Case Study: FERNRIDE accelerates autonomous trucking model training with CVAT Online 3D annotation tools

FERNRIDE accelerates autonomous truck labeling by 30% with CVAT

FERNRIDE, a Munich-based company specializing in autonomous electric trucking, faced significant challenges in labeling 3D LiDAR point cloud data. Their annotation workflow required precise 3D cuboid labeling capabilities for accurate perception model training, but the tools in CVAT Online lacked the necessary precision for rotation control, size input, and orientation indicators, which hindered their progress.

CVAT developed and shipped three custom 3D annotation features for FERNRIDE within five weeks. These new tools provided smoother cuboid rotation, manual size input for centimeter-level precision, and orientation arrows for directional accuracy. As a result of implementing CVAT's solution, FERNRIDE's annotation accuracy improved significantly, their labeling workflows accelerated by 30%, and the higher-quality data directly strengthened their autonomous trucking perception models.

FERNRIDE