3D Object Detection on SUN RGB-D (test)
67.4mAP@0.25DeMF (FCAF3D based)
Evaluation Results
| Method | Links | ||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DeMF (FCAF3D based)Point Backbone=HDResNet34, Input=PC+RGB2022.07 | 67.4 | — | — | — | — | — | — | — | — | — | — | — | 51.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EPNet++Point Backbone=PointNet++, Input=PC+RGB2022.07 | 65.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TokenFusion (L6, O256; S)Backbone=PointNet++, Inputs=Points+RGB2022.04 | 64.9 | — | — | — | — | — | — | — | — | — | — | — | 48.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EPNetPoint Backbone=PointNet++, Input=PC+RGB2022.07 | 64.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TokenFusion (L6, O256; Ti)Backbone=PointNet++, Inputs=Points+RGB2022.04 | 64.5 | — | — | — | — | — | — | — | — | — | — | — | 47.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ImVoteNetPoint Backbone=PointNet++, Input=PC+RGB2022.07 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | 43.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FCAF3DPoint Backbone=HDResNet34, Input=PC2022.07 | 64.2 | — | — | — | — | — | — | — | — | — | — | — | 48.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| imVoteNetbackbone=PointNet++, inputs=point+RGB2021.04 | 63.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| imVoteNetBackbone=PointNet++, Inputs=Points+RGB2022.04 | 63.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LCPFormerReference=N/A, Backbone=w/ LCP2022.10 | 63.2 | — | — | — | — | — | — | — | — | — | — | — | 46.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Group-Free-3Dbackbone=PointNet++, inputs=point, decoder depth (L)=6, number of object candidates (O)=2562021.04 | 63 | — | — | — | — | — | — | — | — | — | — | — | 45.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GF (L6, O256)Backbone=PointNet++, Inputs=Points2022.04 | 63 | — | — | — | — | — | — | — | — | — | — | — | 45.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Group-FreePoint Backbone=PointNet++, Input=PC2022.07 | 63 | — | — | — | — | — | — | — | — | — | — | — | 45.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Group-FreeReference=ICCV 2021, Backbone=PointNet++2022.10 | 62.8 | — | — | — | — | — | — | — | — | — | — | — | 44.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LCPFormerReference=N/A, Backbone=w/o LCP2022.10 | 62.7 | — | — | — | — | — | — | — | — | — | — | — | 43.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GF (L6, O256)Backbone=PointNet++, Inputs=Points+RGB2022.04 | 62.1 | — | — | — | — | — | — | — | — | — | — | — | 42.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RDFC-GANBase Detector=H3DNet2023.06 | 61.75 | — | — | — | — | — | — | — | — | — | — | — | 40.63 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HGNetbackbone=GU-net, inputs=point2021.04 | 61.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| H3DNetBackbone=4xPointNet++, Inputs=Points2022.04 | 61.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HGNetReference=CVPR 2020, Backbone=GU-Net2022.10 | 61.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LCPFormer+ VoteNetReference=N/A, Backbone=w/ LCP2022.10 | 61.4 | — | — | — | — | — | — | — | — | — | — | — | 39.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BRNetPoint Backbone=PointNet++, Input=PC2022.07 | 61.1 | — | — | — | — | — | — | — | — | — | — | — | 43.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Pointformer + VoteNetReference=CVPR 2021, Backbone=Pointformer2022.10 | 61.1 | — | — | — | — | — | — | — | — | — | — | — | 36.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RDF-GANBase Detector=H3DNet2023.06 | 61.03 | — | — | — | — | — | — | — | — | — | — | — | 39.71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RDFC-GANBase Detector=VoteNet2023.06 | 61.02 | — | — | — | — | — | — | — | — | — | — | — | 37.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RDF-GANBase Detector=VoteNet2023.06 | 60.64 | — | — | — | — | — | — | — | — | — | — | — | 37.28 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GTBase Detector=H3DNet2023.06 | 60.51 | — | — | — | — | — | — | — | — | — | — | — | 39.22 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| EQ-NetReference=CVPR 2022, Backbone=EQ-PointNet++2022.10 | 60.5 | — | — | — | — | — | — | — | — | — | — | — | 38.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DeepLidarBase Detector=H3DNet2023.06 | 60.35 | — | — | — | — | — | — | — | — | — | — | — | 39.16 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| H3DNetBase Detector=H3DNet2023.06 | 60.11 | — | — | — | — | — | — | — | — | — | — | — | 39.04 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| H3DNetbackbone=4xPointNet++, inputs=point2021.04 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | 39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HGNetBackbone=GU-net, Inputs=Points2022.04 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | 39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| H3DNetPoint Backbone=4xPointNet++, Input=PC2022.07 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | 39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| H3DNetReference=ECCV 2020, Backbone=4x PointNet++2022.10 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | 39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| T3DReference=TCSVT 2021, Backbone=PointNet++2022.10 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MLCVNetbackbone=PointNet++, inputs=point2021.04 | 59.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MLCVNetBackbone=PointNet++, Inputs=Points2022.04 | 59.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MLCVNetPoint Backbone=PointNet++, Input=PC2022.07 | 59.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MLCVNetReference=CVPR 2020, Backbone=PointNet++2022.10 | 59.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DeepLidarBase Detector=VoteNet2023.06 | 59.73 | — | — | — | — | — | — | — | — | — | — | — | 35.49 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GTBase Detector=VoteNet2023.06 | 59.44 | — | — | — | — | — | — | — | — | — | — | — | 36.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoteNetbackbone=PointNet++, inputs=point2021.04 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 35.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoteNetBackbone=PointNet++, Inputs=Points2022.04 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 35.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 3DETRPoint Backbone=Transformers, Input=PC2022.07 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 32.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 3DETR-mReference=ICCV 20212022.10 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 32.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoteNetReference=ICCV 2019, Backbone=PointNet++2022.10 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 35.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoteNetBase Detector=VoteNet2023.06 | 59.07 | — | — | — | — | — | — | — | — | — | — | — | 35.77 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TupleInfoNCEBackbone=VoteNet2021.07 | 58 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoteNetBackbone=PointNet++, Inputs=Points+RGB2022.04 | 58 | — | — | — | — | — | — | — | — | — | — | — | 34.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 3DETRReference=ICCV 20212022.10 | 58 | — | — | — | — | — | — | — | — | — | — | — | 30.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| InfoNCEBackbone=VoteNet2021.07 | 56.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MISABackbone=VoteNet2021.07 | 56.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CMCBackbone=VoteNet2021.07 | 56.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Train from scratchBackbone=VoteNet2021.07 | 56.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F-PointNetPoint Backbone=PointNet, Input=PC+RGB2022.07 | 54 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| COGInput=PC+RGB2022.07 | 47.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NLSPNBase Detector=VoteNet2023.06 | 47.43 | — | — | — | — | — | — | — | — | — | — | — | 26.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PointFusionPoint Backbone=PointNet, Input=PC+RGB2022.07 | 45.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NLSPNBase Detector=H3DNet2023.06 | 27.1 | — | — | — | — | — | — | — | — | — | — | — | 9.77 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Uni3DETRlatency=0.52 s, params=26.71 M, FLOPS=458.74 G2023.10 | 0.67 | — | — | — | — | — | — | — | — | — | — | 0.6435 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UVTRlatency=0.51 s, params=26.12 M, FLOPS=451.12 G2023.10 | 0.502 | — | — | — | — | — | — | — | — | — | — | 0.5555 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PillarNetlatency=0.31 s, params=12.55 M, FLOPS=100.10 G2023.10 | 0.282 | — | — | — | — | — | — | — | — | — | — | 0.44 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CenterPointlatency=0.32 s, params=9.17 M, FLOPS=121.10 G2023.10 | 0.189 | — | — | — | — | — | — | — | — | — | — | 0.3775 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxelNeXtlatency=0.29 s, params=7.12 M, FLOPS=42.57 G2023.10 | 0.181 | — | — | — | — | — | — | — | — | — | — | 0.393 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 3DGPInput=Point Cloud2021.06 | — | 5.62 | 2.31 | 3.24 | 1.23 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 3DGP2021.03 | — | 5.62 | 2.31 | 3.24 | 1.23 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 4DContrast + VoteNetInput=Geo2021.12 | — | — | — | — | — | — | — | — | — | — | — | — | 38.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Choi et al.2018.10 | — | 5.62 | 2.31 | 3.24 | 1.23 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CooPInput=Point Cloud2021.06 | — | 63.58 | 17.12 | 41.22 | 26.21 | 9.55 | 58.55 | 10.19 | 5.34 | 3.01 | 1.75 | 23.65 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CooPtraining_labels=NYU-37 object labels, source=better results from [34]2021.03 | — | 57.71 | 15.21 | 36.67 | 31.16 | 19.9 | — | — | 15.95 | — | 3.28 | 21.77 | — | — | — | — | — | — | — | — | — | — | — | — | — | 15.98 | 11.36 | 10.47 | |
| CSC + VoteNetInput=Geo2021.12 | — | — | — | — | — | — | — | — | — | — | — | — | 36.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Cube R-CNNTrained on=SUN RGB-D, Backbone=ResNet-342022.07 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 36.2 | 34.7 | — | — | — | |
| Cube R-CNNTrained on=OMNI3DIN, Backbone=ResNet-342022.07 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 37.8 | 35.4 | — | — | — | |
| Cube RCNNTrained on=SUN RGB-D2023.10 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 34.7 | — | — | — | |
| Cube RCNNTrained on=Omni3D2023.10 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 35.4 | — | — | — | |
| DeepContrast + VoteNetInput=Geo, Backbone=PointNet++2021.12 | — | — | — | — | — | — | — | — | — | — | — | — | 35.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F-PointNetRGB=true, PC=true2021.06 | — | — | — | — | — | — | — | — | — | — | — | 54 | — | 43.3 | 81.1 | 33.3 | 64.2 | 24.7 | 32 | 58.1 | 61.1 | 51.1 | 90.9 | — | — | — | — | — | |
| H3DNetRGB=false, PC=true2021.06 | — | — | — | — | — | — | — | — | — | — | — | 60.1 | — | 73.8 | 85.6 | 31 | 76.7 | 29.6 | 33.4 | 65.5 | 66.5 | 50.8 | 88.2 | — | — | — | — | — | |
| HoPRInput=Point Cloud2021.06 | — | 58.29 | 13.56 | 28.37 | 12.12 | 4.79 | 16.5 | 0.63 | 2.18 | 1.29 | 2.41 | 14.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HoPR2021.03 | — | 58.29 | 13.56 | 28.37 | 12.12 | 4.79 | — | — | 2.18 | — | 2.41 | 14.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | 13.71 | 8.8 | 0.48 | |
| Huang et al.2018.10 | — | 58.29 | 13.56 | 28.37 | 12.12 | 4.79 | 16.5 | 0.63 | 2.18 | 1.29 | 2.41 | 14.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ImVoteNetRGB=true, PC=true2021.06 | — | — | — | — | — | — | — | — | — | — | — | 63.4 | — | 75.9 | 87.6 | 41.3 | 76.7 | 28.7 | 41.4 | 69.9 | 70.7 | 51.1 | 90.5 | — | — | — | — | — | |
| ImVoxelNetInput=RGB Only2021.06 | — | 77.87 | 65.94 | 63.89 | 51.17 | 31.91 | 84.53 | 33.35 | 39.91 | 21.65 | 17.19 | 48.74 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ImVoxelNetRGB=true, PC=false2021.06 | — | — | — | — | — | — | — | — | — | — | — | 40.7 | — | 71.7 | 69.6 | 5.7 | 53.7 | 21.9 | 21.2 | 34.6 | 51.5 | 39.1 | 76.8 | — | — | — | — | — | |
| ImVoxelNetTrained on=SUN RGB-D2022.07 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 30.6 | — | — | — | |
| Oursmodel_components=LIEN + SGCN2021.03 | — | 89.32 | 35.14 | 69.1 | 57.37 | 49.03 | — | — | 33.81 | — | 11.9 | 45.21 | — | — | — | — | — | — | — | — | — | — | — | — | — | 29.27 | 41.34 | 33.93 | |
| Ours (cooperative)training_mode=cooperative2018.10 | — | 63.58 | 17.12 | 41.22 | 26.21 | 9.55 | 58.55 | 10.19 | 5.34 | 3.01 | 1.75 | 23.65 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Ours (individual)training_mode=individual2018.10 | — | 53.08 | 7.7 | 27.04 | 22.8 | 5.51 | 28.07 | 0.54 | 5.08 | 2.58 | 0.01 | 15.24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PerspectiveNetInput=Point Cloud2021.06 | — | 79.69 | 40.42 | 62.35 | 44.12 | 20.19 | 81.22 | 22.42 | 41.35 | 8.29 | 13.14 | 39.09 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PointContrast + VoteNetInput=Geo2021.12 | — | — | — | — | — | — | — | — | — | — | — | — | 34.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RandomRooms + VoteNetInput=Geo, Backbone=PointNet++2021.12 | — | — | — | — | — | — | — | — | — | — | — | — | 35.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Scratch + VoteNetInput=Geo2021.12 | — | — | — | — | — | — | — | — | — | — | — | — | 31.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Total3DTrained on=SUN RGB-D, Backbone=ResNet-34, 2D Input=Oracle detections2022.07 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 23.3 | — | — | — | — | |
| Total3D2021.03 | — | 60.65 | 17.55 | 44.9 | 36.48 | 27.93 | — | — | 18.5 | — | 5.04 | 26.38 | — | — | — | — | — | — | — | — | — | — | — | — | — | 21.19 | 17.01 | 14.51 | |
| Uni3DETRTrained on=SUN RGB-D2023.10 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 64.3 | — | — | — | |
| Uni3DETRTrained on=ScanNet2023.10 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.9 | — | — | — | |
| VoteNetRGB=false, PC=true2021.06 | — | — | — | — | — | — | — | — | — | — | — | 57.7 | — | 74.4 | 83 | 28.8 | 75.3 | 22 | 29.8 | 62.2 | 64 | 47.3 | 90.1 | — | — | — | — | — | |
| VoteNetInput=Geo + Height2021.12 | — | — | — | — | — | — | — | — | — | — | — | — | 32.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |