A red-haired man, wearing what appears to be the ultimate Christmas sweater, approaches the camera. A yellow quadrant surrounds him. Facial recognition software instantly identifies the man… a giraffe?
This case of false identity is not accidental, it is literally intentional. This sweater is part of the debut his Manifesto collection by Italian startup Cap_able. In addition to tops, there are hoodies, pants, T-shirts, dresses, and more. Each has patterns known as “adversarial patches” designed by artificial intelligence algorithms to confuse facial recognition software. Either the camera cannot identify the wearer, or he is one of the giraffes, zebras, dogs, or other animals embedded in the pattern.
“When you’re in front of a camera, you don’t have a choice whether to pass your data to the camera or not.” I’m not trying to destroy it.”
Didero, 29, who is studying for his PhD in ‘Textiles and Machine Learning for Privacy’ at the Polytechnic University of Milan, worked in MIT’s Media Lab, but the idea for Cap_able came from fashion. He says he came up with the idea while studying abroad for his master’s degree at the Institute. New York technology. While there, she read about how a tenant in Brooklyn fought back against a landlord’s plan to install a facial recognition entry system in the building.
“This was the first time I heard about facial recognition,” she says. “One of my friends’ girlfriends was a computer science engineer, so together we said, ‘This is a problem. Blending fashion design and computer science to create something you can wear every day to protect your data.’ I might be able to,” he said. ”
Coming up with the idea was the easy part. To do that, they first had to find, and later design, a suitable “adversarial algorithm” that would help create images that would fool facial recognition software. For example, create an image of a giraffe and use an algorithm to adjust it. Or set the color, size and shape of an image or pattern and let the algorithm create it.
“We need a mindset between engineering and fashion,” explains Diderot.
Whichever route I chose, I had to test the images with a well-known object detection system called YOLO, one of the most commonly used algorithms in facial recognition software.
Our current patented process uses a computerized knitwear machine to create a physical version of the pattern. It looks like a cross between a loom and a giant barbecue. With a few tweaks here and there, we were able to achieve the desired look, size, and position of the clothing image, creating a range made entirely in Italy from Egyptian cotton.
Didero Says Current Clothing Works 60% to 90% chance when tested with YOLO. Cap_able’s adversarial algorithm will improve, but so will the software it’s trying to cheat with, possibly even faster.
“This is an arms race,” says Brent Mittelstadt, research director and associate professor at the Oxford Internet Institute. He likens it to the battle between software that generates deep fakes and software designed to detect them. However, clothing cannot download updates.
“You could buy it and it might only last you a year, two years, five years, and no matter how long it takes to actually improve your system, you’re going to ignore the previous approach. You might be cheating them in the first place,” he said.
And with prices starting at $300, he points out, these clothes could become just a niche product.
But their impact can be more than protecting the privacy of those who buy and wear them.
“One of the key benefits is that it helps create a stigma around surveillance, which is very important to encourage legislators to create meaningful rules. It allows you to intuitively resist a very corrosive and dangerous kind of surveillance.”Boston University School of Law.
Cap_able isn’t the first initiative to merge privacy protection and design. At the recent World Cup in Qatar, creative agency Virtue Worldwide devised a flag-themed face painting for fans trying to trick the emirate’s numerous facial recognition cameras.
With a focus on data, privacy, surveillance, and computer vision, Berlin-based artist Adam Harvey has designed cosmetics, clothing, and apps aimed at enhancing privacy. In 2016, he created his Hyperface, a textile that incorporates “computer vision of a fake face his camouflage pattern.” This could be considered an artistic precursor to what Cap_able currently intends to do commercially.
“This is a battle, and the most important aspect is that this battle is not over,” says Shira Rivnai Bahir, a lecturer in the Data, Government and Democracy Program at Reichman University in Israel. “When we participate in street protests, even if it doesn’t protect us completely, it gives us confidence and gives us the idea that we are not completely surrendering ourselves to the camera. They will help you.”
Rivnai Bahir, who is about to submit a doctoral dissertation exploring the role of anonymity and secrecy in digital activism, said the Hong Kong protesters’ use of umbrellas, masks and lasers was more than people’s response to the rise of violence was. It is listed as part of the analog method. machine. However, these are easily discovered and confiscated by the authorities. It can be difficult to do the same thing based on someone’s sweater pattern.
Cap_able launched a Kickstarter campaign late last year. Raised 5,000 euros. The company is now participating in Politecnico’s accelerator program to refine its business model and plans to pitch it to investors later this year.
When Didero got dressed, she said people commented on her “cool” outfit, admitting:
Fortunately, a more modest range is on the horizon with patterns that are less visible to the human eye but that can confuse cameras. Flying under the radar means China, where facial recognition has been a key part of its efforts to identify Uyghurs in the northwest of Xinjiang, and Iran, which reportedly plans to use it. It could also help competently dressed people to evade sanctions from authorities in places like Identify women not wearing hijabs on subways.
Big Brother’s eyes may become more ubiquitous than ever, but perhaps in the future, you’ll see giraffes and zebras instead of you.