
According to local media reports, the Zhengzhou police have been testing the glasses at train stations and intend to put them to the test during the upcoming Lunar New Year holiday, which is the busiest travel season in China. The police also claim that the glasses have helped capture seven suspects in ongoing cases, and another 26 people traveling under stolen identities.
According to the Wall Street Journal, the glasses are manufactured by LLVision Technology. The glasses connect to a mobile unit with a pre-installed database of 10,000 “suspects”. The report claims the glasses can match a face to the database in less than one second, though LLVision chief executive Wu Fei cautioned that accuracy can vary due to “environmental noise”.
Although the database for the glasses is limited in the name of portability, eventually the AR glasses will be able to connect to an even bigger resource: the cloud.
According to a report in SCMP last October, the Ministry of Public Security has been working on a facial recognition system since 2015 that aims to add facial recognition capabilities to China’s massive CCTV networks, connecting cameras to cloud data centers and processing centers. The goal: create a system that can identify every Chinese citizen from any CCTV camera with at least 90% accuracy within three seconds. (Note: China plans to expand its CCTV network from around 170 million cameras at the end of last year to 400 million by 2020.)
Interestingly, the pitch for using AR glasses is that there are limitations to using stationary CCTV cameras to ID people, reports The Verge:
One challenge for facial recognition software is that it struggles when running on CCTV cameras, because the picture is blurry and by the time a target is identified they might already have moved on. The sunglasses, by comparison, given police “the ability to check anywhere,” says [LLVision chief] Wu. “By making wearable glasses, with AI on the front end, you get instant and accurate feedback. You can decide right away what the next interaction is going to be.
Obviously the prospect of police using AR to spot criminals raises privacy concerns, although it’s not so much the glasses themselves as the vast surveillance system they’re eventually going to connect to.
A potentially stickier point is accuracy – 90% sounds pretty good until you start thinking in terms of scale. For example, if you scan 100,000 people passing through a train station, that’s 10,000 people who would be potentially misidentified. A 10% failure rate is still a decent-sized window for a real criminal to escape undetected, while police are kept busy tracking people who have been mistakenly identified as a suspect.
And that’s assuming the government’s database is 100% accurate. That’s rarely the case for most if not all governments, though maybe once we reach the era of digital identities verified by blockchain or some other distributed ledger technology, this might be less of a problem. Maybe.
On the other hand, the facial recognition AR glasses look cool as hell. So, you know, that’s something.
In any case, we’re likely to see more of this – not just in China, but other countries as well. Facial recognition has been a dream surveillance tool that law enforcement agencies have been playing with for years. And as the technology improves, you’re going to see more of it, especially as it becomes more prevalent in the consumer and commercial space, from Apple’s FaceID to smart-building security and payment apps.
In China alone, reports SCMP, facial recognition is already being used for things such as university campus security, virtual boarding passes for flights, and even for paying for food at KFC. And then there’s this:
Some other restaurants have even offered discounts to customers based on a machine that ranks their looks according to an algorithm. Customers with “beautiful” characteristics – such as symmetrical features – get better scores than those with noses that are “too big” or “too small” and those that get better scores will get cheaper meals.
Some public lavatories in Beijing also use facial recognition so that the automatic dispensing machines will deny toilet paper to people who ask for it more than once within a given period.
Wonderful things we can do with technology these days, eh?