Metric: Recall ( correspondence RMSE below 0.2) (lower is better)
| # | Model↕ | Recall ( correspondence RMSE below 0.2)▲ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | OMNet (reported in REGTR) | 35.9 | No | OMNet: Learning Overlapping Mask for Partial-to-... | 2021-03-01 | Code |
| 2 | Predator-NR | 62.7 | No | PREDATOR: Registration of 3D Point Clouds with L... | 2020-11-25 | Code |
| 3 | 3DSN (reported in PREDATOR) | 78.4 | No | The Perfect Match: 3D Point Cloud Matching with ... | 2018-11-16 | Code |
| 4 | D3Feat (reported in PREDATOR) | 81.6 | No | D3Feat: Joint Learning of Dense Detection and De... | 2020-03-06 | Code |
| 5 | FCGF (reported in PREDATOR) | 85.1 | No | - | - | Code |
| 6 | DGR (reported in REGTR) | 85.3 | No | Deep Global Registration | 2020-04-24 | Code |
| 7 | PCAM (reported in REGTR) | 85.5 | No | PCAM: Product of Cross-Attention Matrices for Ri... | 2021-10-04 | Code |
| 8 | Predator-5k | 89 | No | PREDATOR: Registration of 3D Point Clouds with L... | 2020-11-25 | Code |
| 9 | Predator-1k | 90.5 | No | PREDATOR: Registration of 3D Point Clouds with L... | 2020-11-25 | Code |
| 10 | REGTR | 92 | No | REGTR: End-to-end Point Cloud Correspondences wi... | 2022-03-28 | Code |
| 11 | NgeNet | 92.9 | No | Leveraging Inlier Correspondences Proportion for... | 2022-01-28 | Code |