If it were possible to evade facial-recognition systems using just subtle makeup, it might look something like this Makeup carefully applied to the forehead, cheeks, and nose may help you evade facial recognition systems, judging from these computer scientists' experimental work. Their described method is a form of adversarial attack, which generally involves subtly tweaking input data to trick machine-learning algorithms into …
"Wearing makeup can change what you look like."
And for some people, *not* wearing makeup may change what they look like.
Example from previous era: Roy Wood (of Wizzard fame); doubtless others are available.
Still, I don't suppose there's much Venture Capital funding for pointing out these things, whereas putting it in a paper might lead eventually to $$$$.
The Police probably won't hassle you at all if the computer thinks you are accompanied by a right wing Home Secretary, although you might be subject to a bit of kow-towing by any senior officers present. You could just wear a T-shirt with her mug on it of course.
Most facial recognition systems start by trying to find faces anywhere in the image, so you could probably have it on tattooed your arse and it would still work, except you would have to walk backwards without any trousers on.
Once upon a time I have what purported to be Richard Nixons Social Security Number memorized. Doing some odd job where they need something that could plausibly be a SSN? Tricky Dick apparently needed the cash!
So having Priti Patel do all kinds of more or less nefarious things sounds like an interesting notion.
This post has been deleted by its author
If anyone complains tell them it's what you wear instead of a mere paper mouth/nose mask on the grounds that you want to be extra cautious. Apply one of those holographic stickers to the inside of the face plate so that you can see out just fine but anyone looking in only sees the sticker. I like the one of the rotting, maggot infested, zombie as it tends to cause revulsion in nosy buggers. Bonus is that you can remove it for identification purposes (at the bank, for an official photo, to get the cop to stop yammering, etc) but then put it back on the moment such need is done, and almost nobody will dare to tell you to remove it because that would be openly telling someone to remove a face mask during a plague. Now if you'll excuse me I have some eerie green chemical glowsticks to purchase & stick in the inside edges of my mask so I can ask folks if they're my mummy...
If you watch makeup-tutorial YouTube, you will find lots of videos about "contouring" - using different colours of makeup to make your face look like a different shape to what it actually is.
That wouldn't fool a 3d camera, but most of these facial recognition systems don't use 3d cameras.
Unless you wear a different pattern of makeup every time you go out, the AI will simply learn your makeup pattern as another identity and carry on tracking it. It may not immediately connect that identity to "you", but it wouldn't take much for either a human or an AI to make the connection.
To write UNKNOWN on your forehead with mascara (which is guanine, made from fish scales), then the AI knows that nothing fishy is going on, and just lets you pass.
It is a bit like a social engineer in the future walking, no strolling at a leisurely pace, into a high security AI protected area wearing a T-shirt that says "Computer Upgrades" - every door just automagically opens for them.
https://imgur.com/a/5kJi3yR
I'm still not seeing how that would work, AI scans QR code, it sees text. Are saying that the AI is going to run the line of text in an interactive shell on the production environment, that should never happen, unless there is a bug or two.
FaceID works by measuring the distance between the camera and your face, with about 300,000 pixels total. Seems to be totally unaffected by makeup unless you apply filler :-) Should also be fine with different skin colours.
Just haven’t heard of any commercial source for that camera. And it can’t be trained with photos.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
