Point Cloud Classification on ShapeNet (test)
99.3PointNet Instance AccuracyCurveCloudNet
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CurveCloudNetTime (ms)=37, GPU (GB)=0.66, Param (M)=10.32023.03 | 99.3 | 96.3 | — | — | 96 | — | |
| PointMLPTime (ms)=54, GPU (GB)=0.76, Param (M)=13.22023.03 | 99.2 | 94.8 | — | — | 95.3 | — | |
| PointNet++Time (ms)=51, GPU (GB)=0.91, Param (M)=1.62023.03 | 99 | 95.3 | — | — | 95.5 | — | |
| NePSUpsampling Factor=4x2023.12 | 98.94 | 96.2 | 99.12 | 97.94 | — | — | |
| DGCNNTime (ms)=73, GPU (GB)=0.78, Param (M)=0.62023.03 | 98.9 | 93.7 | — | — | 93.6 | — | |
| High-resDescription=Original test point cloud2023.12 | 98.89 | 96.61 | 99.27 | 98.18 | — | — | |
| PUDMUpsampling Factor=4x2023.12 | 98.85 | 96.58 | 99.13 | 97.99 | — | — | |
| Grad-PUUpsampling Factor=4x2023.12 | 98.82 | 96.19 | 99.1 | 97.63 | — | — | |
| Dis-PUUpsampling Factor=4x2023.12 | 98.8 | 96.07 | 99 | 97.15 | — | — | |
| PU-GCNUpsampling Factor=4x2023.12 | 98.78 | 96.06 | 99.03 | 97.42 | — | — | |
| PU-GANUpsampling Factor=4x2023.12 | 98.75 | 95.7 | 99 | 97.15 | — | — | |
| PU-EVAUpsampling Factor=4x2023.12 | 98.72 | 95.69 | 99.07 | 97.58 | — | — | |
| MPUUpsampling Factor=4x2023.12 | 98.03 | 95.92 | 98.94 | 96.81 | — | — | |
| PU-NetUpsampling Factor=4x2023.12 | 97.99 | 95.69 | 98.57 | 96.35 | — | — | |
| Low-resResolution=256/512 points2023.12 | 97.61 | 95.09 | 98.2 | 96.11 | — | — | |
| CleanBackbone=DGCNN, Attack=Point Addition2026.02 | — | — | — | — | — | 0.52 | |
| CleanBackbone=PointNet, Attack=Point Addition2026.02 | — | — | — | — | — | 3.62 | |
| PointCVARBackbone=DGCNN, Attack=Point Addition2026.02 | — | — | — | — | — | 95.48 | |
| PointCVARBackbone=PointNet, Attack=Point Addition2026.02 | — | — | — | — | — | 97.7 | |
| PWavePBackbone=DGCNN, Attack=Point Addition2026.02 | — | — | — | — | — | 96.24 | |
| PWavePBackbone=PointNet, Attack=Point Addition2026.02 | — | — | — | — | — | 98.47 |