Locks and Security News: your weekly locks and security industry newsletter
11th November 2026 Issue no. 827

Your industry news - first

We strongly recommend viewing Locks and Security News full size in your web browser. Click our masthead above to visit our website version.

BrighterAI helps MRT bring new European autonomous driving dataset to researchers

AI-powered anonymisation helps Karlsruhe Institute of Technology’s (KIT’s) Institute for Measurement and Control Technology (MRT) protect identities while preserving realistic visual data for autonomous-driving research.

brighterAI, a Milestone Systems company specialising in advanced anonymisation technology, has helped the MRT in Karlsruhe, Germany, prepare a new autonomous-driving dataset for publication by anonymising faces and license plates captured on public roads. 

The project, called KITScenes, provides researchers with real-world European driving data captured using high-resolution cameras, lidar and radar, together with detailed HD maps. MRT, which created the widely used KITTI dataset in 2011, developed KITScenes to provide more current data reflecting advances in sensor technology and the characteristics of European roads.

Publishing real-world imagery in Europe, however, presents significant privacy challenges. Images recorded on public roads can contain identifiable faces and license plates, making anonymisation essential before the data can be released. At the same time, MRT wanted to avoid conventional blurring because altering images too extensively could reduce their usefulness for machine-learning research.

MRT selected brighterAI’s anonymisation technology, which uses inpainting to replace identifiable faces with different faces and original license plates with randomly generated replacements. This preserves a more realistic appearance than conventional blurring while anonymising personally identifiable elements in the images. 

“The data quality was very important for us,” said Fabian Immel, PhD student and research scientist at MRT. “Old datasets didn’t have to anonymise, so those contain real faces. To use the new datasets, the data quality has to be similarly high.” 

Making real-world research data publishable

KITScenes has now been published for non-commercial research use, giving researchers access to contemporary autonomous-driving data recorded on European roads.

In small internal comparisons, MRT showed participants original and anonymised images without identifying which was which. Participants generally could not distinguish the anonymised images from the originals without zooming in significantly or relying on an external clue, such as a license plate that did not correspond to the location.

For MRT, however, the most important result was straightforward.

“The main positive result was that we were able to publish the dataset in accordance with European and German privacy laws,” said Immel.

MRT also valued being able to use an established commercial anonymisation solution rather than developing its own approach. Immel described the collaboration with brighterAI as responsive and frictionless, particularly when the team needed technical assistance adapting the software to its workflow. 

Looking ahead, MRT plans to continue exploring how KITScenes can support autonomous-driving research. Possible directions include research challenges, extensions to the dataset and the fine-tuning of large foundation models, although no specific plans have been finalised.

For more information: https://brighter.ai/resources/case-study-mrt-kitscenes-autonomous-driving-data-anonymization/ 

11th November 2026