Fashion fights facial recognition systems via garish make-up, face-printed T-shirts, punk hair extensions

South China Morning Post Dipublikasikan 04.11, 14/11/2019
Fashion fights facial recognition systems via garish make-up, face-printed T-shirts, punk hair extensions

Make-up is used for a variety of reasons. To look beautiful, younger, older, professional. To conceal something. And, in an era in which we are closely watched, to hide our identity.

In a world of ubiquitous closed-circuit television cameras and facial recognition software, lipstick, blusher and garish face paint can shield us from some of the world's most sophisticated surveillance technology.

Computer Vision Dazzle - or CV Dazzle - was created by Berlin-based artist Adam Harvey in 2010, when the idea of citywide facial recognition camera networks and deep fake technology existed only in the realm of science fiction. Harvey has spent the past decade proving that thick, pigment-rich make-up painted in shapes across the face - with flourishes and swirls reminiscent of Pablo Picasso artworks - can prevent most facial-recognition algorithms accessing the wearer's biometric profile.

In Hong Kong, anti-government protesters have been wearing masks - now banned under emergency laws - and dressing in black to escape the all-seeing eyes of the authorities. Harvey's findings feel more relevant than ever.

Facial recognition software can be fooled with the strategic application of make-up. Photo: courtesy of CV Dazzle

Fittingly, CV Dazzle is named after an effective camouflage technique developed during World War I, when navies sought to protect their ships by obscuring their size and shape with sophisticated painting techniques that confused the human eye.

The idea behind Harvey's CV Dazzle is similar. Facial recognition algorithms look for patterns of light and darkness across the forehead, cheekbones and chin, and the way colour is distributed across the face. No two people have the exact same rise and fall of shadow and colour distribution. Cover them up, Harvey realised, and algorithms won't be able to separate a face from any other swathe of pixels. Only the human eye is up to that task.

A young Serbian rights group activist has her face painted to confuse Huawei surveillance video cameras with face-recognition software installed in Belgrade. Photo: AP/Darko Vojinovic

"People have been doing this (avoiding recognition) for years: wearing sunglasses, hats, or hoodies to create a veil of anonymity. This is just a more modern version of it," says Tamara Cincik, an analyst specialising in fashion branding.

"A huge part of the popularity around streetwear has been the fact that it can conceal the face of the user with ease and in a fashionable way. One of the most important new areas of subversion is definitely surveillance and, in particular, facial recognition, so I'm not surprised we're moving in this direction."

The entire project began in a more innocent era, when Harvey was trying to create a way for him and his friends to be free of the suddenly ubiquitous Facebook auto tag, which identified fellow users in a photo post without consent.

You can get creative with your camouflage. Photo: courtesy of CV Dazzle/Cha Hyun Seok

What, Harvey wondered, would you do if you didn't want someone to know where you were, but Facebook ratted you out? A decade later, and Mark Zuckerberg's knowledge of our whereabouts seems less important in light of governments and even brands using increasingly sophisticated technology to track citizens and potential clients.

Harvey has now trained a group of make-up artists to perfectly disguise their clients, but anyone can use these techniques to obscure the symmetry, complexion tone, and skin texture of their own faces. In simple terms, that means applying make-up that contrasts with your natural skin tone and using primary colours at the point where the nose, eyes, and forehead intersect, as this is the area facial recognition cameras rely on.

If your make-up skills are not yet cubist-grade, attaching a multicoloured hairpiece in the shape of a fringe or a punk-like spike can interfere with that all-important symmetry, especially if it obscures the top half of the face.

Last year, Harvey upgraded his Dazzle initiative from make-up to fashion. The 2018 HyperFace Project involved printing facial patterns on T-shirts, jumpers and dresses: photographs of eyes, mouths and noses were designed to confuse software that interprets it as a face.

"The aim of the project is to overload an algorithm with what it wants, oversaturating an area with faces to divert the gaze of the computer vision algorithm," Harvey said at a conference in Hamburg, Germany, last year.

"Surveillance, unlike fashion, thrives on conformity. Fashion thrives on being unique. It is this point between fashion and surveillance that I'm interested in."

Unlike the human brain, facial recognition software can be tricked quite easily. Photo: AP/Darko Vojinovic

The only problem with the HyperFace techniques is that the act of becoming invisible to algorithms renders a person almost absurd to the conscious eye, caked in scarlet, silver and deep-blue make-up in the shape of lightning flashes and triangles. Or with garish human faces printed on the chest.

One marginally less obvious option by Harvey is Stealthwear, which guards against thermal imagery and has been used by the military for many years. It hides body warmth from any cameras that use heat detection as a form of surveillance.

The trend is now catching on around the globe. Last month, a group of artists in London known as The Dazzle Club walked around the British capital with primary-coloured stripes painted across their faces in an effort not only to escape the cameras that are so ubiquitous in London - the sixth most-watched city on Earth - but also to publicise their cause. Their walks take place in different parts of the city and are aimed at educating people on the growing use of facial-recognition technology in public spaces.

Juggalos march on Washington wearing camouflage make-up. Photo: Shutterstock

Diehard fans of the rap group Insane Clown Posse, also known as Juggalos or Jugalettes, have put Harvey's findings to good use. They have marched on Washington a few times to protest against US President Donald Trump and did it coated in make-up designed to obscure them from security cameras.

Their technique, which was explained in a series of tweets, focused on the lower half of the face by applying a thick circle of black paint below the mouth and above the chin, making it difficult for facial-recognition technology to identify the shape of the jaw.

In a more amusing turn, one group of research scientists in the US has created an infrared light-projecting baseball cap that can fool a facial-recognition system into thinking the wearer is the musician Moby. Using tiny infrared LEDs wired to the cap, the researchers were able to project dots of light onto the wearer's face in a way that can not only obscure their identity but also "impersonate a different person to pass facial recognition-based authentication".

It may look strange, but this make-up is enough to spoof recognition algorithms. Photo: Shutterstock

Unfortunately for the many who could find it useful, the cap is not for sale, However its invention does mark an important, and frightening, step in the direction of identity fraud.

The high street is also beginning to take notice. E-commerce store Adversarial Fashion now sells shirts, skirts and jumpers covered with fake licence plates that are designed to confuse traffic surveillance cameras by flooding the system with false data in a similar way to HyperFace.

"It could be a major trend of the future," says Liz Flora, the Asia editor of intelligence firm L2. "Even high-end luxury brands are getting in on the act and translating it onto the runways. Fendi's Jackson Wang collection featuring a series of face masks comes to mind."

Perhaps one day soon, making up your face to look like a Picasso painting - or a tribal leader from Papua New Guinea - will be as ubiquitous as wearing a hat or a hoodie. Blending into the crowd just got a lot more garish.

Copyright (c) 2019. South China Morning Post Publishers Ltd. All rights reserved.

Artikel Asli