Research map / 3D vision / 3D generation
EgoPHI: Estimating 3D Hand-Object Contact and Force from Egocentric Vision
Andela Ilic, Rachel Schuchert, Yijing Jiang, Christian Holz
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 Iterative Refinement Module (IRM) for object pose (ablation) | EgoPHI · ARCTIC 2D projected hand force | IoU: 0.021 -> 0.157 | Improved |
| add IRM (ablation) | EgoPHI · ARCTIC s05 object force | vIoU: 0.7 -> 2.0 | Improved |
| add IRM (ablation) | EgoPHI · H2O s4_ego hand force | MAE (N): 4.65 -> 5.16 | Hurt |
| add IRM (ablation) | EgoPHI · H2O s4_ego object force | MAE (N): 4.70 -> 3.88 | Improved |
| iterative pose refinement over iterations (ablation) | EgoPHI · Pose alignment (ARCTIC) | mean per-vertex error: 64 -> 19 cm | Improved |
| use EgoPHI force head instead of fine-tuned HACO | EgoPHI · ARCTIC s05 hand force | MAE (N): 6.62 -> 4.03 | Improved |
| use EgoPHI instead of HACO for hand contact | EgoPHI · ARCTIC s05 hand contact | F1: .196 -> .190 | No clear effect |
| use EgoPHI instead of PressureVision | EgoPHI · ARCTIC 2D projected hand force | F1: .128 -> .271 | Improved |
| fine-tune with HAMER-estimated hand poses instead of GT hand poses (ablation) | EgoPHI · ARCTIC s05 | hand force MAE (N) / hand contact F1: MAE 4.03 -> 3.32; F1 0.190 -> 0.128 | Mixed |
| fine-tune with HAMER-estimated hand poses (ablation) | EgoPHI · ARCTIC s05 object force | MAE (N): 4.42 -> 4.73 | Hurt |
| train on simulated forces, test on real instrumented objects | EgoPHI · Real-world cube | object force MAE (N): 0.48 | Improved |
Closest papers by meaning
- EgoPressure: A Dataset for Hand Pressure and Pose Estimation in Egocentric Vision (CVPR 2025 · similarity 0.92)
- Egocentric World Model for Photorealistic Hand Object Interaction Synthesis (ECCV 2026 · similarity 0.91)
- Benchmarks and Challenges in Pose Estimation for Egocentric Hand Interactions with Objects (ECCV 2024 · similarity 0.91)
- MEgoHand: Multimodal Egocentric Hand-Object Interaction Motion Generation (NeurIPS 2025 · similarity 0.90)
- EgoTac: In-the-wild Tactile Prediction from Egocentric Vision (NeurIPS 2026 · similarity 0.89)
- HOT3D: Hand and Object Tracking in 3D from Egocentric Multi-View Videos (CVPR 2025 · similarity 0.89)
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.
