Research map / 3D vision / Human pose estimation
Category-Level 6D Object Pose and Size Estimation Using Self-Supervised Deep Prior Deformation Networks
Jiehong Lin, Zewei Wei, Changxing Ding, Kui Jia
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 Deep Prior Deformer (ablation) | DPDN with parallel learning · REAL275 test (supervised) | mAP 5deg 2cm: 35.2 -> 46.0 | Improved |
| add Deep Prior Deformer (ablation) | DPDN single input · REAL275 test (supervised) | mAP 5deg 2cm: 26.7 -> 39.6 | Improved |
| add Pose and Size Estimator instead of Umeyama on canonical points (ablation) | DPDN · REAL275 test (supervised) | mAP IoU50: 59.6 -> 83.4 | Improved |
| add parallel dual-input supervision (ablation) | DPDN with deformer · REAL275 test (supervised) | mAP 5deg 2cm: 39.6 -> 46.0 | Improved |
| add parallel dual-input supervision (ablation) | DPDN without deformer · REAL275 test (supervised) | mAP 5deg 2cm: 26.7 -> 35.2 | Improved |
| add parallel dual-input supervision (ablation) | DPDN with deformer · REAL275 test (supervised) | mAP IoU75: 76.2 -> 76.0 | No clear effect |
| add inter-consistency loss (ablation) | DPDN · REAL275 test (unsupervised) | mAP 5deg 2cm: 29.7 -> 36.9 | Improved |
| add intra-consistency losses (ablation) | DPDN · REAL275 test (unsupervised) | mAP 5deg 2cm: 29.7 -> 35.8 | Improved |
| add intra-consistency losses without inter-consistency (ablation) | DPDN · REAL275 test (unsupervised) | mAP IoU75: 60.8 -> 58.6 | Hurt |
| combine inter- and intra-consistency (ablation) | DPDN · REAL275 test (unsupervised) | mAP 5deg 2cm: 29.7 -> 37.8 | Improved |
| add unlabeled real-world images via self-supervision (ablation) | DPDN · REAL275 test (unsupervised) | mAP 5deg 2cm: 29.7 -> 37.8 | Improved |
| increase ratio of unlabeled real training data (ablation) | Self-DPDN · REAL275 test (unsupervised) | mAP | Improved |
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.
- DualPoseNet — DualPoseNet: Category-Level 6D Object Pose and Size Estimation Using Dual Pose Network With Refined Learning of Pose Consistency (ICCV 2021) · better on 4 of 4 shared results
- SGPA — SGPA: Structure-Guided Prior Adaptation for Category-Level 6D Object Pose Estimation (ICCV 2021) · better on 3 of 3 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
- Shape Prior Deformation for Categorical 6D Object Pose and Size Estimation (ECCV 2020 · similarity 0.92)
- FS-Net: Fast Shape-Based Network for Category-Level 6D Object Pose Estimation With Decoupled Rotation Mechanism (CVPR 2021 · similarity 0.89)
- DSC-PoseNet: Learning 6DoF Object Pose Estimation via Dual-Scale Consistency (CVPR 2021 · similarity 0.89)
- DualPoseNet: Category-Level 6D Object Pose and Size Estimation Using Dual Pose Network With Refined Learning of Pose Consistency (ICCV 2021 · similarity 0.89)
- PriorPose: Reference-Guided Joint Deformation and Alignment for Category-Level Object Pose Estimation (ECCV 2026 · similarity 0.89)
- CAP-Net: A Unified Network for 6D Pose and Size Estimation of Categorical Articulated Parts from a Single RGB-D Image (CVPR 2025 · similarity 0.87)
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.
