GDPR blurring of faces and number plates | 360GEO

GDPR blurring of faces and number plates as a fixed part of processing

Imagery of the public realm almost always contains people and vehicles. That makes it personal data, and brings it within the scope of the GDPR. At the same time, anyone managing assets does not need those details: assessing a kerb, a traffic sign or a facade does not depend on who happened to be walking past. GEORIZON therefore obscures faces and number plates automatically during processing, before the imagery becomes available in the viewer, the API or a download.

Blurring happens during processing, not as a separate step afterwards
Detection of both faces and number plates, including partial visibility
Irreversible in the published imagery, with the original held separately
Works on imagery from any make of mobile mapping system

Why this is a processing step and not a viewer setting

Blurring applied only in the viewer is false comfort: the underlying image remains retrievable. GEORIZON therefore applies anonymisation to the imagery itself, during processing and before it is made available. What leaves through the viewer, the API or a download is the anonymised image. Detection models look for faces and for number plates, including where these are at an angle, partly behind an object or small in frame. Because panoramic imagery captures the same person or vehicle from several viewpoints, every capture in the survey round is treated, not only the one in which somebody is most clearly visible.

Lawful basis, retention and what you have to arrange yourself

Technology does not settle the legal side. The GDPR requires your organisation to have a basis for capturing and retaining this imagery, usually a public task or a legitimate interest, together with a record of processing, a retention period and clear information for data subjects. Anonymisation is one of the most important measures in that assessment, because it reduces the processing to what you actually need. Municipalities using panoramic imagery for inspection of the public realm follow the same line. We supply the technical measure and its documentation; the lawful basis and the policy remain yours, and we are glad to think along.

Access, retention and deletion

Beyond anonymisation, how the environment is set up determines how much data is in circulation. Every customer has their own tenant, access is arranged per user or through single sign-on, and an on-premises deployment is possible where the data may not leave your own infrastructure. What goes out through the API and downloads follows the same permissions as the viewer, so no back door appears. If you receive a request from a data subject, a capture can be found by location and date and screened off specifically. Organisations including ProRail (SpoorInBeeld), the municipality of Nijmegen and Kempkes Landmeten work with their imagery this way.