Research map / 3D vision / Human pose estimation
IKOL: Inverse Kinematics Optimization Layer for 3D Human Pose and Shape Estimation via Gauss-Newton Differentiation
Read the paperCheck 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 |
|---|---|---|---|
| Use IKOL instead of the strongest listed regression baseline | 3DPW human pose and shape estimation · 3DPW | MPJPE: 74.7 -> 73.3 | Helped |
| Use IKOL instead of HybrIK | MPI-INF-3DHP human pose and shape estimation · MPI-INF-3DHP | MPJPE: 91.0 -> 88.8 | Helped |
| Fine-tune IKOL with the HybrIK model | IKOL · 3DPW | PVE: 86.4 -> 84.1 | Helped |
| Fine-tune IKOL with the HybrIK model | IKOL · MPI-INF-3DHP | PCK: 87.9 -> 87.0 | Hurt |
| Replace the HybrIK analytical IK module with iterative IKOL optimization (ablation) | HybrIK · 3DPW | MPJPE: 80.0 -> 78.2 | Helped |
| Add GN-Diff and its optimization loss during training (ablation) | IKOL · 3DPW | MPJPE: 78.4 -> 73.3 | Helped |
| Add GN-Diff and its optimization loss during training (ablation) | IKOL · 3DPW | PVE: 91.5 -> 86.4 | Helped |
| Add GN-Diff and its optimization loss during training (ablation) | IKOL · 3DPW | PA-MPJPE: 48.7 -> 45.5 | Helped |
| Use iterative IKOL optimization instead of analytical inverse kinematics (ablation) | HybrIK · 3DPW | PVE: 94.5 -> 92.5 | Helped |
| Use iterative IKOL optimization instead of analytical inverse kinematics (ablation) | HybrIK · 3DPW | PA-MPJPE: 48.8 -> 48.5 | Helped |
| Use IKOL instead of SMPLify optimization | Optimization branch · 3DPW | Runtime (s): 74 -> 0.182 | Helped |
| Use IKOL instead of HybrIK analytical solver | Optimization branch · 3DPW | Runtime (s): 0.04 -> 0.182 | Hurt |
Closest papers by meaning
- NIKI: Neural Inverse Kinematics With Invertible Neural Networks for 3D Human Pose and Shape Estimation (CVPR 2023 · similarity 0.88)
- HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation (CVPR 2021 · similarity 0.87)
- Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · similarity 0.85)
- ProPose: Probabilistic 3D Human Pose Estimation with Instance-Level Distribution and Normalizing Flow (AAAI 2025 · similarity 0.85)
- Reconstructing Humans with a Biomechanically Accurate Skeleton (CVPR 2025 · similarity 0.85)
- Toward Approaches to Scalability in 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.85)
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.