PandaSet, a set of data sets for autonomous driving was made available on GitHub by sensor manufacturer Hesai & start-up Scale AI. It allows researchers to study difficult urban driving situations using the complete sensor combination of a real autonomous car.
The Lidar database aims to promote and advance research and development in autonomous driving and machine learning.
The first open source dataset made available for both academic and commercial use, PandaSet combines the best LiDAR sensors from Hesai with high quality data annotation from Scale AI. It also offers data collected using a forward-facing LiDAR with image-like resolution (PandarGT) as well as a mechanically rotated LiDAR (Pandar64). The collected data has been annotated with a combination of cuboid annotation and segmentation (Scale 3D Sensor Fusion Segmentation).
The PandaSet platform includes :
- 48,000 camera images
- 16,000 Lidar scans
- +100 scenes of 8s each
- 28 annotation classes
- 37 semantic segmentation labels
- Full suite of sensors: 1x mechanical LiDAR, 1x solid-state LiDAR, 6x cameras, GPS / IMU on board
Complex urban driving scenarios
To create PandaSet, Scale AI and Hesai planned routes and selected scenes that would present complex urban driving scenarios, including steep hills, construction areas, heavy traffic and pedestrians, as well as a variety of times of day and lighting conditions in the morning, afternoon, dusk and evening. The data was retrieved from a Chrystler Pacifica minivan on which several cameras and Hesai lidar were installed on two routes through Silicon Valley.
Translated from Véhicule autonome : La base de données issues des Lidar PandatSet de Scale AI est disponible sur GitHub
