CAPABLE: Engineering, textile, and fashion Collaboration, for citizens' Awareness and Privacy Protection

Abstract

Many private companies and public bodies in authoritarian and democratic states have joined facial recognition technology, used for various purposes. This situation is due to the general absence of a specific regulation that monitors its use. There is no consensus in society regarding the ethics of this technology. Furthermore, there are many doubts concerning the long-term ethical sustainability of facial recognition and its compliance with the law. A problem that emerges from the use of this technology is its obscurity. We do not know who is responsible for the decision automatically made; we do not know how the data is used by those who collect it, how long this data is kept, who can have access to it, to whom it is sent, and how this is used to create a profile. In addition, facial recognition systems are powered by numerous images collected from the Internet and social media without users' permission: it is, therefore, impossible to trace the origin of the data. Consequently, any citizen could be classified, most likely discriminated against, and become the victim of an algorithm. The boundary between security and control is decidedly blurred: many cameras do not respect the privacy of individuals and often harm human rights when they are used to discriminate, accuse, power, and manipulate people. From this discussion on privacy and human rights, it was born first the desire to create awareness, in particular regarding these technologies and the possible issues linked to them. Secondly, it was born the will to create a product that would be the spokesperson for these concerns and allow citizens to protect themselves. On this basis, a collaboration between fashion, engineering, and textile has developed to produce fabric and then garments, which confuse facial recognition systems in real-time. The technological innovation aims to create a system capable of generating adversarial knitted patches that can confuse the systems that capture biometric data. By integrating an adversarial algorithm into their jacquard motifs, the garments prevent the wearers from being identified, preserving their privacy. The adversarial textile is made with computerized knitting machines. Compared to a printed image, knitwear acquires texture, durability, wearability, and practicability. Furthermore, a knitted fabric allows modifying the single yarn material based on the results and performance we want to obtain. These fabrics have been tested on Yolo, the fastest and most advanced algorithm for real-time object recognition. The project was born in New York in 2019; the first experiments with computerized knitting machines were carried out at the Politecnico di Milano in January 2020. The textile was developed in the workshops of the Shenkar College of Tel Aviv. On February 8, 2021, the patent of the industrial process to produce the adversarial knitted textile was filed, with the patronage of the Politecnico di Milano. Today, the research on this fabric and these thematics has carried on within a Ph.D. that combines human-centric design and engineering.

Keywords: Human-Centric Design, Knitwear, Machine Learning, Data Science, Computerized Knitting Machine, Facial Recognition, Biometric Data, Adversarial Fashion

DOI: 10.54941/ahfe1001536