Tel Aviv
CNN
—
The redheaded man wearing what appears to be the best Christmas sweater approaches the camera. A yellow quadrant surrounds it. Facial recognition software immediately identifies the man as… a giraffe?
This case of mistaken identity is not an accident, it is literally by design. The sweater is part of Italian startup Cap_able’s debut Manifesto collection. In addition to tops, it includes hoodies, pants, t-shirts, and dresses. Each sports a pattern, known as an “adversarial patch,” designed by AI algorithms to confuse facial recognition software: cameras either fail to identify the wearer or think they are a giraffe, zebra, dog, or one of the other animals embedded in the pattern.
“When I’m in front of a camera, I don’t have a choice whether or not to give it my details,” says co-founder and CEO Rachele Didero. “So we are creating garments that can give you the ability to make this choice. We are not trying to be subversive.”
Didero, 29, who is studying for a PhD in “Textile and Machine Learning for Privacy” at Milan Polytechnic, with a stint at MIT’s Media Lab, says the idea for Cap_able came to him when he was on an exchange. from the Fashion Institute of Technology in New York. While there, she read about how tenants in Brooklyn had fought the landlord’s plans to install a facial recognition entry system for his building.
“This was the first time I heard about facial recognition,” she says. “One of my friends was a computer engineer, so together we said, ‘This is a problem and maybe we can merge fashion design and computer science to create something you can use every day to protect your data.’”
Coming up with the idea was the easy part. To make it a reality, they first had to find, and then design, the right “adversarial algorithms” to help them create images that would fool facial recognition software. Or they would create the image, of our giraffe for example, and then use the algorithm to adjust it. Or they set the colors, size, and shape they wanted the image or pattern to take, and then had the algorithm create it.
“You need a mindset between engineering and fashion,” Didero explains.
Regardless of which route they took, they had to test the images on a well-known object detection system called YOLO, one of the most widely used algorithms in facial recognition software.
In a now-proprietary process, they would then create a physical version of the pattern, using a computerized knitting machine, that looks like a cross between a loom and a giant barbecue. A few tweaks here and there to achieve the desired look, size and position of the images on the garment, and then they were able to create their range, all made in Italy, from Egyptian cotton.
Didero says today’s clothes work 60% to 90% of the time when testing with YOLO. Cap_able’s contradictory algorithms will get better, but the software you’re trying to cheat might get better too, perhaps even faster.
“It’s an arms race,” says Brent Mittelstadt, research director and associate professor at the Oxford Internet Institute. He likens it to the battle between the software that produces deepfakes and the software designed to detect them. Except that clothes can’t download updates.
“It could be that you buy it, and then it’s only good for one year, two years or five years, or however long it takes to improve the system to such an extent that it would ignore the approach being used to fool them in the first place,” he said.
And with prices starting at $300, he notes, these clothes may end up simply being a niche product.
However, their impact can go beyond preserving the privacy of those who buy and use them.
“One of the key advantages is that it helps create a stigma around surveillance, which is really important to encourage lawmakers to create meaningful rules, so that the public can more intuitively resist really corrosive types of surveillance. and dangerous,” said Woodrow Hartzog, a professor at Boston University Law School.
Cap_able is not the first initiative to merge privacy protection and design. At the recent World Cup in Qatar, creative agency Virtue Worldwide came up with a flag-themed face paint for fans looking to fool the emirate’s legion of facial recognition cameras.
Adam Harvey, a Berlin-based artist focused on data, privacy, surveillance, and computer vision, has designed makeup, clothing, and apps aimed at enhancing privacy. In 2016, he created Hyperface, a textile that incorporates “false face computer vision camouflage patterns,” and what could qualify as an artistic precursor to what Cap_able is now trying to do commercially.
“It is a fight, and the most important aspect is that this fight is not over,” says Shira Rivnai Bahir, a professor in the Data, Government and Democracy program at Israel’s Reichman University. “When we go to street protests, even if it doesn’t fully protect us, it gives us more confidence, or a way of thinking that we’re not fully surrendering to the cameras.”
Rivnai Bahir, who is about to submit her doctoral thesis exploring the role of anonymity and covert practices in digital activism, cites the use of umbrellas, masks and lasers by Hong Kong protesters as some of the most analogues in which people have fought against the rise of machines. But these are easily detected and confiscated by the authorities. Doing the same thing based on someone’s sweater pattern can be trickier.
Cap_able launched a Kickstarter campaign late last year. He raised €5,000. The company now plans to join the Politécnico’s acceleration program, to refine its business model, before launching investors later in the year.
When Didero wears the garments, he says people comment on his “cool” clothes, before admitting, “Maybe it’s because I live in Milan or New York, where it’s not the craziest thing!”
Fortunately, more demure ranges are ahead, with patterns that are less visible to the human eye, but can still confuse cameras. Flying under the radar can also help Cap_able-clad individuals avoid penalty from authorities in places like China, where facial recognition was a key part of efforts to identify Uyghurs in the northwestern region of Xinjiang, or Iran, which reportedly plans to use it. to identify women without hijabs in the subway.
Big Brother’s eyes may become more and more ubiquitous, but maybe in the future he’ll see giraffes and zebras instead of you.
Source: news.google.com