PRISM: Polarimetric Road-Surface Dataset with 47k Synchronized Frames

47,098 time-synchronized frames combine trichromatic linear polarization, co-boresighted RGB, 128-channel LiDAR, and RTK-GNSS/INS data. The PRISM dataset is a benchmark for polarimetric road-surface sensing and measurement, captured from an in-vehicle sensor rig. It was submitted anonymously to the NeurIPS 2026 Evaluations & Datasets Track.

Use Cases

• Developing road condition classification models based on polarimetric imagery.
• Training sensor fusion algorithms for autonomous driving using synchronized LiDAR, RGB, and polarization data.
• Benchmarking perception systems on challenging road-surface scenarios.
• Researching material property inference for road surfaces using polarimetric data.

Strengths

• 47,098 time-synchronized frames provide a substantial volume for model training.
• Multimodal data includes four synchronized sensor types: polarization, RGB, LiDAR, and RTK-GNSS/INS.
• Data is presented as a benchmark for a NeurIPS 2026 track, suggesting a defined evaluation standard.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset's geographic and temporal coverage is not specified in the provided metadata.

Provenance

Source

NeurIPS-2026-PRISM

Collection Method

Captured from an in-vehicle sensor rig across unspecified locations.

Freshness

Last updated 2026-05-06 09:24:45; freshness should be verified.