HGGT: Synthetic Multi-View Hand-Object Interaction Images

HGGT is a synthetic multi-view hand-object interaction dataset introduced by researchers including Yumeng Liu and Marc Habermann. It contains diverse photorealistic hand-object interactions rendered with randomized camera viewpoints to provide critical viewpoint diversity. The dataset was created by author catmint123 and was last updated on Hugging Face on 2026-05-02.

Use Cases

• Training 3D hand pose estimation models based on photorealistic synthetic hand-object interactions.
• Benchmarking robustness of reconstruction algorithms based on the dataset's randomized camera viewpoints.
• Developing multi-view computer vision systems based on the described viewpoint diversity.

Strengths

• Dataset provides critical viewpoint diversity through randomized camera viewpoints.
• Data is photorealistic and designed for hand-object interaction research.

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

Hugging Face dataset uploaded by catmint123, associated with the HGGT research paper.

Collection Method

Synthetically rendered photorealistic images.

Freshness

Last updated 2026-05-02 15:16:16; freshness should be verified.