Research map / 3D vision / Human pose estimation
Explaining the Ambiguity of Object Detection and 6D Pose From Visual Data
Fabian Manhardt, Diego Martin Arroyo, Christian Rupprecht, Benjamin Busam, Tolga Birdal, Nassir Navab, Federico Tombari
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 30 pose hypotheses instead of a single one (ablation) | SSD-6D-style detector on synthetic cube · Synthetic toy dataset (cube) | ADI: 15.6 -> 100 | Improved |
| predict 30 pose hypotheses instead of a single one (ablation) | SSD-6D-style detector on synthetic cube · Synthetic toy dataset (cube) | VSS: 87.4 -> 98.6 | Improved |
| predict 30 pose hypotheses instead of a single one (ablation) | SSD-6D-style detector on synthetic cup · Synthetic toy dataset (cup) | ADI: 100 -> 100 | No clear effect |
| predict 5 hypotheses instead of 1 (ablation) | SSD-6D-style detector · LineMOD unambiguous (synthetic training) | ADD: 31.2 -> 35.3 | Improved |
| predict 5 hypotheses instead of 1 (ablation) | SSD-6D-style detector · LineMOD unambiguous (synthetic training) | Translation error (mm): 45.6 -> 39.5 | Improved |
| use 10 hypotheses instead of 1 (ablation) | SSD-6D-style detector · LineMOD unambiguous (synthetic training) | ADD: 31.2 -> 31.3 | No clear effect |
| use 40 hypotheses instead of 1 (ablation) | SSD-6D-style detector · LineMOD unambiguous (synthetic training) | Rotation error (deg): 17.9 -> 22.5 | Hurt |
| predict 30 hypotheses instead of 1 (ablation) | SSD-6D-style detector · Ambiguous LineMOD objects (eggbox, glue; mean) | F1: 78.5 -> 94.1 | Improved |
| predict 30 hypotheses instead of 1 (ablation) | SSD-6D-style detector · Ambiguous LineMOD objects (eggbox, glue; mean) | ADI: 57.4 -> 55.2 | Hurt |
| predict 30 hypotheses instead of 1 (ablation) | SSD-6D-style detector · Ambiguous T-LESS objects (mean of 4) | ADI: 36.6 -> 56.4 | Improved |
| predict 30 hypotheses instead of 1 (ablation) | SSD-6D-style detector · Ambiguous T-LESS objects (mean of 4) | VSS: 77.3 -> 81.2 | Improved |
| train with real data split instead of synthetic data only | Multiple hypotheses (M=5) · LineMOD unambiguous | ADD: 35.3 -> 44.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SSD-6D — SSD-6D: Making RGB-Based 3D Detection and 6D Pose Estimation Great Again (ICCV 2017) · better on 9 of 10 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
- UNOPose: Unseen Object Pose Estimation with an Unposed RGB-D Reference Image (CVPR 2025 · similarity 0.87)
- 3D-Aware Hypothesis & Verification for Generalizable Relative Object Pose Estimation (ICLR 2024 · similarity 0.87)
- Rethinking Correspondence-based Category-Level Object Pose Estimation (CVPR 2025 · similarity 0.87)
- SecondPose: SE(3)-Consistent Dual-Stream Feature Fusion for Category-Level Pose Estimation (CVPR 2024 · similarity 0.86)
- Confronting Ambiguity in 6D Object Pose Estimation via Score-Based Diffusion on SE(3) (CVPR 2024 · similarity 0.86)
- Pos3R: 6D Pose Estimation for Unseen Objects Made Easy (CVPR 2025 · 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.
