Research map / 3D vision / Human pose estimation
WeLSA: Learning to Predict 6D Pose from Weakly Labeled Data Using Shape Alignment
Shishir Reddy Vutukur, Ivan Shugurov, Benjamin Busam, Andreas Hutter, Slobodan Ilic
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 decoder (color) losses (ablation) | WeLSA Stage 2 · LineMOD driller, weakly-labeled samples | pose labeling accuracy: 8.9% increase | Improved |
| add SDF shape loss L'_S (ablation) | WeLSA Stage 2 · LineMOD | pose accuracy: drop of 1.5% when removed | Improved |
| add Stage 2 weakly-labeled training after Stage 1 (ablation) | WeLSA · LineMOD | ADD(-S) accuracy (%): 75.3 -> 91.5 | Improved |
| add Stage 3 ICP-refined pseudo-label training (ablation) | WeLSA · LineMOD | ADD(-S) accuracy (%): 91.5 -> 98.1 | Improved |
| add ICP refinement with reconstructed mesh (ablation) | WeLSA · LineMOD | ADD(-S) accuracy (%): 1.0% improvement, 99.1% | Improved |
| add ICP refinement to Stage 2 labels (ablation) | WeLSA labels · LineMOD weakly-labeled | labeling accuracy: 90.6% -> 99.2% | Improved |
| increase ground truth data from 1% to 15% (ablation) | WeLSA · LineMOD | ADD(-S) accuracy (%): 99.1 -> 99.2 | Improved |
| add ICP refinement (ablation) | WeLSA · TLess (1.5% GT) | VSD recall: 0.34 -> 0.46 | Improved |
| use 1% GT with weak labels instead of 5% GT supervised G2L-Net | WeLSA vs G2L-Net · LineMOD (5% total) | ADD(-S) accuracy (%): 89% -> 94% | Improved |
| use 1.5% GT instead of 100% GT | WeLSA · TLess | BOP average recall: 0.49 -> 0.30 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 2 on most shared results.
- DpodV2 — DPODv2: Dense Correspondence-Based 6 DoF Pose Estimation (TPAMI 2021) · better on 0 of 2 shared results
- Pix2Pose — Pix2Pose: Pixel-Wise Coordinate Regression of Objects for 6D Pose Estimation (ICCV 2019) · better on 0 of 2 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
- Pose without Guesses: Generalizable Object Pose Estimation From a Single Reference (IJCV 2026 · similarity 0.88)
- Learning-Based Point Cloud Registration for 6D Object Pose Estimation in the Real World (ECCV 2022 · similarity 0.88)
- SAR-Net: Shape Alignment and Recovery Network for Category-Level 6D Object Pose and Size Estimation (CVPR 2022 · similarity 0.88)
- Unsupervised Learning of Category-Level 3D Pose from Object-Centric Videos (CVPR 2024 · similarity 0.87)
- One2Any: One-Reference 6D Pose Estimation for Any Object (CVPR 2025 · similarity 0.86)
- Co-op: Correspondence-based Novel Object Pose Estimation (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.
