Background Anonymization for Privacy Preservation

This project combines research in computer vision and artificial intelligence with a focus on regulatory compliance, usability and technology transfer, strengthening the security, ethics and autonomy of the TRAK team.

BARRIER has made a valuable technological, scientific and strategic contribution to TRAK's remote rehabilitation solution. As an overall summary of the project, specialised technical advice has been provided, enabling the design and integration of a visual anonymisation architecture in both the edge environment (real-time web interface) and the cloud environment (batch processing of recordings), providing its remote rehabilitation platform with a greater degree of privacy and reliability. 

These solutions have been developed without compromising existing functionalities, ensuring the compatibility and efficiency of the system. The different areas in which this project has brought tangible value are detailed below: 

• In the field of research and development, VICOMTECH has transferred advanced knowledge in computer vision and privacy preservation techniques, resulting in the development of a functional visual anonymisation prototype. This prototype combines pose estimation and artificial intelligence technologies to automatically blur faces and the background of the scene, keeping only the parts relevant to rehabilitation visible. In addition, a validation protocol was designed and implemented with end users, which made it possible to evaluate the performance, usability and robustness of the system, as well as to collect valuable information for its future optimisation. 
• From a legal and ethical perspective, the project has contributed to better compliance with the General Data Protection Regulation (GDPR) and the European Artificial Intelligence Regulation (AI Act) by TRAK's solution, helping the developed technology to comply with the principles of privacy, transparency and responsible use of AI, and reinforcing TRAK as a reliable and ethical provider. 
• A knowledge transfer process has also been carried out with TRAK's technical team, including detailed documentation, code integration guidelines and training sessions. This transfer has enabled TRAK to acquire technological autonomy to maintain and evolve the anonymisation modules, consolidating its internal development capabilities.
• Finally, the collaboration has had a strategic impact for TRAK, providing its product with an additional layer of privacy that increases user confidence and facilitates its entry into regulated markets in the European healthcare sector.

Vicomtech's role:

Vicomtech has played a key role as a technological and scientific partner in the BARRIER project, contributing its expertise in computer vision and artificial intelligence to develop and integrate visual anonymisation solutions that reinforce privacy and regulatory compliance in TRAK's remote rehabilitation platform. Our contribution has ranged from the design and validation of a functional prototype to the transfer of knowledge and training of the TRAK team, ensuring the compatibility, efficiency and technological sustainability of the system, as well as a strategic impact by increasing confidence and acceptance of the product in the European healthcare sector.

This project has been supported by the Ministry of Industry, Trade and Tourism and the EOI Foundation.