Research map / 3D vision / Human pose estimation
Occ^2Net: Robust Image Matching Based on 3D Occupancy Estimation for Occluded Regions
Miao Fan, Mingrui Chen, Chen Hu, Shuchang Zhou
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 only visible-occluded supervision to the baseline without architecture changes (ablation) | QuadTree LoFTR · ScanNet | AUC@5: 24.9 -> 24.3 | No clear effect |
| match visible and occluded points rather than visible-visible only (ablation) | Occ2Net · ScanNet | AUC@20: 63.2 -> 66.9 | Improved |
| add 3D occupancy estimation (ablation) | Occ2Net · ScanNet | AUC@5: 25.7 -> 28.0 | Improved |
| add OA module Rotation Alignment to baseline (ablation) | QuadTree LoFTR · ScanNet | AUC@20: 61.8 -> 64.0 | Improved |
| replace LoFTR feature extraction with Occ2Net feature extraction (fewer parameters) (ablation) | QuadTree LoFTR · ScanNet | AUC@5: 24.9 -> 22.6 | Hurt |
| use full Occ2Net-l instead of QuadTree LoFTR on TartanAir | Occ2Net-l · TartanAir | AUC@20: 34.7 -> 47.3 | Improved |
| Occ2Net-s smaller variant vs ASpanFormer on ScanNet AUC@5 | Occ2Net-s · ScanNet | AUC@5: 25.0 vs 25.6 (ASpanFormer) | Mixed |
Closest papers by meaning
- Peeking into occluded joints: A novel framework for crowd pose estimation (ECCV 2020 · similarity 0.88)
- GS-Occ3D: Scaling Vision-only Occupancy Reconstruction with Gaussian Splatting (ICCV 2025 · similarity 0.87)
- Scene as Occupancy (ICCV 2023 · similarity 0.87)
- A Global Occlusion-Aware Approach to Self-Supervised Monocular Visual Odometry (AAAI 2021 · similarity 0.87)
- CSV-Occ: Fusing Multi-frame Alignment for Occupancy Prediction with Temporal Cross State Space Model and Central Voting Mechanism (ICML 2025 · similarity 0.86)
- 3D Occupancy Prediction with Low-Resolution Queries via Prototype-aware View Transformation (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.
