b-plus technologies GmbH: Post-Processing Anonymization for Autonomous Driving

The Challenge: The 99% Data Problem

Over 99% of data collected during AD test drives is irrelevant for training. To build robust perception systems, developers must selectively capture only the crucial 1% of edge cases and critical traffic scenarios.

However, storing raw camera data exposes companies to severe GDPR violations. Developers need a reliable way to anonymize personal identifiable information (PII) before using these datasets for training.

The Solution: Automated Post-Processing Anonymization

During the jbData project, b-plus deployed test vehicles equipped with multimodal sensors (cameras, LiDAR) and the b-plus DATALynx ATX4 platform to selectively capture data.

To secure the data pipeline, Celantur Container was integrated as the final anonymization post-processing step.

Celantur's software serves as the final privacy gatekeeper in the b-plus processing chain, automatically blurring faces and license plates to deliver fully GDPR-compliant perception datasets.

Key Results (jbDATA Project)

• Future-Proof Compliance: Enables the legal provision of real-world perception data under strict GDPR requirements.
• Seamless Toolchain Integration: The containerized deployment successfully closed the final loop in b-plus's jbDATA processing pipeline.