Fooling Facial Recognition by Wearing Ugly Clothing, No Mask Necessary

Rachele Didero has an unusual skillset: Multiple degrees in Fashion Design from schools in three different countries, and a Ph.D. in Textiles, Machine Learning and Privacy from the Politecnico di Milano in her native Italy.

Listen beautiful relax classics on our Youtube channel.

The peculiar combination–fashion design skills, textiles knowledge and privacy concerns—has motivated her to create clothing that can defeat facial recognition systems.

Her company, Cap_able, produces a Manifesto Collection of sweaters and pants that we’d read as discordant and ugly; something like a casino carpet you wear on your body. But the patterns have been carefully selected to fool YOLO (You Only Look Once), a neural-network-based Deep Learning algorithm that can identify people and objects.

When you’re wearing a Manifesto garment, YOLO doesn’t even bother looking at your face; instead it assumes (most of the time) that you are an animal.

To be clear, YOLO is just one of many facial recognition systems; it’s the one that Cap_able chose to test against, so your fashion sacrifice isn’t going to work everywhere. Instead “The Manifesto Collection was created to create awareness of the risks associated with the improper use of facial recognition technology,” Cap_able writes.

Here’s Didero explaining Cap_able’s mission:

Source: core77

Listen beautiful relax classics on our Youtube channel.

No votes yet.
Please wait...