Research map / Video / Video segmentation and tracking
Gravity as a Reference for Estimating a Person's Height From Video
Didier Bieler, Semih Gunel, Pascal Fua, Helge Rhodin
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 |
|---|---|---|---|
| fit a degree-two polynomial to the whole COM trajectory instead of using only the peak and last point (ablation) | Ours (curve fit) · Articulated free-fall dataset, on-spot jumps (4m and 7m combined) | height MAE (cm): 15.5 -> 3.9 | Improved |
| curve fitting versus distance-based acceleration on the tennis ball (ablation) | Ours (curve fit) · Tennis ball drops | size error | No clear effect |
| add RANSAC on top of confidence-based outlier removal (ablation) | Ours (curve fit) · Articulated free-fall dataset, on-spot jumps | height MAE (cm): 3.9 -> 7.4 | Hurt |
| add confidence-based outlier removal of AlphaPose points (ablation) | Ours (curve fit) · Articulated free-fall dataset, on-spot jumps | height MAE (cm) | Improved |
| use physics-based gravity reference instead of regression from appearance | Ours (curve fit) · Articulated free-fall dataset, on-spot jumps (4m and 7m combined) | height MAE (cm): 6.5 -> 3.9 | Improved |
| use gravity reference instead of predicting the population mean height | Ours (curve fit) · Articulated free-fall dataset, on-spot jumps (4m and 7m combined) | height MAE (cm): 8.1 -> 3.9 | Improved |
| lower versus higher bounce height of the tennis ball (ablation) | Ours (curve fit) · Tennis ball drops | size error | No clear effect |
| record at 7m instead of 4m (ablation) | Ours (curve fit) · Tennis ball drops | size error | Hurt |
| use low jumps instead of high jumps (4m) (ablation) | Ours (curve fit) · On-spot jump types | height MAE (cm): 8.0 -> 13.1 | Hurt |
| use lateral running instead of frontal jumping (ablation) | Ours (curve fit) · Lateral motion (4m) | height MAE (cm): 6.6 -> 18.9 | Hurt |
| rotate the camera by 30 degrees roll instead of upright (ablation) | Ours (curve fit) · Jumping with camera roll | absolute error (cm): 4.5 -> 0.4 | No clear effect |
| move the simulated jump from fronto-parallel to straight towards the camera at 4m (ablation) | Ours (curve fit) · Simulation, d=4m | absolute error (cm): 0.4 -> 21 | Hurt |
Closest papers by meaning
- Gravity-Aware Monocular 3D Human-Object Reconstruction (ICCV 2021 · similarity 0.86)
- GVHMR: Gravity-View Coordinates for Global Human Motion Recovery From Monocular Videos (TPAMI 2026 · similarity 0.85)
- Force From Motion: Decoding Physical Sensation in a First Person Video (CVPR 2016 · similarity 0.84)
- Trajectory Optimization for Physics-Based Reconstruction of 3D Human Pose From Monocular Video (CVPR 2022 · similarity 0.84)
- WHAM: Reconstructing World-grounded Humans with Accurate 3D Motion (CVPR 2024 · similarity 0.82)
- Seurat: From Moving Points to Depth (CVPR 2025 · similarity 0.80)
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.
