Research map / 3D vision / Human pose estimation
Bones Can't Be Triangles: Accurate and Efficient Vertebrae Keypoint Estimation through Collaborative Error Revision
Jinhee Kim, Taesung Kim, Jaegul Choo
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 KeyBot automatic error revision before user input | Kim et al. interactive keypoint model · AASCE, 0 clicks | MRE: 51.58 -> 41.70 | Improved |
| add KeyBot automatic error revision | Kim et al. interactive keypoint model · AASCE NoC_10@50 | NoC: 1.03 -> 0.53 | Improved |
| add the error detector so corrector outputs are applied selectively (ablation) | KeyBot · AASCE, 0 clicks | MRE: 57.61 -> 44.18 | Improved |
| train corrector on all three error types instead of one (ablation) | KeyBot corrector · AASCE, 0 clicks | MRE: 52.67 -> 44.18 (vs misvertex only) | Improved |
| add real keypoint errors to the synthetic error training data (ablation) | KeyBot training · AASCE, 0 clicks | MRE: 41.70 -> 46.42 | Hurt |
| accumulate false predictions across iterations (ablation) | KeyBot · AASCE, 0 clicks | MRE: 43.54 -> 41.70 (w/o fp vs 3 iterations) | Improved |
| run detector on small keypoint groups (k=8) instead of all 68 keypoints (ablation) | KeyBot detector · BUU-AP, 0 clicks | MRE: 39.32 -> 31.87 | Improved |
| vary detector stride s with fixed window k (ablation) | KeyBot detector · BUU-AP and BUU-LA | MRE | No clear effect |
| increase KeyBot iterations from one to three (ablation) | KeyBot · AASCE, 0 clicks | MRE: 44.18 -> 41.70 | Improved |
| increase KeyBot iterations from one to three (ablation) | KeyBot · AASCE, 2 user clicks | MRE: 20.66 -> 25.02 | Hurt |
| plug KeyBot into Click-Pose backbone without retraining | Click-Pose · AASCE, 0 clicks | MRE: 54.65 -> 51.24 | Improved |
| use KeyBot during annotation (user study) | interactive annotation workflow · BUU-LA user study | annotation time (s): 21.61 -> 11.71 | Improved |
Closest papers by meaning
- Orientation Keypoints for 6D Human Pose Estimation (TPAMI 2021 · similarity 0.82)
- SIMSPINE: A Biomechanics-Aware Simulation Framework for 3D Spine Motion Annotation and Benchmarking (CVPR 2026 · similarity 0.82)
- Key-Grid: Unsupervised 3D Keypoints Detection using Grid Heatmap Features (NeurIPS 2024 · similarity 0.80)
- SpineBench: A Clinically Salient, Level-Aware Benchmark Powered by the SpineMed-450k Corpus (ICLR 2026 · similarity 0.80)
- KITRO: Refining Human Mesh by 2D Clues and Kinematic-tree Rotation (CVPR 2024 · similarity 0.79)
- Coupled Laplacian Eigenmaps for Locally-Aware 3D Rigid Point Cloud Matching (CVPR 2024 · similarity 0.79)
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.
