SHOW3D: Multi-View Scenes of 3D Hands and Objects in the Wild

SHOW3D is a large-scale multi-view dataset of hand–object interactions captured in the wild. It was created by researchers from Facebook and published at CVPR 2026 to advance research on egocentric 3D hand–object interaction understanding and model generalization to real-world scenarios.

Use Cases

• Train 3D hand pose estimation models based on multi-view hand-object interactions.
• Develop models for 3D object reconstruction from egocentric video sequences.
• Benchmark generalization of perception models to real-world, in-the-wild scenes.
• Research on understanding hand-object manipulation and interaction dynamics.

Strengths

• Large-scale collection of multi-view data, as stated in the description.
• Focus on real-world, 'in-the-wild' capture scenarios, which is a key feature.
• Published at a major computer vision conference (CVPR 2026), indicating peer-reviewed quality.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic or temporal bias inherent to its collection method.

Provenance

Source

facebook

Collection Method

Multi-view capture of hand-object interactions in the wild.

Freshness

Last updated 2026-06-03 17:37:53; freshness should be verified.