Shape Classification on ModelNet40
94.7AccuracyMOST
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| MOSTBackbone=ReCon [38], #P (M)=1.3, #F (G)=4.8, Training Strategy=with Pre-Training (Parameter-Efficient Fine-Tuning)2025.03 | 94.7 | — | — | — | — | — | — | — | |
| MOSTBackbone=Mamba3D [20], #P (M)=1.2, #F (G)=3.9, Training Strategy=with Pre-Training (Parameter-Efficient Fine-Tuning)2025.03 | 94.7 | — | — | — | — | — | — | — | |
| PointMLP2022.05 | 94.1 | — | 91.5 | — | — | — | — | — | |
| Point-MAE2022.05 | 94 | — | — | — | — | — | — | — | |
| Simpleview2022.05 | 93.9 | — | 91.8 | — | — | — | — | — | |
| GBNet2022.05 | 93.8 | — | 91 | — | — | — | — | — | |
| MVTN2022.05 | 93.8 | — | 92.2 | — | — | — | — | — | |
| CurveNet2022.05 | 93.8 | — | — | — | — | — | — | — | |
| PointBERT2022.05 | 93.8 | — | — | — | — | — | — | — | |
| PointER-MMbaseline=TAP2026.06 | 93.73 | — | — | — | — | — | — | — | |
| PointGLABackbone=GLA, Inference E-T (s)=42.04, Param.(M)=22.242026.06 | 93.73 | — | — | — | — | — | — | — | |
| PRA-Net2022.05 | 93.7 | — | 91.2 | — | — | — | — | — | |
| TAPreproduced=true2026.06 | 93.67 | — | — | — | — | — | — | — | |
| RSMixReference Venue=CVPR 20212022.10 | 93.5 | — | — | — | — | — | — | — | |
| PatchFormerReference Venue=CVPR 20222022.10 | 93.5 | — | — | — | — | — | — | — | |
| PointGLA (Sup.)Backbone=GLA, Inference E-T (s)=42.04, Param.(M)=22.242026.06 | 93.45 | — | — | — | — | — | — | — | |
| PointERbaseline=Point-MAE2026.06 | 93.41 | — | — | — | — | — | — | — | |
| PointERBackbone=RWKV, Inference E-T (s)=27.15, Param.(M)=25.72026.06 | 93.41 | — | — | — | — | — | — | — | |
| PointER (Sup.)Backbone=RWKV, Inference E-T (s)=27.15, Param.(M)=25.72026.06 | 93.34 | — | — | — | — | — | — | — | |
| PointStack2022.05 | 93.3 | — | 89.6 | — | — | — | — | — | |
| PSNetReference Venue=T-CSVT 20222022.10 | 93.3 | — | — | — | — | — | — | — | |
| Point-MAEreproduced=true2026.06 | 93.24 | — | — | — | — | — | — | — | |
| Point-MAE (Rep.)Backbone=Transformer, Inference E-T (s)=38.49, Param.(M)=22.12026.06 | 93.24 | — | — | — | — | — | — | — | |
| EQ-NetReference Venue=CVPR 20222022.10 | 93.2 | — | — | — | — | — | — | — | |
| DRNet2022.05 | 93.1 | — | — | — | — | — | — | — | |
| DRNetReference Venue=WACV 20212022.10 | 93.1 | — | — | — | — | — | — | — | |
| InterpCNNReference Venue=ICCV 20192022.10 | 93 | — | — | — | — | — | — | — | |
| MOST (Mamba3D)#P (M)=5.1, #F (G)=2.6, Training Strategy=Supervised Learning Only (Sparse Training)2025.03 | 92.91 | — | — | — | — | — | — | — | |
| KPConvInput Representation=Points2019.04 | 92.9 | — | — | — | — | — | — | — | |
| DGCNN2022.05 | 92.9 | — | 90.2 | — | — | — | — | — | |
| KPConv2022.05 | 92.9 | — | — | — | — | — | — | — | |
| KPConvReference Venue=ICCV 20192022.10 | 92.9 | — | — | — | — | — | — | — | |
| RS-CNNReference Venue=CVPR 20192022.10 | 92.9 | — | — | — | — | — | — | — | |
| PointANSLReference Venue=CVPR 20202022.10 | 92.9 | — | — | — | — | — | — | — | |
| MOST (Transformer)#P (M)=5.6, #F (G)=2.8, Training Strategy=Supervised Learning Only (Sparse Training)2025.03 | 92.83 | — | — | — | — | — | — | — | |
| Point Transformer2Reference Venue=Access 20212022.10 | 92.8 | — | — | — | — | — | — | — | |
| Point-TnT2022.05 | 92.6 | — | — | — | — | — | — | — | |
| A-CNNReference Venue=CVPR 20192022.10 | 92.6 | — | — | — | — | — | — | — | |
| P2SequenceReference Venue=AAAI 20192022.10 | 92.6 | — | — | — | — | — | — | — | |
| ConvPointInput Representation=Points, Number of Points=2048, k=162019.04 | 92.5 | — | 89.6 | — | — | — | — | — | |
| PointCNN2022.05 | 92.5 | — | 88.1 | — | — | — | — | — | |
| PointConvReference Venue=CVPR 20192022.10 | 92.5 | — | — | — | — | — | — | — | |
| SpiderCNN2022.05 | 92.4 | — | — | — | — | — | — | — | |
| SpiderCNNReference Venue=ECCV 20182022.10 | 92.4 | — | — | — | — | — | — | — | |
| PCNNReference Venue=TOG 20182022.10 | 92.3 | — | — | — | — | — | — | — | |
| DGCNNInput Representation=Points2019.04 | 92.2 | — | 90.2 | — | — | — | — | — | |
| PointCNNInput Representation=Points2019.04 | 92.2 | — | 88.1 | — | — | — | — | — | |
| SpecGCNReference Venue=ECCV 20182022.10 | 92.1 | — | — | — | — | — | — | — | |
| PointGridReference Venue=CVPR 20182022.10 | 92 | — | — | — | — | — | — | — | |
| PointNet++Input=pc, with normal information=true2017.06 | 91.9 | — | — | — | — | — | — | — | |
| PointNet++Reference Venue=NeurIPS 20172022.10 | 91.9 | — | — | — | — | — | — | — | |
| ConvPointInput Representation=Points, Number of Points=1024, k=162019.04 | 91.8 | — | 88.5 | — | — | — | — | — | |
| Kd-NetReference Venue=ICCV 20172022.10 | 91.8 | — | — | — | — | — | — | — | |
| Point TransformerReference Venue=ICCV 20212022.10 | 91.7 | — | — | — | — | — | — | — | |
| SO-NetReference Venue=CVPR 20182022.10 | 90.9 | — | — | — | — | — | — | — | |
| LCPFormer2022.10 | 90.7 | — | — | — | — | — | — | — | |
| PointNet++Input=pc2017.06 | 90.7 | — | — | — | — | — | — | — | |
| PointNet++Input Representation=Points2019.04 | 90.7 | — | — | — | — | — | — | — | |
| PointNet++2022.05 | 90.7 | — | — | — | — | — | — | — | |
| Point-PlaneNetReference Venue=DSP 20202022.10 | 90.5 | — | — | — | — | — | — | — | |
| Point2SequenceInput=1024 x 32018.11 | 90.4 | 92.6 | — | — | — | — | — | — | |
| DGCNNReference Venue=TOG 20192022.10 | 90.2 | — | — | — | — | — | — | — | |
| DGCNNInput=1024 x 32018.11 | 90.2 | 92.2 | — | — | — | — | — | — | |
| MVCNNInput=img2017.06 | 90.1 | — | — | — | — | — | — | — | |
| MVCNNInput Representation=Mesh or voxels2019.04 | 90.1 | — | — | — | — | — | — | — | |
| MVCNNReference Venue=ICCV 20152022.10 | 90.1 | — | — | — | — | — | — | — | |
| PCT (w/o LCP)Reference Venue=CVM 20212022.10 | 90 | — | — | — | — | — | — | — | |
| A-SCNReference Venue=CVPR 20182022.10 | 89.8 | — | — | — | — | — | — | — | |
| PointWebReference Venue=CVPR 20192022.10 | 89.4 | — | — | — | — | — | — | — | |
| SubvolumeInput=vox2017.06 | 89.2 | — | — | — | — | — | — | — | |
| PointNetInput=pc2017.06 | 89.2 | — | — | — | — | — | — | — | |
| SubvolumeInput Representation=Mesh or voxels2019.04 | 89.2 | — | — | — | — | — | — | — | |
| PointNetInput Representation=Points2019.04 | 89.2 | — | 86.2 | — | — | — | — | — | |
| PointNet2022.05 | 89.2 | — | 86 | — | — | — | — | — | |
| PointNet#P (M)=3.5, #F (G)=0.5, Training Strategy=Supervised Learning Only (Dense Training)2025.03 | 89.2 | — | — | — | — | — | — | — | |
| HOLA-PointBERT#Params=72.1M, FLOPs=84G, FPS=152, Training data=Ensembled no LVIS (829,460 triplets), Evaluation protocol=Zero-shot2026.05 | 89 | — | — | — | — | 97.6 | 99 | — | |
| HOLA-PointBERT#Params=72.1M, FLOPs=84G, FPS=152, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 88.8 | — | — | — | — | 97.6 | 98.8 | — | |
| Kd-NetInput=2^15 x 32018.11 | 88.5 | 91.8 | — | — | — | — | — | — | |
| HOLA-PointBERT#Params=32.3M, FLOPs=29G, FPS=202, Training data=Ensembled no LVIS (829,460 triplets), Evaluation protocol=Zero-shot2026.05 | 88.3 | — | — | — | — | 97.6 | 98.9 | — | |
| Uni3D †training shape source=Ensembled, best achieved=true2023.10 | 88.2 | — | — | — | — | 98.4 | 99.3 | — | |
| UNI3D-G#Params=1020.0M, FLOPs=1130G, FPS=34, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 88.2 | — | — | — | — | 98.4 | 99.3 | — | |
| PointCNNReference Venue=NeurIPS 20182022.10 | 88.1 | — | — | — | — | — | — | — | |
| HOLA-PointBERT#Params=32.3M, FLOPs=29G, FPS=202, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 88.1 | — | — | — | — | 97.4 | 98.6 | — | |
| ShapeContextNetInput=1024 x 32018.11 | 87.6 | 90 | — | — | — | — | — | — | |
| VIT-LENS-G#Params=1543.0M, FLOPs=806G, FPS=32, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 87.6 | — | — | — | — | 96.6 | 98.4 | — | |
| HOLA-PointBERT#Params=26.0M, FLOPs=7G, FPS=264, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 87.5 | — | — | — | — | 97.4 | 98.7 | — | |
| SO-NetInput=2048 x 32018.11 | 87.3 | 90.9 | — | — | — | — | — | — | |
| Uni3Dtraining shape source=Ensembled2023.10 | 87.3 | — | — | — | — | 98.1 | 99.2 | — | |
| RECON++-L#Params=658.9M, FLOPs=423G, FPS=34, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 87.3 | — | — | — | — | 95.4 | 96.1 | — | |
| PointNet (vanilla)Input=pc2017.06 | 87.2 | — | — | — | — | — | — | — | |
| Uni3Dtraining shape source=Ensembled (no LVIS)2023.10 | 86.8 | — | — | — | — | 97.3 | 98.4 | — | |
| VIT-LENS-G#Params=2000.0M, FLOPs=1050G, FPS=27, Training data=Ensembled no LVIS (829,460 triplets), Evaluation protocol=Zero-shot2026.05 | 86.8 | — | — | — | — | 96.8 | 97.8 | — | |
| UNI3D-G#Params=1020.0M, FLOPs=1130G, FPS=34, Training data=Ensembled no LVIS (829,460 triplets), Evaluation protocol=Zero-shot2026.05 | 86.8 | — | — | — | — | 97.3 | 98.4 | — | |
| MixCon3D-PointBERT#Params=30.9M, FLOPs=7G, FPS=153, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 86.8 | — | — | — | — | 96.9 | 98.3 | — | |
| RECON++-B#Params=201.5M, FLOPs=137G, FPS=90, Training data=Ensembled (875,665 triplets), Evaluation protocol=Zero-shot2026.05 | 86.5 | — | — | — | — | 94.7 | 95.8 | — | |
| TAMM-PointBERT#Params=35.4M, FLOPs=29G, FPS=85, Training data=Ensembled no LVIS (829,460 triplets), Evaluation protocol=Zero-shot2026.05 | 86.3 | — | — | — | — | 96.6 | 98.1 | — | |
| PointNetReference Venue=CVPR 20172022.10 | 86.2 | — | — | — | — | — | — | — | |
| PointNetInput=1024 x 32018.11 | 86.2 | 89.2 | — | — | — | — | — | — | |
| SubvolumeReference Venue=CVPR 20162022.10 | 86 | — | — | — | — | — | — | — | |
| OpenShape-PointBERTtraining shape source=Ensembled (no LVIS)2023.10 | 85.3 | — | — | — | — | 96.2 | 97.4 | — |