Object Classification on ModelNet40 (test)
95.54AccuracyVRN
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| VRNEnsemble=true2016.08 | 95.54 | — | — | — | — | — | |
| VRN ensemble2017.11 | 95.54 | — | — | — | — | — | |
| PointMLPLearning Protocol=Supervised Learning Only, #TP (M)=14.9, Data Type=1k Points2023.04 | 94.5 | — | — | — | — | — | |
| Point-MAE w/ IDPTLearning Protocol=IDPT, Standard Transformer=true, #TP (M)=1.7, Data Type=1k Points2023.04 | 94.4 | — | — | — | — | — | |
| CurveNetInput=xyz, #point=1024, voting strategy=true2021.05 | 94.2 | — | — | — | — | — | |
| RPNet-W9Modality=Points+Normals2021.08 | 94.1 | — | — | — | — | — | |
| Point-M2AELearning Protocol=Full Fine-tuning, Standard Transformer=false, #TP (M)=15.3, Data Type=1k Points2023.04 | 94 | — | — | — | — | — | |
| CLIP2PointLearning Protocol=Full Fine-tuning, Data Type=10 Images2023.04 | 94 | — | — | — | — | — | |
| P2PLearning Protocol=Full Fine-tuning, #TP (M)=1.2, Data Type=1 Images2023.04 | 94 | — | — | — | — | — | |
| ACT w/ IDPTLearning Protocol=IDPT, Standard Transformer=true, #TP (M)=1.7, Data Type=1k Points2023.04 | 94 | — | — | — | — | — | |
| PAConvInput=1K points, Backbone=DGCNN, voting=true2021.03 | 93.9 | — | — | — | — | — | |
| PAConvInput=xyz, #point=1024, voting strategy=true2021.05 | 93.9 | — | — | — | — | — | |
| SimpleViewLearning Protocol=Supervised Learning Only, Data Type=6 Images2023.04 | 93.9 | — | — | — | — | — | |
| RPNet-W9Modality=Points2021.08 | 93.9 | — | — | — | — | — | |
| RPNet-W7Modality=Points+Normals2021.08 | 93.9 | — | — | — | — | — | |
| CurveNetInput=xyz, #point=1024, voting strategy=false2021.05 | 93.8 | — | — | — | — | — | |
| Point-MAESupervision Protocol=Self-supervised, Model Architecture Annotation=standard Transformers2022.03 | 93.8 | — | — | — | — | — | |
| Point-BERT#point=8k, Architecture=Standard Transformer [ST]2021.11 | 93.8 | — | — | — | — | — | |
| MVTNLearning Protocol=Supervised Learning Only, #TP (M)=11.2, Data Type=12 Images2023.04 | 93.8 | — | — | — | — | — | |
| Point-MAELearning Protocol=Full Fine-tuning, Standard Transformer=true, #TP (M)=22.1, Data Type=1k Points2023.04 | 93.8 | — | — | — | — | — | |
| MaskPointLearning Protocol=Full Fine-tuning, Standard Transformer=true, #TP (M)=22.1, Data Type=1k Points2023.04 | 93.8 | — | — | — | — | — | |
| RPNet-W7Modality=Points2021.08 | 93.8 | — | — | — | — | — | |
| IAESupervision Protocol=Self-supervised2022.03 | 93.7 | — | — | — | — | — | |
| PointTransformerSupervision Protocol=Supervised, Model Architecture Annotation=modified Transformers2022.03 | 93.7 | — | — | — | — | — | |
| PTInput=xyz+nor2021.11 | 93.7 | — | — | — | — | — | |
| Point Transformer#point=Not specified, Architecture=Transformer with special designs [T]2021.11 | 93.7 | — | — | — | — | — | |
| PointTransformerLearning Protocol=Supervised Learning Only, Standard Transformer=false, Data Type=1k Points2023.04 | 93.7 | — | — | — | — | — | |
| ACTLearning Protocol=Full Fine-tuning, Standard Transformer=true, #TP (M)=22.1, Data Type=1k Points2023.04 | 93.7 | — | — | — | — | — | |
| RS-CNNInput=1K points, voting=true2021.03 | 93.6 | — | — | — | — | — | |
| PAConvInput=1K points, Backbone=DGCNN, voting=false2021.03 | 93.6 | — | — | — | — | — | |
| RS-CNNInput=xyz, #point=1024, voting strategy=true2021.05 | 93.6 | — | — | — | — | — | |
| PAConvInput=xyz, #point=1024, voting strategy=false2021.05 | 93.6 | — | — | — | — | — | |
| PVTSupervision Protocol=Supervised, Model Architecture Annotation=modified Transformers2022.03 | 93.6 | — | — | — | — | — | |
| PAConvInput=xyz, Backbone=DGCNN, is_reimplementation=false2021.11 | 93.6 | — | — | — | — | — | |
| PVTLearning Protocol=Supervised Learning Only, Standard Transformer=false, Data Type=1k Points2023.04 | 93.6 | — | — | — | — | — | |
| PVRNetModality=Points+Views2021.08 | 93.6 | — | — | — | — | — | |
| RS-CNN w/ vot.Modality=Points2021.08 | 93.6 | — | — | — | — | — | |
| DSPointInput=xyz2021.11 | 93.5 | — | — | — | — | — | |
| SO-NetInput=xyz, nr, #point=5000, voting strategy=false2021.05 | 93.4 | — | — | — | — | — | |
| PAConvInput=xyz, Backbone=DGCNN, is_reimplementation=true2021.11 | 93.4 | — | — | — | — | — | |
| Point-BERT#point=4k, Architecture=Standard Transformer [ST]2021.11 | 93.4 | — | — | — | — | — | |
| Point-BERT w/ IDPTLearning Protocol=IDPT, Standard Transformer=true, #TP (M)=1.7, Data Type=1k Points2023.04 | 93.4 | — | — | — | — | — | |
| SO-NetModality=Points+Normals (5k)2021.08 | 93.4 | — | — | — | — | — | |
| DensePointInput=1K points2021.03 | 93.2 | — | — | — | — | — | |
| PosPoolInput=5K points2021.03 | 93.2 | — | — | — | — | — | |
| PAConvInput=1K points, Backbone=PointNet, voting=false2021.03 | 93.2 | — | — | — | — | — | |
| PCTInput=xyz, #point=1024, voting strategy=false2021.05 | 93.2 | — | — | — | — | — | |
| PosPoolInput=xyz, #point=2048, voting strategy=true2021.05 | 93.2 | — | — | — | — | — | |
| Point-BERTSupervision Protocol=Self-supervised, Model Architecture Annotation=standard Transformers2022.03 | 93.2 | — | — | — | — | — | |
| PCTSupervision Protocol=Supervised, Model Architecture Annotation=modified Transformers2022.03 | 93.2 | — | — | — | — | — | |
| PAConvInput=xyz, Backbone=PointNet, is_reimplementation=false2021.11 | 93.2 | — | — | — | — | — | |
| PCTInput=xyz, is_reimplementation=false2021.11 | 93.2 | — | — | — | — | — | |
| PointASNLInput=xyz+nor2021.11 | 93.2 | — | — | — | — | — | |
| PCT#point=1k, Architecture=Transformer with special designs [T]2021.11 | 93.2 | — | — | — | — | — | |
| Point-BERT#point=1k, Architecture=Standard Transformer [ST]2021.11 | 93.2 | — | — | — | — | — | |
| PCTLearning Protocol=Supervised Learning Only, Standard Transformer=false, #TP (M)=2.9, Data Type=1k Points2023.04 | 93.2 | — | — | — | — | — | |
| Point-BERTLearning Protocol=Full Fine-tuning, Standard Transformer=true, #TP (M)=22.1, Data Type=1k Points2023.04 | 93.2 | — | — | — | — | — | |
| PointASNLModality=Points+Normals2021.08 | 93.2 | — | — | — | — | — | |
| Grid-GCNInput=1K points2021.03 | 93.1 | — | — | — | — | — | |
| Grid-CNNInput=xyz, #point=1024, voting strategy=false2021.05 | 93.1 | — | — | — | — | — | |
| STRLSupervision Protocol=Self-supervised2022.03 | 93.1 | — | — | — | — | — | |
| Grid-GCNModality=Points+Normals2021.08 | 93.1 | — | — | — | — | — | |
| InterpCNNInput=1K points2021.03 | 93 | — | — | — | — | — | |
| Point2NodeInput=1K points2021.03 | 93 | — | — | — | — | — | |
| OcCoSupervision Protocol=Self-supervised2022.03 | 93 | — | — | — | — | — | |
| KPConvInput=1K points2021.03 | 92.9 | — | — | — | — | — | |
| DGCNNInput=1K points2021.03 | 92.9 | — | — | — | — | — | |
| DGCNNInput=xyz, #point=1024, voting strategy=false2021.05 | 92.9 | — | — | — | — | — | |
| PointASNLInput=xyz, #point=1024, voting strategy=false2021.05 | 92.9 | — | — | — | — | — | |
| RS-CNNInput=xyz, #point=1024, voting strategy=false2021.05 | 92.9 | — | — | — | — | — | |
| KPConvSupervision Protocol=Supervised2022.03 | 92.9 | — | — | — | — | — | |
| DGCNNSupervision Protocol=Supervised2022.03 | 92.9 | — | — | — | — | — | |
| RS-CNNSupervision Protocol=Supervised2022.03 | 92.9 | — | — | — | — | — | |
| DGCNNInput=xyz2021.11 | 92.9 | — | — | — | — | — | |
| KPConvInput=xyz2021.11 | 92.9 | — | — | — | — | — | |
| DGCNN#point=1k2021.11 | 92.9 | — | — | — | — | — | |
| RSCNN#point=1k2021.11 | 92.9 | — | — | — | — | — | |
| KPConv#point=~6.8k2021.11 | 92.9 | — | — | — | — | — | |
| DGCNNLearning Protocol=Supervised Learning Only, #TP (M)=1.8, Data Type=1k Points2023.04 | 92.9 | — | — | — | — | — | |
| EPCLLearning Protocol=Full Fine-tuning, Data Type=1k Points2023.04 | 92.9 | — | — | — | — | — | |
| KPConvModality=Grid2021.08 | 92.9 | — | — | — | — | — | |
| PointASNLModality=Points2021.08 | 92.9 | — | — | — | — | — | |
| PCTInput=xyz, is_reimplementation=true2021.11 | 92.8 | — | — | — | — | — | |
| DensePoint#point=1k2021.11 | 92.8 | — | — | — | — | — | |
| A-CNNInput=xyz, nr, #point=1024, voting strategy=false2021.05 | 92.6 | — | — | — | — | — | |
| PointCNNInput=1K points2021.03 | 92.5 | — | — | — | — | — | |
| PointConvInput=1K points+normal2021.03 | 92.5 | — | — | — | — | — | |
| FPConvInput=1K points2021.03 | 92.5 | — | — | — | — | — | |
| PointCNNSupervision Protocol=Supervised2022.03 | 92.5 | — | — | — | — | — | |
| FPConvInput=xyz2021.11 | 92.5 | — | — | — | — | — | |
| PAConvInput=xyz, Backbone=PointNet, is_reimplementation=true2021.11 | 92.5 | — | — | — | — | — | |
| SO-Net#point=1k2021.11 | 92.5 | — | — | — | — | — | |
| FPConvModality=Points+Normals2021.08 | 92.5 | — | — | — | — | — | |
| SpiderCNN (4-layer)Input=1024 points+normal, layers=42018.03 | 92.4 | — | — | — | — | — | |
| DGCNNNumber of points=1024, Initialization=Self-supervised Pre-Training on ShapeNet2019.01 | 92.4 | — | — | — | — | — | |
| SpiderCNNInput=1K points+normal2021.03 | 92.4 | — | — | — | — | — | |
| RS-CNNInput=1K points, voting=false2021.03 | 92.4 | — | — | — | — | — | |
| RS-CNN w/o vot.Modality=Points2021.08 | 92.4 | — | — | — | — | — | |
| PCNNInput=1K points2021.03 | 92.3 | — | — | — | — | — | |
| PointWebInput=1K points+normal2021.03 | 92.3 | — | — | — | — | — |