HRDX: A Large-Scale Vector HD-Map Dataset

HRDX is a newly released large-scale dataset for vector HD-map construction, comprising approximately 40 hours of driving data (1,400 km) captured with six synchronized cameras, a 128-beam LiDAR, and centimeter-level RTK GNSS/IMU. It includes annotations for 10 vector map classes and over 20 semantic attributes, alongside aligned aerial orthoimagery, which enhances geometric map quality during training and inference. This dataset aims to facilitate reproducible research in HD-map learning and multimodal fusion, providing a valuable resource for practitioners in autonomous driving and related fields.