3D Object Classification on ModelNet40 (test)
97.7AccuracyCARNet
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| CARNetMethod Category=View-based, Views=202021.10 | 97.7 | — | — | — | — | — | — | — | |
| CAR-NetViews=202022.11 | 97.7 | — | — | — | — | — | — | — | |
| R2-MLP-36Views=202022.11 | 97.7 | — | — | — | — | — | — | — | |
| View-GCNViews=202022.11 | 97.6 | — | — | — | — | — | — | — | |
| MVT-smallMethod Category=View-based, Views=202021.10 | 97.5 | — | — | — | — | — | — | — | |
| MVT-smallViews=202022.11 | 97.5 | — | — | — | — | — | — | — | |
| RotationNetMethod Category=View-based, Views=202021.10 | 97.4 | — | — | — | — | — | — | — | |
| RotationNetViews=202022.11 | 97.4 | — | — | — | — | — | — | — | |
| Ours (repr. + graph)Number of input points=20482021.03 | 96.9 | 94.1 | — | — | — | — | — | — | |
| Ours (repr. + graph)Number of input points=10242021.03 | 95.9 | 93.1 | — | — | — | — | — | — | |
| HybridNetpre-training=MAP, parameters_M=19.3, flops_G=4.4, voting=true2024.10 | 95.9 | — | — | — | — | — | — | — | |
| Mamba3dpre-training=MAP, parameters_M=16.9, flops_G=3.9, voting=true2024.10 | 95.6 | — | — | — | — | — | — | — | |
| Mamba3dpre-training=Point-MAE, parameters_M=16.9, flops_G=3.9, voting=true2024.10 | 95.4 | — | — | — | — | — | — | — | |
| HybridNetpre-training=MAP, parameters_M=19.3, flops_G=4.4, voting=false2024.10 | 95.4 | — | — | — | — | — | — | — | |
| CARNetMethod Category=View-based, Views=122021.10 | 95.2 | — | — | — | — | — | — | — | |
| CAR-NetViews=122022.11 | 95.2 | — | — | — | — | — | — | — | |
| Ours (repr.)Number of input points=10242021.03 | 95.1 | 92 | — | — | — | — | — | — | |
| Mamba3dpre-training=MAP, parameters_M=16.9, flops_G=3.9, voting=false2024.10 | 95.1 | — | — | — | — | — | — | — | |
| R2-MLP-36Views=122022.11 | 95 | — | — | — | — | — | — | — | |
| R2-MLP-36Views=62022.11 | 94.7 | — | — | — | — | — | — | — | |
| PointMLP + ULIPvoting technique=true2022.12 | 94.7 | 92.4 | — | — | — | — | — | — | |
| Mamba3dpre-training=Point-MAE, parameters_M=16.9, flops_G=3.9, voting=false2024.10 | 94.7 | — | — | — | — | — | — | — | |
| PointMLPLevel of supervision=Supervised, No. of class obtained=402022.10 | 94.5 | — | — | — | — | — | — | — | |
| PointMLPvoting technique=true2022.12 | 94.5 | 91.4 | — | — | — | — | — | — | |
| PointMLPpre-training=None, parameters_M=12.6, flops_G=31.42024.10 | 94.5 | — | — | — | — | — | — | — | |
| MVT-smallMethod Category=View-based, Views=122021.10 | 94.4 | — | — | — | — | — | — | — | |
| MVT-smallViews=122022.11 | 94.4 | — | — | — | — | — | — | — | |
| Point-MAEpre-training=IDPT, parameters_M=22.1+1.7, flops_G=4.82024.10 | 94.4 | — | — | — | — | — | — | — | |
| Mamba3dpre-training=Point-BERT, parameters_M=16.9, flops_G=3.9, voting=false2024.10 | 94.4 | — | — | — | — | — | — | — | |
| Relation NetworkMethod Category=View-based, Views=122021.10 | 94.3 | — | — | — | — | — | — | — | |
| Relation NetworkViews=122022.11 | 94.3 | — | — | — | — | — | — | — | |
| PointMLP + ULIPvoting technique=false2022.12 | 94.3 | 92.3 | — | — | — | — | — | — | |
| HybridNetpre-training=None, parameters_M=19.3, flops_G=4.4, voting=true2024.10 | 94.3 | — | — | — | — | — | — | — | |
| MLVCNNViews=122022.11 | 94.2 | — | — | — | — | — | — | — | |
| CurveNet2022.12 | 94.2 | — | — | — | — | — | — | — | |
| RPNet2022.12 | 94.1 | — | — | — | — | — | — | — | |
| PointBERT + ULIP2022.12 | 94.1 | — | — | — | — | — | — | — | |
| PointMLPvoting technique=false2022.12 | 94.1 | 91.3 | — | — | — | — | — | — | |
| PointMLPParams. (M)=13.2, Input Points=1024, Voting Strategy=false2024.03 | 94.1 | 91.3 | — | — | — | — | — | — | |
| Mamba3dpre-training=None, parameters_M=16.9, flops_G=3.9, voting=true2024.10 | 94.1 | — | — | — | — | — | — | — | |
| PointNeXtpre-training=None, parameters_M=1.4, flops_G=3.62024.10 | 94 | — | — | — | — | — | — | — | |
| P2P-HorNetpre-training=checkmark, flops_G=34.62024.10 | 94 | — | — | — | — | — | — | — | |
| DeLApre-training=None, parameters_M=5.3, flops_G=1.52024.10 | 94 | — | — | — | — | — | — | — | |
| Point-M2AEpre-training=Point-M2AE, parameters_M=15.3, flops_G=3.62024.10 | 94 | — | — | — | — | — | — | — | |
| SimpleViewpre-training=None2024.10 | 93.9 | — | — | — | — | — | — | — | |
| GDANetInput=1K points2020.12 | 93.8 | — | — | — | — | — | — | — | |
| CurveNetInput modality=raw point clouds, Voting strategy=false2023.02 | 93.8 | — | — | — | — | — | — | — | |
| DeltaConvInput modality=raw point clouds, Voting strategy=false2023.02 | 93.8 | — | — | — | — | — | — | — | |
| APES (global-based)Input modality=raw point clouds, Voting strategy=false2023.02 | 93.8 | — | — | — | — | — | — | — | |
| GDANet2022.12 | 93.8 | — | — | — | — | — | — | — | |
| GBNet2022.12 | 93.8 | 91 | — | — | — | — | — | — | |
| MTVN2022.12 | 93.8 | 92 | — | — | — | — | — | — | |
| CurveNetParams. (M)=2, Input Points=1024, Voting Strategy=false2024.03 | 93.8 | — | — | — | — | — | — | — | |
| Point-MAEParams. (M)=22.1, Input Points=1024, Voting Strategy=false2024.03 | 93.8 | — | — | — | — | — | — | — | |
| GBNetpre-training=None, parameters_M=8.82024.10 | 93.8 | — | — | — | — | — | — | — | |
| MVTNpre-training=None, parameters_M=11.2, flops_G=43.72024.10 | 93.8 | — | — | — | — | — | — | — | |
| MaskPointpre-training=MaskPoint, parameters_M=22.1, flops_G=4.82024.10 | 93.8 | — | — | — | — | — | — | — | |
| Point-MAEpre-training=Point-MAE, parameters_M=22.1, flops_G=4.82024.10 | 93.8 | — | — | — | — | — | — | — | |
| PT²Input modality=raw point clouds, Voting strategy=false2023.02 | 93.7 | — | — | — | — | — | — | — | |
| PRA-NetInput modality=raw point clouds, Voting strategy=false2023.02 | 93.7 | — | — | — | — | — | — | — | |
| PRA-Nepre-training=None, flops_G=2.32024.10 | 93.7 | — | — | — | — | — | — | — | |
| RS-CNNInput Data=pnt, Number of Points=1k2020.03 | 93.6 | — | — | — | — | — | — | — | |
| RS-CNNInput=point cloud, Main operator=Local feature, Input size=1024 x 32021.02 | 93.6 | — | — | — | — | — | — | — | |
| RS-CNNInput=1K points2020.12 | 93.6 | — | — | — | — | — | — | — | |
| PAConvInput modality=raw point clouds, Voting strategy=false2023.02 | 93.6 | — | — | — | — | — | — | — | |
| RS-CNNvoting technique=true2022.12 | 93.6 | — | — | — | — | — | — | — | |
| RECON++Evaluation Protocol=Linear SVM, Hierarchical=false, Voting=Without voting2024.02 | 93.6 | — | — | — | — | — | — | — | |
| DGCNNNumber of input points=20482021.03 | 93.5 | 90.7 | — | — | — | — | — | — | |
| APES (local-based)Input modality=raw point clouds, Voting strategy=false2023.02 | 93.5 | — | — | — | — | — | — | — | |
| PointConTpre-training=None2024.10 | 93.5 | — | — | — | — | — | — | — | |
| HybridNetpre-training=None, parameters_M=19.3, flops_G=4.4, voting=false2024.10 | 93.5 | — | — | — | — | — | — | — | |
| SO-NetInput Data=pnt, nor, Number of Points=5k2020.03 | 93.4 | — | — | — | — | — | — | — | |
| Geo-CNNInput=1K points2020.12 | 93.4 | — | — | — | — | — | — | — | |
| 3D2SeqViewsMethod Category=View-based, Views=122021.10 | 93.4 | — | — | — | — | — | — | — | |
| SeqViews2SeqLabelsMethod Category=View-based, Views=122021.10 | 93.4 | — | — | — | — | — | — | — | |
| 3D2SeqViewsViews=122022.11 | 93.4 | — | — | — | — | — | — | — | |
| SeqViews2SeqLabelsViews=122022.11 | 93.4 | — | — | — | — | — | — | — | |
| PointNet++(ssg) + ULIP2022.12 | 93.4 | 91.2 | — | — | — | — | — | — | |
| I2P-MAEEvaluation Protocol=Linear SVM, Hierarchical=true, Voting=Without voting2024.02 | 93.4 | — | — | — | — | — | — | — | |
| RECONEvaluation Protocol=Linear SVM, Hierarchical=false, Voting=Without voting2024.02 | 93.4 | — | — | — | — | — | — | — | |
| PCMpre-training=None, parameters_M=34.2, flops_G=45.02024.10 | 93.4 | — | — | — | — | — | — | — | |
| Mamba3dpre-training=None, parameters_M=16.9, flops_G=3.9, voting=false2024.10 | 93.4 | — | — | — | — | — | — | — | |
| Point-BERTpre-training=IDPT, parameters_M=22.1+1.7, flops_G=4.82024.10 | 93.4 | — | — | — | — | — | — | — | |
| GSNNumber of input points=20482021.03 | 93.3 | — | — | — | — | — | — | — | |
| PointASNLInput Data=pnt, nor, Number of Points=1k2020.03 | 93.2 | — | — | — | — | — | — | — | |
| DensePointInput=1K points2020.12 | 93.2 | — | — | — | — | — | — | — | |
| DensePoint2021.03 | 93.2 | — | — | — | — | — | — | — | |
| PointASNL2021.03 | 93.2 | — | — | — | — | — | — | — | |
| PointASNLInput modality=raw point clouds, Voting strategy=false2023.02 | 93.2 | — | — | — | — | — | — | — | |
| PCTInput modality=raw point clouds, Voting strategy=false2023.02 | 93.2 | — | — | — | — | — | — | — | |
| PCT2022.12 | 93.2 | — | — | — | — | — | — | — | |
| PointBERT2022.12 | 93.2 | — | — | — | — | — | — | — | |
| PCTParams. (M)=2.9, Input Points=1024, Voting Strategy=false2024.03 | 93.2 | — | — | — | — | — | — | — | |
| Point-BERTParams. (M)=22.1, Input Points=1024, Voting Strategy=false2024.03 | 93.2 | — | — | — | — | — | — | — | |
| PCTpre-training=None, parameters_M=2.9, flops_G=2.32024.10 | 93.2 | — | — | — | — | — | — | — | |
| (AF)2-S3Net (AF2M)Input=voxels, Main operator=Sparse 3D Operation, Input size=1024 x 32021.02 | 93.16 | — | — | — | — | — | — | — | |
| DPC# Points=4k, Features=with normals2019.07 | 93.1 | 91.4 | — | — | — | — | — | — | |
| GVCNNMethod Category=View-based, Views=122021.10 | 93.1 | — | — | — | — | — | — | — | |
| GVCNNViews=122022.11 | 93.1 | — | — | — | — | — | — | — | |
| ACTEvaluation Protocol=Linear SVM, Hierarchical=false, Voting=Without voting2024.02 | 93.1 | — | — | — | — | — | — | — |