3D Object Classification on ScanNet (test)
84.61Test AccuracyPointNet
Evaluation Results
| Method | Links | |
|---|---|---|
| PointNetAttack type=No attack2022.03 | 84.61 | |
| PointNetDefense method=DUP-Net, Attack type=No attack2022.03 | 83.62 | |
| LPCBackbone=EfficientNet, Attack type=No attack2022.03 | 83.16 | |
| LPCBackbone=EfficientNet, Attack type=FGSM2022.03 | 83.16 | |
| LPCBackbone=EfficientNet, Attack type=JGBA2022.03 | 83.16 | |
| PointNetDefense method=IF-Defense, Attack type=No attack2022.03 | 80.19 | |
| PointNetDefense method=RPL, Attack type=No attack2022.03 | 76.02 | |
| PointNetDefense method=DUP-Net, Attack type=FGSM2022.03 | 73.67 | |
| PointNetDefense method=IF-Defense, Attack type=FGSM2022.03 | 71.14 | |
| PointNetAttack type=FGSM2022.03 | 45.66 | |
| PointNetDefense method=IF-Defense, Attack type=JGBA2022.03 | 13.45 | |
| PointNetDefense method=DUP-Net, Attack type=JGBA2022.03 | 7.77 | |
| PointNetDefense method=RPL, Attack type=FGSM2022.03 | 1.7 | |
| PointNetAttack type=JGBA2022.03 | 0 | |
| PointNetDefense method=RPL, Attack type=JGBA2022.03 | 0 |