Research map / 3D vision / Human pose estimation
STAR: Sparse Trained Articulated Human Body Regressor
Ahmed A. A. Osman, Timo Bolkart, Michael J. Black
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 learned sparse per-joint activation masks (ReLU) to pose correctives | STAR · model size | pose corrective parameters: 4.28x10^6 -> 3.37x10^5 | Improved |
| use sparse local STAR correctives instead of SMPL dense ones, both trained on CAESAR | STAR · held-out Dyna and 3DBodyTex scans | vertex-to-vertex error (mm) | Improved |
| add SizeUSA scans to CAESAR training data (ablation) | SMPL and STAR · held-out Dyna and 3DBodyTex scans | vertex-to-vertex error (mm) | Improved |
| condition pose correctives on body shape (second PCA coefficient, correlated with BMI) (ablation) | STAR · held-out Dyna and 3DBodyTex scans | vertex-to-vertex error (mm) | Improved |
| train STAR rather than SMPL on CAESAR plus SizeUSA | STAR · held-out Dyna and 3DBodyTex scans | vertex-to-vertex error (mm) | Improved |
| switch pose features from Rodrigues rotation matrices to quaternions (ablation) | STAR · model compactness | accuracy: 207 -> 93 | No clear effect |
| shape space trained on one dataset applied to the other (CAESAR vs SizeUSA) | SMPL shape space · explained variance | explained variance | Hurt |
Later papers that compared with this method
2 later accepted papers report STAR as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Heatmap Regression without Soft-Argmax for Facial Landmark Detection (ICCV 2025 · better on 11 of 14 shared results)
- VAREN: Very Accurate and Realistic Equine Network (CVPR 2024 · better on 4 of 4 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Keep It SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image (ECCV 2016 · similarity 0.92)
- Expressive Body Capture: 3D Hands, Face, and Body From a Single Image (CVPR 2019 · similarity 0.90)
- Learning to Reconstruct 3D Human Pose and Shape via Model-Fitting in the Loop (ICCV 2019 · similarity 0.90)
- Convolutional Mesh Regression for Single-Image Human Shape Reconstruction (CVPR 2019 · similarity 0.90)
- SMPLer-X: Scaling Up Expressive Human Pose and Shape Estimation (NeurIPS 2023 · similarity 0.89)
- SUPR: A Sparse Unified Part-Based Human Representation (ECCV 2022 · similarity 0.89)
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.
