RoboX-EgoGrasp-v0.1: 10 Annotated Egocentric Grasping Video Clips

EgoGrasp is a crowdsourced egocentric video dataset of human grasping interactions, built for robotics imitation learning. The repository contains a sample of 10 annotated clips from the full dataset, which covers over 620 unique everyday object categories. The sample is provided by RoboXTechnologies to help researchers evaluate data quality and annotation.

Use Cases

• Train robotic imitation learning models based on first-person grasp action videos.
• Benchmark computer vision models for action recognition based on egocentric grasping clips.
• Develop and test robotic control policies based on human demonstration data.
• Analyze human-object interaction patterns based on the diverse object categories.

Strengths

• Covers over 620 unique everyday object categories.
• Clips are filmed from a first-person perspective, providing an egocentric view.
• Sample of 10 annotated clips is available for initial evaluation.

Limitations

• The provided sample is only 10 clips; the full dataset size is unknown.
• Column-level documentation is absent; field semantics must be inferred after download.
• License information is unknown, which may restrict usage.

Provenance

Source

RoboXTechnologies via Hugging Face.

Collection Method

Crowdsourced filming of grasp actions using smartphones.

Freshness

Last updated 2026-04-08 14:18:10; freshness should be verified.