Privacy-Preserving Security Cameras based on Metalenses

Informazioni relative al progetto

MetaCam

ID dell’accordo di sovvenzione: 101123231

DOI

10.3030/101123231

Progetto chiuso

Data della firma CE 8 Giugno 2023

Data di avvio 1 Ottobre 2023

Data di completamento 31 Marzo 2025

Finanziato da

• European Research Council (ERC)

Costo totale

Nessun dato

Contributo UE

€ 150 000,00

Coordinato da

FONDAZIONE ISTITUTO ITALIANO DI TECNOLOGIA
Italy

Periodic Reporting for period 1 - MetaCam (Privacy-Preserving Security Cameras based on Metalenses)

Periodo di rendicontazione: 2023-10-01 al 2025-03-31

Sintesi del contesto e degli obiettivi generali del progetto

The MetaCam project addresses the pressing need for privacy-preserving surveillance in an era of growing concern over pervasive monitoring. While surveillance cameras are instrumental in maintaining security in public and private spaces, they often raise ethical and legal questions, especially with the increasing use of AI for video analysis. MetaCam proposes a novel technological solution that transforms the conventional model of video surveillance. Rather than relying on software to de-identify recorded footage—which still necessitates capturing sensitive data—MetaCam introduces a fundamentally different approach: preventing the collection of such data altogether at the hardware level. By integrating purpose-designed metasurface optical elements with tailored image-processing algorithms, the project seeks to enable only the information necessary for activity recognition, such as human pose estimation, while suppressing all personally identifying visual content. This innovation is particularly relevant in the context of GDPR and other data protection frameworks. The ultimate objective is to establish a foundation for next-generation security systems that inherently respect privacy, aiming for applications in smart cities, commercial infrastructures, and other settings where public trust and regulatory compliance are critical.

Lavoro eseguito dall’inizio del progetto fino alla fine del periodo coperto dalla relazione e principali risultati finora ottenuti

Over the course of the project, significant progress was made in both the theoretical design and experimental implementation of a privacy-preserving imaging system. The technical work centered around the co-design of a novel optical metalens and a machine learning model capable of interpreting intentionally distorted visual inputs. A range of simulations were conducted to define the optical characteristics needed to perform the desired image transformations. These were followed by nanofabrication processes that yielded prototype metasurfaces. Parallel to this, neural network models were trained to interpret the distorted imagery and extract relevant pose-related information. The interplay between optical aberration and algorithmic decoding forms the core innovation of the system. A laboratory-scale prototype was assembled and used to demonstrate the feasibility of capturing privacy-preserving data in real-time while retaining performance for human activity recognition. Extensive characterization confirmed the ability of the system to block personal identifiers while maintaining pose estimation capabilities. This work has culminated in a fully functioning proof-of-concept that validates the technical approach and establishes the groundwork for further development toward a deployable product.

Progressi oltre lo stato dell’arte e potenziale impatto previsto (incluso l’impatto socioeconomico e le implicazioni sociali più ampie del progetto fino ad ora)

MetaCam advances the state of the art by proposing a disruptive shift in how surveillance imaging is conceived—moving from software-based anonymization to optically enforced privacy. Existing methods for privacy protection typically involve post-processing, which cannot prevent the capture of sensitive information. MetaCam eliminates this vulnerability by engineering the light path itself through a metasurface lens, creating a secure optical barrier that ensures only anonymized data reaches the sensor. The success of this approach opens a range of commercial and societal possibilities. In particular, it addresses growing demand from stakeholders who must comply with strict privacy regulations while still ensuring public safety. The system has potential for integration into security infrastructures in smart cities, transportation hubs, commercial facilities, and other high-traffic environments. To ensure uptake, the project identified key next steps, including optimizing the prototype for manufacturability, validating the system with potential users, and securing intellectual property protections. While details of the technical implementation remain confidential to protect potential patent filings and publications, the results clearly indicate that this hardware-software co-design strategy represents a new class of secure sensing technologies. Continued development and strategic partnerships will be key to enabling widespread adoption and market impact.

MetaCam prototype