Italian fashion start-up Cap_able has launched the Manifesto Collection, a line of knitted clothing designed to protect individuals’ biometric data without concealing their faces.
Using AI algorithms, the collection features various patterns, known as adversarial patches, that deceive real-time facial recognition software. Cap_able’s garments were tested with YOLO, the fastest real-time object detection system. The results were impressive, with wearers either remaining undetectable or being identified as animals due to the embedded animal prints in the adversarial patches. The technology allows individuals to avoid recognition by the software, which instead identifies dogs, zebras, or giraffes within the fabric.
The Manifesto Collection is the brainchild of Cap_able’s founders, Rachele Didero and Federica Busani. Over nine months of research, they explored various images, algorithms, knitting machines, and materials. Didero developed a system to convert digital adversarial patches into 3D knitted textiles using a single yarn, ensuring precise control over the knit and maintaining the patch’s adversarial properties.
Beyond its cutting-edge design, Cap_able’s collection aims to spark conversations about protecting individuals from the non-ethical use of facial recognition cameras. The garments address concerns regarding personal expression and movement in public spaces, empowering wearers to safeguard their privacy. By merging fashion and technology, Cap_able encourages a dialogue on ethical considerations surrounding facial recognition, ultimately championing personal freedom and privacy.
ADD WORKGROUP