Unconditional 3D Point Cloud Generation on ShapeNet Chair v1 (test)
2.409MMD (CD)PointFlow
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| PointFlow2022.12 | 2.409 | 1.595 | 42.9 | 50 | |
| SoftFlow2022.12 | 2.528 | 1.682 | 41.39 | 47.43 | |
| DPF-Net2022.12 | 2.536 | 1.632 | 44.71 | 48.79 | |
| l-GAN (CD)Training metric=CD2022.12 | 2.589 | 2.007 | 41.99 | 29.31 | |
| PVD2022.12 | 2.622 | 1.556 | 49.84 | 50.6 | |
| PSFSampling steps (N)=12022.12 | 2.624 | 1.573 | 46.71 | 49.84 | |
| PVD-DDIMSampling protocol=DDIM, Sampling steps (N)=1002022.12 | 2.758 | 1.703 | 46.32 | 48.19 | |
| Shape-GF2022.12 | 2.889 | 1.702 | 46.67 | 48.03 |