A Vision-Centric Approach for Static Map Element Annotation

TL;DR

CAMA is a vision-based framework that generates high-quality, spatial-temporally consistent 3D annotations for static map elements without LiDAR, improving accuracy over existing datasets.

Contribution

It introduces a novel vision-centric method for accurate, consistent 3D map annotations without relying on LiDAR, enhancing data quality for autonomous driving.

Findings

Lower reprojection errors with CAMA annotations (e.g., 4.73 vs. 8.03 pixels)

Achieves high spatial-temporal consistency in annotations

Enhances training data quality for autonomous vehicle perception

Abstract

The recent development of online static map element (a.k.a. HD Map) construction algorithms has raised a vast demand for data with ground truth annotations. However, available public datasets currently cannot provide high-quality training data regarding consistency and accuracy. To this end, we present CAMA: a vision-centric approach for Consistent and Accurate Map Annotation. Without LiDAR inputs, our proposed framework can still generate high-quality 3D annotations of static map elements. Specifically, the annotation can achieve high reprojection accuracy across all surrounding cameras and is spatial-temporal consistent across the whole sequence. We apply our proposed framework to the popular nuScenes dataset to provide efficient and highly accurate annotations. Compared with the original nuScenes static map element, models trained with annotations from CAMA achieve lower reprojection errors (e.g., 4.73 vs. 8.03 pixels).