Table 3.
Reconstruction error for 3D shapes for the three datasets
| Dim | Datasets | SPHARM-RPDM | WSPHARM | Diffeomorphic | AE | SRAE | VAE | O-AE |
|---|---|---|---|---|---|---|---|---|
| 7 | HeLa | 8.38 | 20.4 | 14.8* | 16.2 | 16.9(17.0) | 16.2 | 40.9 |
| SNL 3D | 8.64 | 21.4 | – | 52.7 | 132.0(42.5) | 52.7 | 57.9 | |
| SNL NR2 | 12.7 | 24.6 | – | 80.1 | 143.8(51.6) | 80.2 | 73.1 | |
| 100 | HeLa | 4.89 | 20.2 | – | 7.93 | 16.9(17.6) | 10.4 | 42.4 |
| SNL 3D | 4.02 | 16.3 | – | 7.45 | 147.3(10.0) | 28.4 | 56.5 | |
| SNL NR2 | 5.28 | 21.3 | – | 8.19 | 135.0(13.1) | 80.1 | 81.5 |
Note: Same as 2D datasets, we only show the results for 7 dimensions for HeLa 3D dataset for diffeomorphic model. In the parenthesis, we show the reconstruction errors after filtering out isolated voxels in the reconstruction. *The reconstruction error for the diffeomorphic model was calculated with only a subset of the shapes because it failed for many of them.