An Italian startup fashion company has put together the ‘Manifesto Collection’. This collection consists of knitted pieces that have the ability to screen or ‘hide’ the person wearing them from facial recognition cameras.
It is not by chance. It was designed for this purpose but it is not going to keep you from your fair go login. Take the following situation: A man with red hair, wearing a brightly colored Christmas-like sweater stands in front of the camera. The facial recognition camera identifies him as a giraffe! How is this possible?
The Manifesto collection
This sweater is part of the new Manifesto collection, being presented by the Italian startup, Cap_able. This is their debut collection and includes not only sweaters but many other garments such as pants, t-shirts and hoodies. All these garments share one thing; they all have an “adversarial patch”.
These patches have been designed using artificial intelligence algorithms and have the ability to cause the facial recognition software to make mistakes. The camera has difficulty identifying the person or mistakes them for the animal, a dog, giraffe or whichever animal is set into the pattern on the patch.
The co-founder and CEO of the company, Rachele Didero says “When I’m in front of a camera, I don’t have a choice of whether I give it my data or not. So, we’re creating garments that can give you the possibility of making this choice. We’re not trying to be subversive.”
The birth of an idea
The 29-year-old is currently studying for her PhD in Textile and Machine Learning for Privacy in Milano. She says that she first came up with the idea for Cap_able while studying on a Masters exchange in New York at the Fashion Institute of Technology.
During her time in New York, she heard about a group of tenants fighting their landlord over his plans to put in facial recognition cameras in the entrance to their building. Didero said she had never heard of these cameras before but she said “One of my friends was a computer science engineer, so together we said ‘This is a problem and maybe we can merge fashion design and computer science to create something you can wear everyday to protect your data’.”
Apparently coming up with the idea was not difficult, far more complicated was finding and designing the perfect “adversarial algorithms” in order to make the idea work: to design and create images that would confuse the facial recognition software.
They had two options: one was to create the image, say of a zebra and then adjust it using the algorithm. The second was to choose the size colors and image they wanted and then it would be created by the algorithm. Didero says “You need a mindset in between engineering and fashion.”
It didn’t matter which of the two they chose, it was still necessary to test the images on YOLO, an object detection system.
The process is now patented and involves creating a physical example of the pattern. It is made on a computerized knitting machine, similar to a loom. Once they had reached the desired look and had tweaked the size wanted and the place where they wanted the image to appear on the garment, they could go ahead an create the collection. The garments are made from Egyptian cotton and are all made in Italy.
Currently the garments with these images work 60% to 90% of the time when they are tested using YOLO. However, facial recognition software is advancing all the time and therefore the company’s adversarial algorithms will also need to be improved.
According to Brent Mittelstadt, director of research and associated professor at the Oxford Internet Institute “It’s an arms race.” Mittelstadt compares it to the conflict between software that creates deep fakes and the software created to find them. But the difference being that garments cannot download updates.
He says “It may be that you purchase it, and then it’s only good for a year, or two years or five years, or however long it’s going to take to actually improve the system to such a degree where it would ignore the approach being used to fool them in the first place.”
However, these garments are not cheap; they start at around $300 so it could be that they will remain a luxury product that many can’t afford.
But the fact that they are available may create more awareness of issues surrounding privacy. Woodrow Hartzog, professor at Boston University School of Law thinks “One of the key advantages is it helps create a stigma around surveillance, which is really important to encourage lawmakers to create meaningful rules, so the public can more intuitively resist really corrosive and dangerous kinds of surveillance.”
Others are doing the same or similar
The use of design has been used recently by another company to protect the privacy of individuals for instance during the World Cup in Qatar. Virtue Worldwide created flag-themed face paint so that fans could evade the facial recognition cameras.
Another example is Adam Harvey, a Berlin artist who concentrates his art on data, privacy, surveillance and computer vision. He has created makeup, clothing items and apps with the ability to enhance privacy. Having this in mind he created Hyperface in 2016, a textile which has ‘false-face computer vision camouflage patterns.’
According to Shira Rivnai Bahir, lecturer at the Data, Government and Democracy program at Israel’s Reichman University, “It’s a fight, and the most important aspect is that this fight is not over. When we go to protests on the street, even if it doesn’t fully protect us, it gives us more confidence, or a way of thinking that we are not fully giving ourselves to the cameras.”
In Hong Kong protesters use umbrellas as well as masks and lasers to resist the cameras but the authorities often confiscate them. The pattern on a person’s sweater may be a little more difficult to contend with.
Last year Cap_able initiated a Kickstarter campaign which brought in 5000 Euros. The latest plan is to participate in the Politecnico’s accelerator program and then later in the year to attract investors. Didero says that when wearing the garments, people often remark on how “cool” they are. But she does say “Maybe that’s because I live in Milan or New York, where it’s not the craziest thing!”
‘Quieter’ images and patterns are being considered but that will still have the ability to confuse the cameras. Perhaps Cap_able clothing can be used in countries like Iran where it is said the authorities may begin using cameras to detect women outside without wearing the hijab. Perhaps they can be tricked into seeing zebras and giraffes instead.