Benchmarks

Task NameDataset NameSOTA ResultTrendResults

Point Cloud Alignment

ModelNet10 (test)

0.947Pearson Correlation

Point Cloud Alignment

ModelNet10 (train)

0.959Pearson Correlation

3D Point Cloud Attack

ModelNet40

0.012Hausdorff Distance

Dexterous Grasping

ModelNet40

87.89GSR

Graph Classification

MODELNET

39.81Accuracy

Retrieval

ModelNet40

57.3Top-1 Accuracy (Instance)

Point cloud registration

ModelNet40 8 axisymmetrical categories PRNet/RPMNet generation (twice-sampled (TS))

6.258RMSE (R)

Point cloud registration

ModelNet40 8 axisymmetrical categories PRNet/RPMNet generation (once-sampled (OS))

0.312RMSE (Rotation)

Topological Feature Selection

ModelNet40 6 (test)

0.024Mean WD (H1)

3D Object Classification

ModelNet10

75.7Top-1 Accuracy

Few-Shot Learning

ModelNet40

97.1Acc (5w10s)

Point-cloud reconstruction

ModelNet40 Epoch 200 (test)

2.04SW2 Error

Point-cloud reconstruction

ModelNet40 Epoch 100 (test)

2.18SW2

Point-cloud reconstruction

ModelNet40 Epoch 20 (test)

2.68SW2

3D Object Classification

ModelNet40 (val)

0.731Top-1 Accuracy

3D Classification

ModelNet with Mix3D

91.8Accuracy

Multi-instance Point Cloud Registration

ModelNet40 Synthetic (90%~99% Outlier Ratio)

14.7MHR

Multi-instance Point Cloud Registration

Synthetic ModelNet40 10%~50% Outlier Ratio

96.08MHR (%)

Backdoor Attack

ModelNet40

14.74MFLOPs

Point Cloud Recognition

ModelNet-C

48.78Acc (Add Global)

Open-ended 3D Object Classification

Modelnet Open

47.9Accuracy

1-Wasserstein distance approximation

ModelNet large

8Training Time

1-Wasserstein distance approximation

ModelNet small

7Training Time (min)

3D Generation

ModelNet40 Table (test)

1,468.43FPD

3D Generation

ModelNet40 Chair (test)

342.23FPD