GOLD-BEV: GrOund and aeriaL Data for Dense Semantic BEV Mapping of Dynamic Scenes

TL;DR

GOLD-BEV is a framework that learns dense, scene-centric BEV maps of dynamic road scenes using aerial imagery supervision during training, enabling scalable and consistent semantic mapping.

Contribution

It introduces a novel approach combining aerial imagery supervision with ego-centric sensors for dense semantic BEV mapping of dynamic scenes.

Findings

Achieves dense semantic mapping with minimal manual annotation.

Effectively supervises moving traffic participants using synchronized aerial data.

Supports pseudo-labeling and synthesis of aerial views from ego sensors.

Abstract

Understanding road scenes in a geometrically consistent, scene-centric representation is crucial for planning and mapping. We present GOLD-BEV, a framework that learns dense bird's-eye-view (BEV) semantic environment maps-including dynamic agents-from ego-centric sensors, using time-synchronized aerial imagery as supervision only during training. BEV-aligned aerial crops provide an intuitive target space, enabling dense semantic annotation with minimal manual effort and avoiding the ambiguity of ego-only BEV labeling. Crucially, strict aerial-ground synchronization allows overhead observations to supervise moving traffic participants and mitigates the temporal inconsistencies inherent to non-synchronized overhead sources. To obtain scalable dense targets, we generate BEV pseudo-labels using domain-adapted aerial teachers, and jointly train BEV segmentation with optional pseudo-aerial BEV reconstruction for interpretability. Finally, we extend beyond aerial coverage by learning to synthesize pseudo-aerial BEV images from ego sensors, which support lightweight human annotation and uncertainty-aware pseudo-labeling on unlabeled drives.