The Italian start-up Cap_able wants to give people the power to choose whether or not to be identified by facial recognition systems.
The latter, very practical for unlocking your phone in an instant or for unlocking a lock with panache, are insinuating themselves more and more into public spaces and becoming an omnipresent surveillance tool. With the risk of jeopardizing the privacy of more and more citizens, this trend is giving rise to ever more inventive counter-attacks.
Force the algorithm to identify the wearer as a giraffe
Cap_able, founded and managed by Rachele Didero in Milan, offers a range of ready-to-wear with an… original design. Dresses, pants, sweaters or t-shirts, it’s a good part of a wardrobe that can go there. If these clothes are not necessarily to everyone’s taste, the goal is above all to deceive the algorithms.
It’s about fooling the cameras by requiring them to identify anything other than a human being. The wearer can thus be assimilated to a giraffe, a zebra, food or any other object, thus preventing any attempt at identification. Didero explains that ” it is the transposition of certain images onto the fabric that is capable of confusing facial recognition algorithms “.
Merge styling and IT
To create the patterns, the startup relied on the popular YOLO facial recognition system to design ” conflicting algorithms “. They can refine a model created by designers from a decoy image, or design one according to predefined parameters of size, color or shape.
However, engineers were forced to push the innovation further. Indeed, after the creation of the patterns comes the manufacture of the clothes, which is more complex than it seems. The Cap_able founder explains that after designing the illustration, ” comes the programming of the weaving machine, which must respect certain criteria in order to be able to reproduce the fabric – which can also be composed of different threads “. This step would be essential to allow their algorithm to be “ integrated » directly into the garment.
The company does not want to stop there, “…we plan to launch a collection of accessories and then apply the technology to a smaller area. We will also work on the material composition of the wires, so that in the future we will be able to consider other types of detection than visual recognition”. However, if the technology is impressive, one wonders how long it will take facial recognition algorithms to find a response.
Source : Wired, Able, Capable on YouTube
6