1 minute read
3.1 Non-Intrusive Surveillance Systems - Norway
Videos and images collected by law enforcement through surveillance systems, such as closed-circuit television (CCTV) and body or car cameras, are often essential for law enforcement to prevent and investigate crimes and secure the prosecution of offenders within the court system. At the same time, however, images, videos and sounds contain information that can disclose the identity of individuals, the handling and use of which may present concerns regarding privacy.
In aiming for transparency, the Oslo Police District of Norway has been working with partners both within the police force and externally with industry and academia to explore the application of AI for the creation of heavily user-sensitive non-intrusive surveillance systems that can be employed in smart cities. The anonymization of videos and images collected using AI is at the core of this. More specifically, the images of people captured in surveillance footage are automatically anonymized by the AI system, by covering their face with a cartoon character or emoji. This enables the anonymous datasets to be handled and shared by the police and with police partners for use in a non-intrusive manner, for example, the anonymized data can be freely used for pattern recognition to identify acts such as vandalism, street fighting and movements that indicate intoxication. This project marks one of the first trials by law enforcement in which pattern recognition is combined with means of automated anonymization. This pilot is also notable in that it could even present a partial solution to addressing compliance with the General Data Protection Regulation of the European Union (EU), whereby law enforcement is required to be able to perform non-intrusive monitoring and evidence analysis with regards to privacy of the ‘bystanders’ or non-person of interest (non-POI). 15
While non-intrusive surveillance presents opportunities for law enforcement and society to rethink surveillance, it nevertheless still requires careful consideration of privacy, data protection, de-anonymization and the practicalities of the use of anonymized data by law enforcement.
15 See GDPR, inter alia, Article 5(1)(c), which sets the ‘data minimisation’ principle. The full text of the GDPR is available at https://gdpr-info.eu