Research map / 3D vision / Human pose estimation
Synthetic Depth Transfer for Monocular 3D Object Pose Estimation in the Wild
Yueying Kao, Weiming Li, Qiang Wang, Zhouchen Lin, Wooshik Kim, Sunghoon Hong
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 synthetic data to real training (ablation) | RGB-Net · PASCAL 3D+ GT boxes | Acc pi/6 (%): 81.8 -> 85.7 | Improved |
| add RGB-to-Depth Embedding (PD-Net) with real+syn data (ablation) | RGB-Net · PASCAL 3D+ GT boxes | Acc pi/6 (%): 85.7 -> 88.2 | Improved |
| add Synthetic-Real Adaptation (SDA) with real+syn data (ablation) | RGB-Net · PASCAL 3D+ GT boxes | Acc pi/6 (%): 85.7 -> 87.2 | Improved |
| add both PD-Net and SDA (ablation) | RGB-Net · PASCAL 3D+ GT boxes | Acc pi/6 (%): 85.7 -> 89.4 | Improved |
| fuse RGB and depth features from synthetic data only (ablation) | RGB-Net (Syn) · PASCAL 3D+ GT boxes | Acc pi/6 (%): 43.9 -> 76.1 | Improved |
| add depth fusion to synthetic-only training vs PD-Net alone (ablation) | PD-Net (Syn) · PASCAL 3D+ GT boxes | Acc pi/6 (%): 48.8 -> 76.1 | Improved |
| add unsupervised domain adaptation (UDA) to synthetic-only RGB-Net with unlabeled real data (ablation) | RGB-Net (Syn) · PASCAL 3D+ GT boxes | Acc pi/6 (%): 43.9 -> 79.5 | Improved |
| add UDA to RGB-Net+PD-Net (ablation) | RGB-Net+PD-Net (Syn) · PASCAL 3D+ GT boxes | Acc pi/6 (%): 76.1 -> 80.8 | Improved |
| use SSD512 detector instead of RCNN | Ours(final) · PASCAL 3D+ joint detection | Acc pi/6 (%): 53.4 -> 74.1 | Improved |
| evaluate on small vs large objects | Ours(final) · PASCAL 3D+ GT boxes | Acc pi/6 (%): 92.4 -> 85.3 | Hurt |
Closest papers by meaning
- Feature Mapping for Learning Fast and Accurate 3D Pose Inference From Synthetic Images (CVPR 2018 · similarity 0.88)
- SingRef6D: Monocular Novel Object Pose Estimation with a Single RGB Reference (NeurIPS 2025 · similarity 0.87)
- Free-Moving Object Reconstruction and Pose Estimation with Virtual Camera (AAAI 2025 · similarity 0.86)
- NOPE: Novel Object Pose Estimation from a Single Image (CVPR 2024 · similarity 0.85)
- RGBD Objects in the Wild: Scaling Real-World 3D Object Learning from RGB-D Videos (CVPR 2024 · similarity 0.85)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.85)
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.
