Research map / Point clouds / Registration and matching
Sampling Network Guided Cross-Entropy Method for Unsupervised Point Cloud Registration
Haobo Jiang, Yaqi Shen, Jin Xie, Jun Li, Jianjun Qian, Jian Yang
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 sampling network prior to differentiable CEM (without future reward) (ablation) | CEMNet · ModelNet40 unseen category | MAE(R): 1.8625 -> 0.1781 | Improved |
| add sampling network prior to differentiable CEM (without future reward) (ablation) | CEMNet · ICL-NUIM | MAE(R): 1.8333 -> 0.3345 | Improved |
| add sampling network when future reward is used (ablation) | CEMNet · ModelNet40 unseen category | MAE(R): 0.3675 -> 0.0804 | Improved |
| add sampling network when future reward is used (ablation) | CEMNet · ICL-NUIM | MAE(R): 0.8364 -> 0.0211 | Improved |
| add ICP-driven future reward to score function (with SN) (ablation) | CEMNet · ModelNet40 unseen category | MAE(R): 0.1781 -> 0.0804 | Improved |
| add ICP-driven future reward to score function (with SN) (ablation) | CEMNet · ICL-NUIM | MAE(R): 0.3345 -> 0.0211 | Improved |
| add ICP-driven future reward to score function (without SN) (ablation) | CEMNet · ICL-NUIM | MAE(t): 0.0023 -> 0.0048 | Hurt |
| add CEM planning on top of sampling network (SN alone as DCP-v2) (ablation) | Sampling network · ModelNet40 unseen category | MAE(R): 4.4222 -> 0.1781 | Improved |
| increase iterations using future reward M from 0 to 4 (ablation) | CEMNet · ModelNet40 unseen category | MAE(R): 0.1782 -> 0.0470 | Improved |
| increase iterations using future reward M from 0 to 4 (ablation) | CEMNet · ModelNet40 unseen object | MAE(R): 0.3505 -> 0.1258 | Improved |
| increase number of iterations T and candidates N (ablation) | CEMNet · ICL-NUIM | rotation error | Improved |
| balance current and future reward weights alpha (ablation) | CEMNet · ICL-NUIM | rotation error | Improved |
Closest papers by meaning
- HeMoRa: Unsupervised Heuristic Consensus Sampling for Robust Point Cloud Registration (CVPR 2025 · similarity 0.89)
- Where Precision Meets Efficiency: Transformation Diffusion Model for Point Cloud Registration (AAAI 2025 · similarity 0.89)
- Cross-PCR: A Robust Cross-Source Point Cloud Registration Framework (AAAI 2025 · similarity 0.88)
- Unsupervised Deep Probabilistic Approach for Partial Point Cloud Registration (CVPR 2023 · similarity 0.88)
- FreeReg: Image-to-Point Cloud Registration Leveraging Pretrained Diffusion Models and Monocular Depth Estimators (ICLR 2024 · similarity 0.88)
- GCD-Sampling: A General Cross-scale Decoupled Sampling for Point Cloud (AAAI 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.
