Dimensions overall score 7.0

GitHub Code Pulse

No public code linked for this paper yet.

• SHOW3D is the first large-scale dataset with 3D annotations of hands interacting with objects in diverse real-world environments, including outdoor settings.result 95%
• The SHOW3D capture system uses a novel marker-less multi-camera rig that allows for nearly unconstrained mobility in genuinely in-the-wild conditions.method 90%
• Models trained on SHOW3D generalize better to studio environments than models trained on studio datasets generalize to SHOW3D's in-the-wild data.result 85%
• The SHOW3D dataset spans both in-the-wild and controlled domains, providing broader environmental coverage than existing studio-only datasets.result 80%
• Existing hand-object interaction datasets are predominantly captured in controlled studio settings, which limits environmental diversity and model generalization to real-world scenarios.limitation 90%
• The SHOW3D approach significantly reduces the fundamental trade-off between environmental realism and accuracy of 3D annotations.technical 85%
• Natural language descriptions of hand-object interactions in SHOW3D provide valuable context for predicting future object poses.result 75%
• The SHOW3D dataset contains 4.3 million frames with both hand and object mesh annotations across 21 participants and 38 scenes.result 95%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card