Research map / 3D vision / Human pose estimation
Occlusion-Aware Self-Supervised Monocular 6D Object Pose Estimation
Gu Wang, Fabian Manhardt, Xingyu Liu, Xiangyang Ji, 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 |
|---|---|---|---|
| add self-supervision with unlabeled real RGB-D data | OURS(LB) · Occluded LINEMOD (BOP test) | average recall ADD(-S): 52.9% -> 64.7% | Improved |
| add self-supervision with unlabeled real RGB data only | OURS(LB) · Occluded LINEMOD (BOP test) | average recall ADD(-S): 52.9 -> 59.8 | Improved |
| add self-supervision on a few unlabeled HomebrewedDB images (RGB-D) | OURS(LB) · HomebrewedDB | average recall ADD(-S): 3.1% -> 84.4% | Improved |
| add self-supervision (RGB-D) | OURS(LB) · LINEMOD | average recall ADD(-S): 77.4 -> 88.5 | Improved |
| add self-supervision (RGB-D) | OURS(LB) · YCB-Video | AUC of ADD(-S): 77.8 -> 80.0 | Improved |
| add self-supervision (RGB-D) | OURS(LB) · Cropped LINEMOD | mean angle error (deg): 11.2 -> 3.9 | Improved |
| add geometric loss Lgeom (ablation) | Self6D++ RGB-D · Occluded LINEMOD (BOP test) | average recall ADD(-S): 5.1% -> 64.7% | Improved |
| add amodal mask prediction (ablation) | Self6D++ RGB-D · Occluded LINEMOD (BOP test) | average recall ADD(-S): 55.1% -> 64.7% | Improved |
| add data augmentation on student input (noisy student) (ablation) | Self6D++ RGB-D · Occluded LINEMOD (BOP test) | average recall ADD(-S): 62.1% -> 64.7% | Improved |
| add visual loss Lvisual (ablation) | Self6D++ RGB-D · Occluded LINEMOD (BOP test) | average recall ADD(-S): 63.2 -> 64.7 | Improved |
| add point matching loss Lpm (ablation) | Self6D++ RGB-D · Occluded LINEMOD (BOP test) | average recall ADD(-S): 64.3 -> 64.7 | Improved |
| predict both amodal and visible masks instead of visible only (ablation) | Self6D++ · LINEMOD | average recall ADD(-S): 88.3 -> 88.5 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Self6D — Self6D: Self-Supervised Monocular 6D Object Pose Estimation (ECCV 2020) · better on 3 of 3 shared results
- DPOD — DPOD: 6D Pose Object Detector and Refiner (ICCV 2019) · better on 1 of 1 shared result
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
- Self6D: Self-Supervised Monocular 6D Object Pose Estimation (ECCV 2020 · similarity 0.97)
- SO-Pose: Exploiting Self-Occlusion for Direct 6D Pose Estimation (ICCV 2021 · similarity 0.93)
- Pseudo Flow Consistency for Self-Supervised 6D Object Pose Estimation (ICCV 2023 · similarity 0.91)
- One2Any: One-Reference 6D Pose Estimation for Any Object (CVPR 2025 · similarity 0.89)
- SingRef6D: Monocular Novel Object Pose Estimation with a Single RGB Reference (NeurIPS 2025 · similarity 0.89)
- ONDA-Pose: Occlusion-Aware Neural Domain Adaptation for Self-Supervised 6D Object Pose Estimation (CVPR 2025 · 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.
