Research map / 3D vision / Human pose estimation
H+O: Unified Egocentric Recognition of 3D Hand-Object Poses and Interactions
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add temporal modeling of hand and object poses (ablation) | H+O on FPHA-HO · FPHA-HO | Action Accuracy (%): 85.56 -> 94.73 | Helped |
| Add explicit hand-object interaction modeling to temporal pose features (ablation) | H+O on FPHA-HO · FPHA-HO | Action Accuracy (%): 94.73 -> 96.99 | Helped |
| Combine hand and object pose inputs instead of using either pose alone (ablation) | H+O temporal model · FPHA-HO | Action Accuracy (%): 89.47 -> 94.73; 85.71 -> 94.73 | Helped |
| Add object class probabilities to hand-pose and action-class temporal features (ablation) | H+O on FPHA · FPHA | Accuracy: 74.20 -> 82.43 | Helped |
| Add action category probabilities to hand-pose temporal features (ablation) | H+O on FPHA · FPHA | Accuracy: 62.54 -> 74.20 | Helped |
| Add interaction modeling to joint hand-object pose estimation (ablation) | H+O on FPHA · FPHA | HP error (mean 3D distance in mm): 16.87 -> 15.81 | Helped |
| Add interaction modeling to joint hand-object pose estimation (ablation) | H+O on FPHA · FPHA | OP error (mean 3D distance in mm): 25.54 -> 24.89 | Helped |
| Jointly predict hand and object poses rather than train separate pose-only networks (ablation) | H+O · FPHA | OP error (mean 3D distance in mm): 28.27 -> 25.54 | Helped |
| Jointly predict hand and object poses rather than use hand-only training (ablation) | H+O · FPHA | HP error (mean 3D distance in mm): 16.15 -> 16.87 | Hurt |
| Use a learned interaction mapping before the temporal LSTM rather than direct pose inputs (ablation) | H+O · FPHA-HO | Action Accuracy (%): 94.73 -> 96.99 | Helped |
| Replace the LSTM with a GRU (ablation) | H+O temporal module · FPHA-HO | Action recognition | No clear effect |
Closest papers by meaning
- Collaborative Learning for 3D Hand-Object Reconstruction and Compositional Action Recognition from Egocentric RGB Videos Using Superquadrics (AAAI 2025 · similarity 0.90)
- Modeling Fine-Grained Hand-Object Dynamics for Egocentric Video Representation Learning (ICLR 2025 · similarity 0.90)
- HOT3D: Hand and Object Tracking in 3D from Egocentric Multi-View Videos (CVPR 2025 · similarity 0.89)
- MEgoHand: Multimodal Egocentric Hand-Object Interaction Motion Generation (NeurIPS 2025 · similarity 0.88)
- EgoPressure: A Dataset for Hand Pressure and Pose Estimation in Egocentric Vision (CVPR 2025 · similarity 0.88)
- How Do I Do That? Synthesizing 3D Hand Motion and Contacts for Everyday Interactions (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.