Hair and Makeup Outsmarts Computers, Makes You Un-taggable on Social Media
Photo: Adam Harvey / cvdazzle.com
Have you ever been tagged in a Facebook or Instagram photo against your wishes? That all changes now, thanks to a creative new beauty invention. Adam Harvey, an artist, designer, and entrepreneur based in Brooklyn, N.Y., created Computer Vision Dazzle (or CV Dazzle) for his thesis project at New York University’s Interactive Telecommunications Program. By applying black and white makeup or arranging hair in the right pattern combinations it blocks those intuitive facial-recognition algorithms from detecting a face in any photo.
“My goal, back when I first created it in 2010, was to prove that face detection could be blocked using only hair and makeup,” says Harvey. “By reverse engineering a computer vision profile I can understand what the algorithm is looking for to detect a face— symmetry, contours in the cheekbones, and the nose bridge—and obscure those areas and create an ‘anti-face.’” The look is an outrageous one, but Harvey intended it as an eccentric, high-fashion look that could work in a club setting. “An interesting parallel is that fashion is all about staying a season ahead of the latest trend, and surveillance is about staying ahead as well. Fashion by its nature is about not conforming and reinventing itself, and it actually provides a partial answer to some surveillance issues.”
Photo: Adam Harvey / cvdazzle.com
Preventing social media tagging is just one benefit of this invention. When computer vision programs detect a face, they can extract data about your gender, age, emotions, and identity and share this information with everyone from the FBI to advertisers. “Your face print is like your social security number, and when you walk around in public you don’t typically expect people to be accessing something like that,” says Harvey. There’s a program called Janus that collects images in the wild and pieces them into identity profiles and shares them with intelligence agencies. Advertisers are more of a threat to the mainstream, though. “If they’re collecting data on your gender, age, mood, and identity, how do you opt out?”
Here’s where CV Dazzle comes in, to work in the opposite way of traditional makeup. “When you define your mouth, outline your eyes, or contour your cheeks, you’re actually making it easier to detect your face,” says Harvey. “Decontouring will decrease the probability that your face will show up.”
And unlike most elevated beauty looks, this one is dirt cheap for consumers. You can use any makeup as long as you use contrasting shades. If you have light skin or hair, add dark makeup and hair to create patterns, that help prevent computers from spotting you. And if you have dark skin or hair, do the opposite. Harvey works with a makeup artist, but says the best cosmetics to use are those that won’t crack apart on the skin. Metallic or reflective makeup will only further confuse the tracking algorithms. “There’s no one way to wear it.” In fact, it’s better camouflage when people come up with a unique design,” says Harvey. “The irony is, you create patterns, and computers are great at learning patterns.” To that end, he’s been working on a software application that will allow people to create their own designs.
As technology continues to change and become more sophisticated, innovators like Harvey need to keep moving with the times. Early this fall he’ll be launching a surveillance trend report with new fashion items and products, but he’s happy with the progress being made on CV Dazzle. “These algorithms are a multibillion dollar industry, but here’s a $1 solution.”
