Research map / 3D vision / Human pose estimation
RenderIH: A Large-Scale Synthetic Dataset for 3D Interacting Hand Pose Estimation
Lijun Li, Linrui Tian, Xindi Zhang, Qi Wang, Bang Zhang, Liefeng Bo, Mengyuan Liu, Chen Chen
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 RenderIH synthetic data to IH2.6M training | IntagHand · IH2.6M test | MPJPE (mm): 10.9 -> 9.72 | Improved |
| add RenderIH synthetic data to IH2.6M training | InterNet · IH2.6M test | MPJPE (mm): 18.28 -> 17.19 | Improved |
| add RenderIH synthetic data to IH2.6M training | TransHand · IH2.6M test | MPJPE (mm): 10.6 -> 10.06 | Improved |
| pretrain on RenderIH then finetune on IH2.6M | TransHand · IH2.6M test | PAMPJPE (mm): 6.76 -> 5.79 | Improved |
| pretrain on synthetic then finetune rather than mixing datasets | TransHand · IH2.6M test | MPJPE | Improved |
| add correlation encoder fusing other-hand features (ablation) | TransHand · IH2.6M test | PAMPJPE (mm): 7.32 -> 6.76 | Improved |
| add natural discriminator to pose optimization (ablation) | RenderIH pose generation · user study | natural rate: 54.68% -> 81.25% | Improved |
| train on RenderIH instead of H2O-3D or Ego3d (40K samples each) | TransHand · Tzionas | MPJPE (mm): 35.61 -> 32.80 (vs H2O-3D) | Improved |
| increase synthetic data beyond about 900K images | TransHand · IH2.6M test | MPJPE | No clear effect |
| finetune RenderIH-pretrained model on only 1% real data (ablation) | TransHand · IH2.6M test | MPJPE (mm): 27.73 -> 12.6 | Improved |
| train on RenderIH only instead of IH2.6M | TransHand · Tzionas | MPJPE (mm): 11.38 -> 22.11 | Hurt |
| use anchors instead of all vertices for hand attraction | RenderIH pose optimization · qualitative | pose quality | No clear effect |
Closest papers by meaning
- Analyzing the Synthetic-to-Real Domain Gap in 3D Hand Pose Estimation (CVPR 2025 · similarity 0.91)
- InterHand2.6M: A Dataset and Baseline for 3D Interacting Hand Pose Estimation from a Single RGB Image (ECCV 2020 · similarity 0.91)
- 3D Interacting Hand Pose Estimation by Hand De-Occlusion and Removal (ECCV 2022 · similarity 0.90)
- EasyHOI: Unleashing the Power of Large Models for Reconstructing Hand-Object Interactions in the Wild (CVPR 2025 · similarity 0.89)
- HandBooster: Boosting 3D Hand-Mesh Reconstruction by Conditional Synthesis and Sampling of Hand-Object Interactions (CVPR 2024 · similarity 0.89)
- SiMHand: Mining Similar Hands for Large-Scale 3D Hand Pose Pre-training (ICLR 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.
