PanoBench: A Panoptic Occupancy Prediction Benchmark for Autonomous Driving

Key points

• Panoptic occupancy unifies 3D geometry, semantics, and instance identities, but its development is limited by scarce instance-aware annotations.1
• The synthetic branch derives complete panoptic geometry from simulated meshes.1
• For real-world data, a new pipeline reconstructs backgrounds using geometric priors and topology constraints, generates foreground mesh assets from multi-frame multi-view images, and fuses them into sparse panoptic voxels.1
• We benchmark representative methods on semantic occupancy prediction, panoptic occupancy prediction, and 4D occupancy forecasting, and study voxel resolution, synthetic-to-real transfer, and storage efficiency.1

Sources (1)

• [1]PanoBench: A Panoptic Occupancy Prediction Benchmark for Autonomous Driving
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 16, 02:34 PM

  We introduce PanoBench, a camera-centric panoptic occupancy prediction benchmark spanning synthetic and real-world driving scenes.

  Panoptic occupancy unifies 3D geometry, semantics, and instance identities, but its development is limited by scarce instance-aware annotations.

Extractive summary: sentences quoted from the sources.