Training object detection algorithms for perception in autonomous driving is a complicated task that requires large amounts of labeled data. Recently the IES proposed a novel concept at the Interactive Adaptive Learning Workshop ECMLPKDD 2021. In the paper “A Concept for Highly Automated Pre-Labeling via Cross-Domain Label Transfer for Perception in Autonomous Driving” Maarten Bieshaar, Marek Herde, Denis Huseljic, and Bernhard Sick propose an innovative concept to leverage labeling tasks in autonomous driving and to transfer labels between different sensors.

The video of the paper presentation can be found online: https://youtu.be/BTk8o1IFG24 Paper: https://www.activeml.net/ial2021/pdf/ialatecml_paper5.pdf

This work originates from the KI-Data Tooling project (part of the KI-Familie).