3D Shape Classification on ModelNet40 (test)
97.6AccuracyViewGCN
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ViewGCNData Type=20 Views2020.11 | 97.6 | 96.5 | — | — | — | — | — | |
| RotationNetInput=multi-views2020.06 | 97.3 | — | — | — | — | — | — | |
| VRN ensembleInput Feature=mesh structure2018.03 | 95.5 | — | — | — | — | — | — | |
| MVCNN-MultiRes2016.05 | 93.8 | — | — | — | — | — | — | |
| MVCNN-MultiResInput Feature=mesh structure2018.03 | 93.8 | — | — | — | — | — | — | |
| MVTNData Type=12 Views, rendering_setup=differentiable pipeline2020.11 | 93.8 | 92 | — | — | — | — | — | |
| PT2Input=8×1024, Latency=530.2ms2021.10 | 93.7 | — | — | — | — | — | — | |
| PT2Params=9.14M, FLOPs=17.1G, SDA=65.4%2021.10 | 93.7 | — | — | — | — | — | — | |
| PTransformerData Type=Points2020.11 | 93.7 | 90.6 | — | — | — | — | — | |
| RS-CNNinput=xyz, #points=1k2019.04 | 93.6 | — | — | — | — | — | — | |
| RS-CNNInput=point cloud2020.06 | 93.6 | — | — | — | — | — | — | |
| RSMix#Points=1k, Backbone=DGCNN, Evaluation Protocol=Single-View2021.02 | 93.5 | — | — | — | — | — | — | |
| PatchFormerInput=16×1024, Latency=34.3ms2021.10 | 93.5 | — | — | — | — | — | — | |
| PatchFormerParams=2.45M, FLOPs=1.62G, SDA=6.3%2021.10 | 93.5 | — | — | — | — | — | — | |
| MVTNData Type=20 Views, rendering_setup=differentiable pipeline2020.11 | 93.5 | 92.2 | — | — | — | — | — | |
| SO-Netinput=xyz, normal, #points=5k2019.04 | 93.4 | — | — | — | — | — | — | |
| SO-Netinput=point coordinates, normal, #points=5k2019.09 | 93.4 | — | — | — | — | — | — | |
| 3D2SeqViewsInput=multi-views2020.06 | 93.4 | — | — | — | — | — | — | |
| ViewGCNData Type=20 views, rendering_setup=differentiable pipeline2020.11 | 93.3 | 91.3 | — | — | — | — | — | |
| DensePointinput=point coordinates, #points=1k2019.09 | 93.2 | — | — | — | — | — | — | |
| PointASNLInput=16×1024, Latency=923.6ms2021.10 | 93.2 | — | — | — | — | — | — | |
| PCTInput=16×1024, Latency=92.4ms2021.10 | 93.2 | — | — | — | — | — | — | |
| PCTParams=2.88M, FLOPs=2.17G, SDA=24.6%2021.10 | 93.2 | — | — | — | — | — | — | |
| SGASData Type=Points2020.11 | 93.2 | — | — | — | — | — | — | |
| GVCNNInput=multi-views2020.06 | 93.1 | — | — | — | — | — | — | |
| PointASNLParams=3.98M, FLOPs=5.92G, SDA=39.8%2021.10 | 93.1 | — | — | — | — | — | — | |
| GVCNNData Type=12 Views2020.11 | 93.1 | 90.7 | — | — | — | — | — | |
| InterpCNNInput=1024 points2019.08 | 93 | — | — | — | — | — | — | |
| MID-FC(Finetune)Training Paradigm=Unsupervised Pre-training, Evaluation Protocol=Fine-tuned2020.08 | 93 | — | — | — | — | — | — | |
| ViewGCNData Type=12 views, rendering_setup=differentiable pipeline2020.11 | 93 | 90.7 | — | — | — | — | — | |
| KPConvKernel Type=rigid, Shape Normals=false2019.04 | 92.9 | — | — | — | — | — | — | |
| KPConvInput=point cloud2020.06 | 92.9 | — | — | — | — | — | — | |
| MID-FC(NoPre)Training Paradigm=Supervised, Evaluation Protocol=No Pre-training2020.08 | 92.9 | — | — | — | — | — | — | |
| DGCNNTraining Paradigm=Supervised2020.08 | 92.9 | — | — | — | — | — | — | |
| KPConvTraining Paradigm=Supervised2020.08 | 92.9 | — | — | — | — | — | — | |
| RS-CNNTraining Paradigm=Supervised2020.08 | 92.9 | — | — | — | — | — | — | |
| KPConvInput=16×6500, Latency=120.5ms2021.10 | 92.9 | — | — | — | — | — | — | |
| DGCNNInput=16×1024, Latency=85.8ms2021.10 | 92.9 | — | — | — | — | — | — | |
| DGCNNParams=1.81M, FLOPs=2.43G, SDA=57.2%2021.10 | 92.9 | — | — | — | — | — | — | |
| KPConvData Type=Points2020.11 | 92.9 | — | — | — | — | — | — | |
| DGCNN#Points=1k, Backbone=DGCNN, Evaluation Protocol=Single-View2021.02 | 92.8 | — | — | — | — | — | — | |
| PT1Input=16×1024, Latency=320.6ms2021.10 | 92.8 | — | — | — | — | — | — | |
| PT1Params=21.1M, FLOPs=5.05G, SDA=32.5%2021.10 | 92.8 | — | — | — | — | — | — | |
| KPConvKernel Type=deformable, Shape Normals=false2019.04 | 92.7 | — | — | — | — | — | — | |
| 3DCapsuleInput=1024 points2019.08 | 92.7 | — | — | — | — | — | — | |
| LDGCNNInput=16×10242021.10 | 92.7 | — | — | — | — | — | — | |
| Point2SeqInput=1024 points2019.08 | 92.6 | — | — | — | — | — | — | |
| PointConvInput=1024 points+normal2019.08 | 92.5 | — | — | — | — | — | — | |
| SpiderCNNinput=xyz, normal, #points=5k2019.04 | 92.4 | — | — | — | — | — | — | |
| SpiderCNNInput=1024 points+normal2019.08 | 92.4 | — | — | — | — | — | — | |
| SpiderCNNinput=point coordinates, normal, #points=5k2019.09 | 92.4 | — | — | — | — | — | — | |
| SpiderCNNInput=8×1024, Latency=82.6ms2021.10 | 92.4 | — | — | — | — | — | — | |
| PVCNNInput=16×1024, Latency=24.2ms2021.10 | 92.4 | — | — | — | — | — | — | |
| PCNN# points=10242018.03 | 92.3 | — | — | — | — | — | — | |
| PCNNinput=xyz, #points=1k2019.04 | 92.3 | — | — | — | — | — | — | |
| PCNN by ExtShape Normals=false2019.04 | 92.3 | — | — | — | — | — | — | |
| PCNNinput=point coordinates, #points=1k2019.09 | 92.3 | — | — | — | — | — | — | |
| MeshWalkerInput=mesh2020.06 | 92.3 | — | — | — | — | — | — | |
| PointWebInput=16×10242021.10 | 92.3 | — | — | — | — | — | — | |
| DGCNNinput=xyz, #points=1k2019.04 | 92.2 | — | — | — | — | — | — | |
| PointCNNShape Normals=false2019.04 | 92.2 | — | — | — | — | — | — | |
| DGCNNShape Normals=false2019.04 | 92.2 | — | — | — | — | — | — | |
| PointCNNInput=1024 points2019.08 | 92.2 | — | — | — | — | — | — | |
| DGCNNInput=1024 points2019.08 | 92.2 | — | — | — | — | — | — | |
| DGCNNinput=point coordinates, #points=1k2019.09 | 92.2 | — | — | — | — | — | — | |
| PointCNNinput=point coordinates, #points=1k2019.09 | 92.2 | — | — | — | — | — | — | |
| PointCNNTraining Paradigm=Supervised2020.08 | 92.2 | — | — | — | — | — | — | |
| PointCNNInput=16×1024, Latency=221.2ms2021.10 | 92.2 | — | — | — | — | — | — | |
| DGCNNData Type=Points2020.11 | 92.2 | 90.2 | — | — | — | — | — | |
| RSMix#Points=1k, Backbone=PointNet++, Evaluation Protocol=Multi-View2021.02 | 92.1 | — | — | — | — | — | — | |
| VRN singleInput Feature=mesh structure2018.03 | 92 | — | — | — | — | — | — | |
| MVCNNInput Feature=mesh structure2018.03 | 92 | — | — | — | — | — | — | |
| VIPGANModality=View, Learning paradigm=Unsupervised2019.07 | 91.98 | — | — | — | — | — | — | |
| pointnet++# points=5K, Input Feature=normals2018.03 | 91.9 | — | — | — | — | — | — | |
| MeshNetModality=mesh2018.11 | 91.9 | — | — | — | — | — | — | |
| PointNet++input=xyz, normal, #points=5k2019.04 | 91.9 | — | — | — | — | — | — | |
| PointNet++Input=5000 points+normal2019.08 | 91.9 | — | — | — | — | — | — | |
| PointNet++input=point coordinates, normal, #points=5k2019.09 | 91.9 | — | — | — | — | — | — | |
| MeshNetInput=mesh2020.06 | 91.9 | — | — | — | — | — | — | |
| PointNet++Input=16×1024, Latency=35.3ms2021.10 | 91.9 | — | — | — | — | — | — | |
| PointNet++(MSG)Params=1.74M, FLOPs=4.09G, SDA=47.6%2021.10 | 91.9 | — | — | — | — | — | — | |
| PointNet++Data Type=Points2020.11 | 91.9 | — | — | — | — | — | — | |
| Rotation Prediction + Context PredictionBackbone=DGCNN, Evaluation protocol=Linear SVM, Feature Combination=Concatenation2020.08 | 91.84 | — | — | — | — | — | — | |
| kd-network# points=32k2018.03 | 91.8 | — | — | — | — | — | — | |
| Kd-NetModality=point2018.11 | 91.8 | — | — | — | — | — | — | |
| Kd-Netinput=xyz, #points=32k, depth=152019.04 | 91.8 | — | — | — | — | — | — | |
| Spec-GCNinput=xyz, normal, #points=1k2019.04 | 91.8 | — | — | — | — | — | — | |
| Kd-NetworkInput=1024 points2019.08 | 91.8 | — | — | — | — | — | — | |
| Kd-Net(depth=15)input=point coordinates, #points=32k2019.09 | 91.8 | — | — | — | — | — | — | |
| Spec-GCNinput=point coordinates, normal, #points=1k2019.09 | 91.8 | — | — | — | — | — | — | |
| PointCNNData Type=Points2020.11 | 91.8 | 88.1 | — | — | — | — | — | |
| PointCNNinput=xyz, #points=1k2019.04 | 91.7 | — | — | — | — | — | — | |
| 3DmFV-NetShape Normals=false2019.04 | 91.6 | — | — | — | — | — | — | |
| SNGCInput=mesh2020.06 | 91.6 | — | — | — | — | — | — | |
| RSMix#Points=1k, Backbone=PointNet++, Evaluation Protocol=Single-View2021.02 | 91.6 | — | — | — | — | — | — | |
| 4l-spec-cpDomain=Local k-NN graph, Kernel=graph conv, Pooling=graph-cp2018.03 | 91.5 | — | — | 92.1 | — | — | — | |
| Spec-GCNinput=xyz, #points=1k2019.04 | 91.5 | — | — | — | — | — | — | |
| SpecGCNShape Normals=false2019.04 | 91.5 | — | — | — | — | — | — | |
| Spec-GCNinput=point coordinates, #points=1k2019.09 | 91.5 | — | — | — | — | — | — | |
| VRN SingleInput=voxels2019.08 | 91.3 | — | — | — | — | — | — |