CALVIN-3D Cam Params: Camera Parameters for Vision-Language-Action Models

FALCON-VLA provides camera parameters for the CALVIN-3D dataset, supporting research on grounding Vision-Language-Action models in spatial foundation priors. The dataset was last updated on 2026-05-25. It accompanies a paper accepted at ICLR 2026.

Use Cases

• Training spatial reasoning modules for Vision-Language-Action models based on camera parameters.
• Benchmarking model performance on tasks requiring 3D spatial understanding based on camera data.
• Grounding language instructions to physical actions in 3D environments based on camera calibration data.

Strengths

• Associated with a peer-reviewed ICLR 2026 publication.
• Last updated on 2026-05-25, indicating recent maintenance.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

FALCON-VLA

Freshness

Last updated 2026-05-25 10:13:56; freshness should be verified.