ViPE wild SDG 1M

From the catalogue. Listed by one researcher from one source page.

• id: vipe-wild-sdg-1m
• kind: world model and video
• robot or device: in-the-wild video with estimated camera pose, depth, intrinsics (class: sensor only)
• how collected: automatic annotation with ViPE
• size as stated: HF size tag 100K<n<1M
• hours: not stated | hours only claimed: none | episodes: not stated
• year: 2025
• organisation: NVIDIA
• licence: CC-BY-NC-4.0 (class: non commercial)
• access: open (class: open)
• link (dataset): https://huggingface.co/datasets/nvidia/vipe-wild-sdg-1m
• Hugging Face repository: nvidia/vipe-wild-sdg-1m, 12,741 downloads in 30 days, https://huggingface.co/datasets/nvidia/vipe-wild-sdg-1m
• note: Pose and depth labels for spatial world model training.

Confirm the licence at the link before relying on it.