Waymo Open Dataset is a dataset containing high-resolution sensor data collected from Waymo Drivers operating their autonomous vehicles in a variety of conditions.

The dataset currently contains:

• 1,950 clips, 20 seconds each, were collected at 10 Hz (390,000 frames) at various geographic locations and conditions.
• Sensor data:
  • 1 medium range radar
  • 4 short range radars
  • 5 cameras (front and side)
  • Synchronizing radar and camera data
  • Radar and camera projection
  • Sensor Calibration
  • Vehicle posture

• Labeled data
  • 4 object class labels — vehicle, pedestrian, cyclist, sign
  • High-quality labels for 1,200 segments of radar data
  • 12.6 million 3D bounding box labels with tracking IDs on radar data
  • High-quality labels for 1,000 segments of camera data
  • 11.8 million 2D bounding box labels with tracking IDs on camera data