03/2018 – 02/2020
Objective of the project: To demonstrate a video surveillance system that is discreet enough to make innocent people feel as free as they would if no digital cameras were in place.
The strained public security situation calls for increased video surveillance in public spaces. Even the technical possibilities of facial recognition are being considered to facilitate automatic recognition of suspects designated as “potential offenders.” However, storing and processing such data without cause is contentious under data protection law. People act differently when they know they may be being watched. The more naturally investigation techniques can be used, the greater the effect on people’s behavior. The project investigates how facial recognition in public spaces needs to be technically designed so that such data is only collected on specific occasions and innocent people can feel as free as they would if no digital cameras were in place.
The overall objective of the project is to investigate whether video surveillance with facial recognition in public spaces is technically possible without violating people’s basic rights. If it turns out that it is possible in principle to build such systems, this would pave the way for significant improvements in public safety, especially in airports, railway stations, tourist hotspots, etc.
One aim is to facilitate, from the images generated by the surveillance cameras, automatic recognition of the faces of suspicious individuals whose names and characteristic data are on a blacklist for legally justified reasons. This would be followed by alerting of the competent authorities and instigation of escalation measures ranging from simple people screening up to evacuations. Here, the data must be processed in a technically sealed environment so that no data from this infrastructure other than the alarms can be leaked, not even to the operators of these systems. It must be possible to demonstrate this capacity in a transparent manner.
The technical research objectives are to clarify the modalities and the effectiveness of this technical sealing and to explore ways in which errors in the identification of individuals (false positives) can be made so unlikely that the privacy advances achieved through the sealing are not invalidated by the revelations that occur in false alarms.
As a result, it is expected that the exact conditions under which operation of e-freedom is possible will be found. To this end, the technological approaches to facial recognition combined multimodally with gait recognition, different camera and lighting settings, and sealing technology are being investigated. The solution is to minimize the false positive rate.
Signs like the example shown are intended to allay the concerns of innocent people as far as possible.
The project is investigating the extent to which technical protection is possible and can create the required level of trust.