PandaSet 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 Hesai’s best-in-class LiDAR sensors with Scale AI’s high-quality data annotation. PandaSet features data collected using a forward-facing LiDAR with image-like resolution (PandarGT) as well as a mechanical spinning LiDAR (Pandar64). The collected data was annotated with a combination of cuboid and segmentation annotation (Scale 3D Sensor Fusion Segmentation).
48,000 camera images
16,000 LiDAR sweeps
+100 scenes of 8s each
28 annotation classes
37 semantic segmentation labels
Full sensor suite: 1x mechanical LiDAR, 1x solid-state LiDAR, 6x cameras, On-board GPS/IMU
For PandaSet we carefully planned routes and selected scenes that would showcase complex urban driving scenarios, including steep hills, construction, dense traffic and pedestrians, and a variety of times of day and lighting conditions in the morning, afternoon, dusk and evening.
PandaSet scenes are selected from 2 routes in Silicon Valley: (1) San Francisco; and (2) El Camino Real from Palo Alto to San Mateo.
We collected data using a Chrysler Pacifica minivan mounted with a suite of cameras and Hesai LiDARs.
To achieve a high quality multi-sensor dataset, it is essential to calibrate the extrinsics and intrinsics of every sensor.
We express extrinsic coordinates relative to the ego frame, i.e. the midpoint of the rear vehicle axle.
The most relevant steps are described below:
Camera intrinsic calibration
Scale’s data annotation platform combines human work and review with smart tools, statistical confidence checks and machine learning checks to ensure the quality of annotations.
The resulting accuracy is consistently higher than what a human or synthetic labeling approach can achieve independently as measured against seven rigorous quality areas for each annotation.
PandaSet includes 3D Bounding boxes for 28 object classes and a rich set of class attributes related to activity, visibility, location, pose. The dataset also includes Point Cloud Segmentation with 37 semantic labels including for smoke, car exhaust, vegetation, and driveable surface.