Research map / 3D vision / Human pose estimation
DeepSimHO: Stable Pose Estimation for Hand-Object Interaction via Physics Simulation
Rong Wang, Wei Mao, Hongdong Li
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 DeepSim-based stability loss (ablation) | base hand-object network · DexYCB test (S0) | SR (%): 28.55 -> 32.89 | Improved |
| add DeepSim-based stability loss (ablation) | base hand-object network · DexYCB test (S0) | SD (cm): 2.77 -> 2.42 | Improved |
| add DeepSim-based stability loss (ablation) | base hand-object network · DexYCB test (S0) | CP (%): 94.19 -> 95.90 | Improved |
| add DeepSim-based stability loss (ablation) | base hand-object network · DexYCB test (S0) | PD (cm): 1.51 -> 1.48 | Improved |
| regress object center (T) instead of scalar stability loss (S) (ablation) | DeepSim · DexYCB test (S0) | SR (%): 32.89 -> 28.51 | Hurt |
| regress 6DoF object pose (RT) instead of scalar stability loss (S) (ablation) | DeepSim · DexYCB test (S0) | SR (%): 32.89 -> 26.60 | Hurt |
| regress object center (T) instead of scalar stability loss (S) (ablation) | DeepSim · DexYCB test (S0) | approximation error (mm): 2.70 -> 13.80 | Hurt |
| LSTM with per-step displacement instead of MLP scalar stability loss (ablation) | DeepSim · DexYCB test (S0) | SR (%): 32.89 -> 30.66 | Hurt |
| LSTM + T versus no DeepSim (ablation) | DeepSim · DexYCB test (S0) | SR (%): 28.55 -> 30.66 | Improved |
| DeepSim vs numerical gradient from MuJoCo finite differences (ablation) | stability loss backprop · DexYCB | gradient quality | Improved |
| DeepSim vs analytical gradient through NimblePhysics (ablation) | stability loss backprop · DexYCB | gradient stability | Improved |
| compare with ArtiBoost on stability | DeepSimHO · DexYCB test | SR (%): 30.07 -> 32.89 | Improved |
Closest papers by meaning
- VPHO: Joint Visual-Physical Cue Learning and Aggregation for Hand-Object Pose Estimation (AAAI 2026 · similarity 0.90)
- Learning a Contact Potential Field for Modeling the Hand-Object Interaction (TPAMI 2024 · similarity 0.90)
- CPF: Learning a Contact Potential Field To Model the Hand-Object Interaction (ICCV 2021 · similarity 0.90)
- Robust Hand Pose Estimation During the Interaction With an Unknown Object (ICCV 2017 · similarity 0.90)
- MEgoHand: Multimodal Egocentric Hand-Object Interaction Motion Generation (NeurIPS 2025 · similarity 0.88)
- EasyHOI: Unleashing the Power of Large Models for Reconstructing Hand-Object Interactions in the Wild (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.
