Research map / 3D vision / Human pose estimation
Making Deep Heatmaps Robust to Partial Occlusions for 3D Object Pose Estimation
Markus Oberweger, Mahdi Rad, Vincent Lepetit
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 Feature Mapping to the method (ablation) | Our method · Occluded LineMOD | 2D Reprojection-5px accuracy: 43.1 -> 60.9 | Improved |
| add Feature Mapping to the method (ablation) | Our method · Occluded LineMOD | ADD/I-10% accuracy: 25.8 -> 30.4 | Improved |
| add Feature Mapping to the method (ablation) | Our method · YCB-Video | ADD/I-10% accuracy: 33.6 -> 53.1 | Improved |
| add Feature Mapping to the method (ablation) | Our method · YCB-Video | AUC: 61.4 -> 72.8 | Improved |
| add Feature Mapping to BB8-style full-image approach | BB8 [5] · Occluded LineMOD | 2D Reprojection error | No clear effect |
| use heatmap patch method instead of PoseCNN | PoseCNN · Occluded LineMOD | 2D Reprojection-5px accuracy: 17.2 -> 60.9 | Improved |
| train with random geometric occlusions (ablation) | BB8 and CPM · synthetic Occluded LineMOD | pose accuracy | Improved |
| train with random geometric occlusions (ablation) | Our method · synthetic Occluded LineMOD | pose accuracy | No clear effect |
| reduce patch size to 8px (ablation) | Our method · Occluded LineMOD | 2D Reprojection: 19% worse | Hurt |
| increase patch size to 128px (ablation) | Our method · Occluded LineMOD | 2D Reprojection: 5% decrease | Hurt |
| use patch sizes between 25px and 40px (ablation) | Our method · Occluded LineMOD | pose accuracy | No clear effect |
| use more than 64 patches (ablation) | Our method · Occluded LineMOD | pose accuracy | No clear effect |
Closest papers by meaning
- BB8: A Scalable, Accurate, Robust to Partial Occlusion Method for Predicting the 3D Poses of Challenging Objects Without Using Depth (ICCV 2017 · similarity 0.91)
- Occlusion-Aware Networks for 3D Human Pose Estimation in Video (ICCV 2019 · similarity 0.89)
- Templates for 3D Object Pose Estimation Revisited: Generalization to New Objects and Robustness to Occlusions (CVPR 2022 · similarity 0.89)
- Human Pose Recognition via Occlusion-Preserving Abstract Images (ECCV 2024 · similarity 0.88)
- 3D-Aware Hypothesis & Verification for Generalizable Relative Object Pose Estimation (ICLR 2024 · similarity 0.86)
- NOPE: Novel Object Pose Estimation from a Single Image (CVPR 2024 · similarity 0.86)
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.
