Research map / 3D vision / Human pose estimation
Improving Sparse IMU-based Motion Capture with Motion Label Smoothing
Zhaorui Meng, Lu Yin, Yangqing Hou, Anjun Chen, Shihui Guo, Yipeng Qin
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 motion label smoothing (skeleton Perlin noise) to GlobalPose | GlobalPose · TotalCapture | SIP error: 9.85 -> 7.84 | Improved |
| add motion label smoothing to PIP | PIP · TotalCapture | SIP error: 11.16 -> 10.54 | Improved |
| add motion label smoothing to TransPose | TransPose · TotalCapture | SIP error: 14.28 -> 12.49 | Improved |
| add motion label smoothing to GlobalPose on ANDY | GlobalPose · ANDY | SIP error: 39.58 -> 36.71 | Improved |
| add motion label smoothing to PIP on ANDY (small gain) | PIP · ANDY | Joint error: 13.56 -> 13.49 | Improved |
| blend plain Gaussian noise (naive label smoothing) (ablation) | GlobalPose · TotalCapture | SIP error: 9.85 -> 8.82 | Improved |
| add temporal smoothness to the noise (ablation) | Gaussian-noise label smoothing on GlobalPose · TotalCapture | SIP error: 8.82 -> 8.59 | Improved |
| add joint correlation along kinematic chains (ablation) | Gaussian-noise label smoothing on GlobalPose · TotalCapture | SIP error: 8.59 -> 8.22 | Improved |
| switch to Perlin noise for low-frequency dominance (ablation) | Gaussian-noise label smoothing on GlobalPose · TotalCapture | SIP error: 8.22 -> 7.84 | Improved |
| blend T-pose vector as label smoothing target (ablation) | GlobalPose · TotalCapture | Mesh error: 4.35 -> 4.73 | Hurt |
| blend T-pose vector as label smoothing target (ablation) | GlobalPose · TotalCapture | Joint error: 3.96 -> 3.96 | No clear effect |
| blend AMASS mean motion as target (naive adaptation) (ablation) | GlobalPose · TotalCapture | SIP error: 9.85 -> 8.75; Mesh 4.35 -> 4.60 | Mixed |
Closest papers by meaning
- MODA: Motion-Drift Augmentation for Inertial Human Motion Analysis (CVPR 2025 · similarity 0.87)
- MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation (AAAI 2024 · similarity 0.87)
- Dynamic Inertial Poser (DynaIP): Part-Based Motion Dynamics Learning for Enhanced Human Pose Estimation with Sparse Inertial Sensors (CVPR 2024 · similarity 0.86)
- I'M HOI: Inertia-aware Monocular Capture of 3D Human-Object Interactions (CVPR 2024 · similarity 0.85)
- Spatial-Related Sensors Matters: 3D Human Motion Reconstruction Assisted with Textual Semantics (AAAI 2024 · similarity 0.85)
- DiffusionPoser: Real-time Human Motion Reconstruction From Arbitrary Sparse Sensors Using Autoregressive Diffusion (CVPR 2024 · similarity 0.84)
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.
