SoylentNews

-- OriginalOwner_ [tinyurl.com] writes:

The Center for American Progress reports [thinkprogress.org]

Maybe [the Hip-Hop duo] Insane Clown Posse [wikipedia.org] isn't your thing, but if you're headed to a protest any time soon, you might want to consider that familiar black and white makeup [google.com].

Two years ago, a Georgetown Law report [theguardian.com] found that half of all American adults are recorded in police facial recognition databases. The databases are, according to the report, made up of "overwhelmingly...non-criminal entities" and are built on state driver's licenses, passports, and visa applications, giving the government the ability to track more than 117 million adults using unregulated software.

[...]In 2016[1], the American Civil Liberties Union (ACLU) told The Guardian that police in Maryland were using the software to identify protesters from photos and that the database is essentially racist, as it relies on mugshots--disproportionately taken of people of color--that remain in the system, even if the people in the photos weren't charged.

A study from last year found that increasingly intelligent software is even learning how to identify protesters who wear hats and use scarves to cover their faces. The software identifies key parts of a face and makes estimates so the software can recreate a covered face.

[...]Here's where the Juggalos come in.

Basically, as The Outline wrote [theoutline.com] [July 2], a computer science blogger known on Twitter as @tahkion, made a breakthrough earlier this week when he realized that Juggalo makeup makes it essentially impossible for facial recognition software to figure out who a person is.

[...]The power is mostly in the fact that Juggalo makeup usually includes putting black makeup below the mouth but not all the way to the chin, which messes with recognition software trying to find a person\u2019s jaw.

Of course, one imagines facial recognition software might just start recognizing a face in consistent Juggalo makeup, but, as @tahkion noted, many Juggalos make small changes to the style and Juggalos only wear their makeup for special occasions.

i don't know why i'm replying to this with a serious answer, but to summarize it's because the neural network is trained on contrast levels on normal facial landmarks. you would need to either have a user set the landmarks manually or train a network just on juggalo facepaint

-- TAKHION (@tahkion) July 1, 2018 [twitter.com]

As The Outline noted, Juggalo makeup won't trick, say, Apple's Face ID, which relies on depth perception rather than visual light, but for law enforcement software that works from, as @tahkion explained it, "contrast levels on normal facial landmarks", Juggalo makeup will do the trick.

[1] Dup'd link in TFA.

Original Submission