Research map / 3D vision / Human pose estimation
BB8: A Scalable, Accurate, Robust to Partial Occlusion Method for Predicting the 3D Poses of Challenging Objects Without Using Depth
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 |
|---|---|---|---|
| predict 2D projections of bounding-box corners instead of direct pose (ablation) | BB8 · LINEMOD (ground-truth center) | 2D Projections accuracy (%): 64.9 -> 85.4 | Improved |
| add mask-rendering pose refinement (ablation) | BB8 · LINEMOD (ground-truth center) | 2D Projections accuracy (%): 85.4 -> 89.7 | Improved |
| add color-rendering pose refinement (ablation) | BB8 · LINEMOD (ground-truth center) | 2D Projections accuracy (%): 85.4 -> 91.3 | Improved |
| add RGB refinement (ablation) | BB8 · LINEMOD (detected center) | 2D Projections accuracy (%): 83.9 -> 89.3 | Improved |
| use segmentation-based localization instead of sliding window (ablation) | BB8 · object detection | correct detection rate (%): 75 -> 98.8 | Improved |
| predict corner projections instead of direct pose (ablation) | BB8 · LINEMOD Driller | 2D Projections accuracy (%): 46.5 -> 77.6 | Improved |
| use BB8 with refinement instead of Brachmann et al. [2] | BB8 · LINEMOD | 2D Projection accuracy (%): 73.7 -> 89.3 | Improved |
| use BB8 with refinement instead of Brachmann et al. [2] | BB8 · LINEMOD | 5cm 5deg accuracy (%): 40.6 -> 69.0 | Improved |
| use random instead of orientation-spread training image selection (ablation) | BB8 · LINEMOD | pose accuracy | No clear effect |
| replace VGG with a specific small network for a single object (ablation) | BB8 · computation time | segmentation time (ms): 140 -> 20 | Improved |
Closest papers by meaning
- Making Deep Heatmaps Robust to Partial Occlusions for 3D Object Pose Estimation (ECCV 2018 · similarity 0.91)
- 3D Human Pose Estimation Using Spatio-Temporal Networks with Explicit Occlusion Training (AAAI 2020 · similarity 0.90)
- Pose without Guesses: Generalizable Object Pose Estimation From a Single Reference (IJCV 2026 · similarity 0.90)
- Omni6D: Large-Vocabulary 3D Object Dataset for Category-Level 6D Object Pose Estimation (ECCV 2024 · similarity 0.89)
- Templates for 3D Object Pose Estimation Revisited: Generalization to New Objects and Robustness to Occlusions (CVPR 2022 · similarity 0.89)
- ImageNet3D: Towards General-Purpose Object-Level 3D Understanding (NeurIPS 2024 · similarity 0.88)
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.
