Research map / Point clouds / Registration and matching
Hg-I2P: Bridging Modalities for Generalizable Image-to-Point-Cloud Registration via Heterogeneous Graphs
Pei An, Junfeng Ding, Jiaqi Yang, Yulong Wang, Jie Ma, Liangliang Nan
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 multi-path edges E2 and E3 to MP-mining (vs E1 only) (ablation) | Hg-I2P · 7-Scenes train Office (ablation) | RR: 0.610 -> 0.661 | Improved |
| add multi-path edges E2 and E3 to MP-mining (vs E1 only) (ablation) | Hg-I2P · 7-Scenes train Office (ablation) | IR: 0.478 -> 0.494 | Improved |
| enable HE-adapting with beta 0.10 (vs beta 0) (ablation) | Hg-I2P · 7-Scenes train Office (ablation) | RR: 0.620 -> 0.661 | Improved |
| raise beta from 0.10 to 0.25 (ablation) | Hg-I2P · 7-Scenes train Office (ablation) | RR: 0.661 -> 0.582 | Hurt |
| add HC-pruning criterion I (reprojection distance) (ablation) | Hg-I2P · 7-Scenes train Office (ablation) | IR: 0.494 -> 0.523 | Improved |
| add HC-pruning criterion II (relative position consistency) (ablation) | Hg-I2P · 7-Scenes train Office (ablation) | IR: 0.523 -> 0.558 | Improved |
| add HC-pruning criterion II (ablation) | Hg-I2P · 7-Scenes train Office (ablation) | RR: 0.682 -> 0.690 | Improved |
| add HC-pruning on top of heterogeneous graph features (ablation) | Hg-I2P · 7-Scenes train Chess | RR (avg): 0.642 -> 0.667 | Improved |
| add HC-pruning (ablation) | Hg-I2P · 7-Scenes train Chess | IR (avg): 0.472 -> 0.581 | Improved |
| use Hg-I2P instead of MinCD | Hg-I2P · 7-Scenes train Office | IR (avg): 0.568 -> 0.558 | Hurt |
| use Hg-I2P instead of MinCD | Hg-I2P · Cross-dataset RGBD-V2 | RR (avg): 0.870 -> 0.868 | No clear effect |
| use Hg-I2P instead of Top-I2P | Hg-I2P · Cross-dataset ScanNet | IR (avg): 0.493 -> 0.586 | Improved |
Closest papers by meaning
- SemReg: Semantics Constrained Point Cloud Registration (ECCV 2024 · similarity 0.89)
- GraphI2P: Image-to-Point Cloud Registration with Exploring Pattern of Correspondence via Graph Learning (CVPR 2025 · similarity 0.88)
- Partial Point Cloud Registration with Multi-view 2D Image Learning (AAAI 2025 · similarity 0.88)
- FreeReg: Image-to-Point Cloud Registration Leveraging Pretrained Diffusion Models and Monocular Depth Estimators (ICLR 2024 · similarity 0.88)
- Mining and Transferring Feature-Geometry Coherence for Unsupervised Point Cloud Registration (NeurIPS 2024 · similarity 0.86)
- HybridReg: Robust 3D Point Cloud Registration with Hybrid Motions (AAAI 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.
