Research map / Point clouds / Registration and matching
3DRegNet: A Deep Neural Network for 3D Point Registration
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 a second, smaller 3DRegNet for refinement (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 1.37 -> 1.19 | Helped |
| Add a second, smaller 3DRegNet for refinement (ablation) | 3DRegNet · ICL-NUIM | classification accuracy: 0.96 -> 0.94 | Hurt |
| Use L1 distance in the registration loss instead of L2 (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 2.44 -> 1.37 | Helped |
| Use L1 distance instead of weighted L2 distance (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 1.89 -> 1.37 | Helped |
| Use L1 distance instead of Geman-McClure distance (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 2.45 -> 1.37 | Helped |
| Use minimal Lie algebra rotation parameters instead of quaternions (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 1.55 -> 1.37 | Helped |
| Use minimal Lie algebra rotation parameters instead of linear matrix parameters (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 5.78 -> 1.37 | Helped |
| Use DNN regression instead of differentiable Procrustes (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 1.65 -> 1.37 | Helped |
| Use more point correspondences (ablation) | 3DRegNet · ICL-NUIM | rotation error (mean, deg): 2.40 -> 1.37 | Helped |
| Use curriculum-learning data augmentation with rotations up to 50° (ablation) | 3DRegNet · ICL-NUIM | rotation error | Helped |
| Initialize ICP with 3DRegNet rather than FGR | ICP registration · ICL-NUIM | rotation error (mean, deg): 1.01 -> 0.55 | Helped |
| Initialize ICP with 3DRegNet rather than FGR | ICP registration · SUN3D unseen sequences | rotation error (mean, deg): 1.49 -> 1.26 | Helped |
Closest papers by meaning
- Deep Confidence Guided Distance for 3D Partial Shape Registration (AAAI 2022 · similarity 0.87)
- PointCNN++: Performant Convolution on Native Points (CVPR 2026 · similarity 0.86)
- Deep Global Registration (CVPR 2020 · similarity 0.86)
- VGGT: Visual Geometry Grounded Transformer (CVPR 2025 · similarity 0.85)
- PosDiffNet: Positional Neural Diffusion for Point Cloud Registration in a Large Field of View with Perturbations (AAAI 2024 · similarity 0.84)
- Scalable 3D Registration via Truncated Entry-wise Absolute Residuals (CVPR 2024 · similarity 0.84)
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.