3D Shape recognition on ModelNet10 (test)
99.3AccuracyMVT-small
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MVT-smallMethod Category=View-based, Views=202021.10 | 99.3 | — | — | |
| CARNetMethod Category=View-based, Views=202021.10 | 99 | — | — | |
| RotationNetMethod Category=View-based, Views=202021.10 | 98.5 | — | — | |
| VRN ensembleInput Feature=mesh structure2018.03 | 97.1 | — | — | |
| DensePointinput=point coordinates, #points=1k2019.09 | 96.6 | — | — | |
| PointMACESymmetry=O(3)2023.05 | 96.1 | — | — | |
| RSMix#Points=1k, Backbone=DGCNN, Evaluation Protocol=Single-View2021.02 | 95.9 | — | — | |
| CARNetMethod Category=View-based, Views=122021.10 | 95.8 | — | — | |
| SO-Netinput=point coordinates, normal, #points=5k2019.09 | 95.7 | — | — | |
| LGANModality=Points, Learning paradigm=Unsupervised2019.07 | 95.3 | — | — | |
| NSamplerModality=Points, Learning paradigm=Unsupervised2019.07 | 95.3 | — | — | |
| Relation NetworkMethod Category=View-based, Views=122021.10 | 95.3 | — | — | |
| MVT-smallMethod Category=View-based, Views=122021.10 | 95.3 | — | — | |
| 3DmFV-NetMethod Category=Point-based2021.10 | 95.2 | — | — | |
| PointNet++2023.05 | 95 | — | — | |
| PCNN# points=10242018.03 | 94.9 | — | — | |
| PCNNinput=point coordinates, #points=1k2019.09 | 94.9 | — | — | |
| MAP-VAEModality=Points, Learning paradigm=Unsupervised2019.07 | 94.82 | — | — | |
| DGCNN#Points=1k, Backbone=DGCNN, Evaluation Protocol=Single-View2021.02 | 94.8 | — | — | |
| SeqViews2SeqLabelsMethod Category=View-based, Views=122021.10 | 94.8 | — | — | |
| 3D2SeqViewsMethod Category=View-based, Views=122021.10 | 94.7 | — | — | |
| FNetModality=Points, Learning paradigm=Unsupervised2019.07 | 94.4 | — | — | |
| KCNetinput=point coordinates, #points=1k2019.09 | 94.4 | — | — | |
| RSMix#Points=1k, Backbone=PointNet++, Evaluation Protocol=Multi-View2021.02 | 94.4 | — | — | |
| RSMix#Points=1k, Backbone=PointNet++, Evaluation Protocol=Single-View2021.02 | 94.3 | — | — | |
| PointNet2023.05 | 94.2 | — | — | |
| SO-Netinput=point coordinates, #points=2k2019.09 | 94.1 | — | — | |
| VIPGANModality=View, Learning paradigm=Unsupervised2019.07 | 94.05 | — | — | |
| kd-network# points=32k2018.03 | 94 | — | — | |
| Kd-Net(depth=15)input=point coordinates, #points=32k2019.09 | 94 | — | — | |
| RotationNetMethod Category=View-based, Views=122021.10 | 94 | — | — | |
| ORIONSupervision=Supervised2017.05 | 93.8 | — | — | |
| PointNet++#Points=1k, Backbone=PointNet++, Evaluation Protocol=Multi-View2021.02 | 93.5 | — | — | |
| kd-network# points=10242018.03 | 93.3 | — | — | |
| Kd-Net(depth=10)input=point coordinates, #points=1k2019.09 | 93.3 | — | — | |
| PointNet++#Points=1k, Backbone=PointNet++, Evaluation Protocol=Single-View2021.02 | 93.3 | — | — | |
| FusionNetInput Feature=mesh structure2018.03 | 93.1 | — | — | |
| PointNet#Points=1k, Backbone=PointNet, Evaluation Protocol=Single-View2021.02 | 93.1 | — | — | |
| RSMix#Points=1k, Backbone=PointNet, Evaluation Protocol=Single-View2021.02 | 93.1 | — | — | |
| DGCNNLayers (L)=42021.08 | 92.8 | — | — | |
| RSMix#Points=1k, Backbone=PointNet, Evaluation Protocol=Multi-View2021.02 | 92.6 | — | — | |
| PointNet#Points=1k, Backbone=PointNet, Evaluation Protocol=Multi-View2021.02 | 92.5 | — | — | |
| PDE-GCNDLayers (L)=42021.08 | 92.2 | — | — | |
| LGAN (MN)Modality=Points, Learning paradigm=Unsupervised, Training dataset version=ModelNet context (reproduced)2019.07 | 92.18 | — | — | |
| VoxNetSupervision=Supervised2017.05 | 92 | — | — | |
| VoxNetMethod Category=Volume-based2021.10 | 92 | — | — | |
| FNet (MN)Modality=Points, Learning paradigm=Unsupervised, Training dataset version=ModelNet context (reproduced)2019.07 | 91.85 | — | — | |
| 3D-GANSupervision=Unsupervised2017.05 | 91 | — | — | |
| VSLSupervision=Unsupervised2017.05 | 91 | — | — | |
| 3DGANModality=Voxel, Learning paradigm=Unsupervised2019.07 | 91 | — | — | |
| VSLModality=Voxel, Learning paradigm=Unsupervised2019.07 | 91 | — | — | |
| OctNetInput Feature=voxels2018.03 | 90.9 | — | — | |
| ECC# points=10002018.03 | 90.8 | — | — | |
| ECCinput=point coordinates, #points=1k2019.09 | 90.8 | — | — | |
| Geometry ImageSupervision=Supervised2017.05 | 88.4 | — | — | |
| DeepPanoSupervision=Supervised2017.05 | 85.5 | — | — | |
| 3D ShapeNetsSupervision=Supervised2017.05 | 83.5 | — | — | |
| 3DShapeNetsMethod Category=Volume-based2021.10 | 83.5 | — | — | |
| VConv-DAESupervision=Unsupervised2017.05 | 80.5 | — | — | |
| Vconv-DAEModality=Voxel, Learning paradigm=Unsupervised2019.07 | 80.5 | — | — | |
| LFDSupervision=Unsupervised2017.05 | 79.9 | — | — | |
| SPHSupervision=Unsupervised2017.05 | 79.8 | — | — | |
| T-L NetworkSupervision=Unsupervised2017.05 | 74.4 | — | — | |
| GCNIILayers (L)=42021.08 | 65.4 | — | — | |
| 3DShapeNets2017.04 | — | 83.5 | — | |
| ECC2017.04 | — | 90 | 90.8 | |
| FusionNet2017.04 | — | — | 93.1 | |
| Kd-Netdepth=102017.04 | — | 92.8 | 93.3 | |
| Kd-Netdepth=152017.04 | — | 93.5 | 94 | |
| MVCNNEnsemble vs Single=Single2017.04 | — | — | 90.1 | |
| OctNet2017.04 | — | 90.1 | 90.9 | |
| VRNEnsemble vs Single=Single2017.04 | — | — | 93.6 | |
| VRNEnsemble vs Single=Ensemble2017.04 | — | — | 97.1 |