3D Object Detection on SUN RGB-D v1 (val)
68.9mAP@0.25Ours (GroupFree3D)
Evaluation results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Ours (GroupFree3D)Pre-trained=Yes, Backbone=GroupFree3D2024.10 | 68.9 | — | — | — | — | — | — | — | — | — | — | — | 52.1 | |
| Bridge3D (GroupFree3D)Pre-trained=Yes, Backbone=GroupFree3D2024.10 | 67.9 | — | — | — | — | — | — | — | — | — | — | — | 48.5 | |
| IAELearning Paradigm=Self-supervised, Backbone=CAGroup3D2022.01 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | 51.3 | |
| IAEBase Architecture=CAGroup3D, Pre-training Setting=Self-supervised2025.12 | 67.6 | — | — | — | — | — | — | — | — | — | — | — | 51.3 | |
| Uni3DETRModality=Point Clouds2023.10 | 67 | — | — | — | — | — | — | — | — | — | — | — | 50.3 | |
| CAGroup3DLearning Paradigm=Supervised2022.01 | 66.8 | — | — | — | — | — | — | — | — | — | — | — | 50.2 | |
| CAGroup3DPre-training Setting=Supervised2025.12 | 66.8 | — | — | — | — | — | — | — | — | — | — | — | 50.2 | |
| PIMAE (GroupFree3D)Pre-trained=Yes, Backbone=GroupFree3D2024.10 | 65 | — | — | — | — | — | — | — | — | — | — | — | 46.8 | |
| TokenFusionModality=Point Clouds + RGB2023.10 | 64.9 | — | — | — | — | — | — | — | — | — | — | — | 48.3 | |
| FCAF3DLearning Paradigm=Supervised2022.01 | 64.2 | — | — | — | — | — | — | — | — | — | — | — | 48.9 | |
| FCAF3DModality=Point Clouds2023.10 | 64.2 | — | — | — | — | — | — | — | — | — | — | — | 48.9 | |
| FCAF3DPre-training Setting=Supervised2025.12 | 64.2 | — | — | — | — | — | — | — | — | — | — | — | 48.9 | |
| RBGNetModality=Point Clouds2023.10 | 64.1 | — | — | — | — | — | — | — | — | — | — | — | 47.2 | |
| Point-MAE (GroupFree3D)Pre-trained=Yes, Backbone=GroupFree3D2024.10 | 63.9 | — | — | — | — | — | — | — | — | — | — | — | 46.1 | |
| HyperDet3DModality=Point Clouds2023.10 | 63.5 | — | — | — | — | — | — | — | — | — | — | — | 47.3 | |
| Ours (3DETR)Pre-trained=Yes, Backbone=3DETR2024.10 | 63.5 | — | — | — | — | — | — | — | — | — | — | — | 39.5 | |
| ImVoteNetModality=Point Clouds + RGB2023.10 | 63.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| GroupFreeLearning Paradigm=Supervised2022.01 | 63 | — | — | — | — | — | — | — | — | — | — | — | 45.2 | |
| GroupFreeModality=Point Clouds2023.10 | 63 | — | — | — | — | — | — | — | — | — | — | — | 45.2 | |
| GroupFree3DPre-trained=None2024.10 | 63 | — | — | — | — | — | — | — | — | — | — | — | 45.2 | |
| GroupFreePre-training Setting=Supervised2025.12 | 63 | — | — | — | — | — | — | — | — | — | — | — | 45.2 | |
| VENetModality=Point Clouds2023.10 | 62.5 | — | — | — | — | — | — | — | — | — | — | — | 39.2 | |
| Plain Transformer (GroupFree3D)Pre-trained=None, Backbone=GroupFree3D2024.10 | 62.2 | — | — | — | — | — | — | — | — | — | — | — | 45 | |
| Bridge3D (3DETR)Pre-trained=Yes, Backbone=3DETR2024.10 | 61.8 | — | — | — | — | — | — | — | — | — | — | — | 37.1 | |
| DepthContrastPre-trained=Yes2024.10 | 61.6 | — | — | — | — | — | — | — | — | — | — | — | 35.5 | |
| BRNetLearning Paradigm=Supervised2022.01 | 61.1 | — | — | — | — | — | — | — | — | — | — | — | 43.7 | |
| BRNetModality=Point Clouds2023.10 | 61.1 | — | — | — | — | — | — | — | — | — | — | — | 43.7 | |
| BRNetPre-training Setting=Supervised2025.12 | 61.1 | — | — | — | — | — | — | — | — | — | — | — | 43.7 | |
| DepthContrastLearning Paradigm=Self-supervised, Backbone=VoteNet2022.01 | 60.4 | — | — | — | — | — | — | — | — | — | — | — | 35.4 | |
| IAELearning Paradigm=Self-supervised, Backbone=VoteNet2022.01 | 60.4 | — | — | — | — | — | — | — | — | — | — | — | 36 | |
| DepthContrastPre-training Setting=Self-supervised2025.12 | 60.4 | — | — | — | — | — | — | — | — | — | — | — | 35.4 | |
| IAEBase Architecture=VoteNet, Pre-training Setting=Self-supervised2025.12 | 60.4 | — | — | — | — | — | — | — | — | — | — | — | 36 | |
| DPCOPre-trained=Yes2024.10 | 60.2 | — | — | — | — | — | — | — | — | — | — | — | 35.5 | |
| H3DNetRGB=false2020.06 | 60.1 | 73.8 | 85.6 | 31 | 76.7 | 29.6 | 33.4 | 65.5 | 66.5 | 50.8 | 88.2 | — | — | |
| H3DNetInput=Geo only2020.06 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | 39 | |
| H3DNetModality=Point Clouds2023.10 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | 39 | |
| H3DNet2021.09 | 60.1 | — | — | — | — | — | — | — | — | — | — | — | 39 | |
| PIMAE (3DETR)Pre-trained=Yes, Backbone=3DETR2024.10 | 59.9 | — | — | — | — | — | — | — | — | — | — | — | 33.7 | |
| MLCVNetInput=Geo only2020.04 | 59.8 | 79.2 | 85.8 | 31.9 | 75.8 | 26.5 | 31.3 | 61.5 | 66.3 | 50.4 | 89.1 | — | — | |
| MLCVNetLearning Paradigm=Supervised2022.01 | 59.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| MLCVNetPre-training Setting=Supervised2025.12 | 59.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| RandomRoomsLearning Paradigm=Self-supervised2022.01 | 59.2 | — | — | — | — | — | — | — | — | — | — | — | 35.4 | |
| RandomRoomsPre-training Setting=Self-supervised2025.12 | 59.2 | — | — | — | — | — | — | — | — | — | — | — | 35.4 | |
| 3DETRLearning Paradigm=Supervised2022.01 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 32.7 | |
| 3DETRModality=Point Clouds2023.10 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 32.7 | |
| 3DETRPre-training Setting=Supervised2025.12 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 32.7 | |
| 3DETR-mencoder=masked Transformer encoder2021.09 | 59.1 | — | — | — | — | — | — | — | — | — | — | — | 32.7 | |
| H3DNet w/o refineRGB=false2020.06 | 58.5 | 74.1 | 86.4 | 31.3 | 76.1 | 27.1 | 26.3 | 57.9 | 64.9 | 51.6 | 89.3 | — | — | |
| H3DNet (w/o refine)Input=Geo only2020.06 | 58.5 | — | — | — | — | — | — | — | — | — | — | — | 34.2 | |
| VoteNetimplementation=improved2021.09 | 58.3 | — | — | — | — | — | — | — | — | — | — | — | 33.4 | |
| STRLLearning Paradigm=Self-supervised2022.01 | 58.2 | — | — | — | — | — | — | — | — | — | — | — | 35 | |
| STRLPre-training Setting=Self-supervised2025.12 | 58.2 | — | — | — | — | — | — | — | — | — | — | — | 35 | |
| 3DETRPre-trained=None2024.10 | 58 | — | — | — | — | — | — | — | — | — | — | — | 30.3 | |
| 3DETR2021.09 | 58 | — | — | — | — | — | — | — | — | — | — | — | 30.3 | |
| VoteNetInput=Geo-only2020.04 | 57.7 | 74.4 | 83 | 28.8 | 75.3 | 22 | 29.8 | 62.2 | 64 | 47.3 | 90.1 | — | — | |
| VoteNetRGB=false2020.06 | 57.7 | 74.7 | 83 | 28.8 | 75.3 | 22 | 29.8 | 62.2 | 64 | 47.3 | 90.1 | — | — | |
| VoteNetInput=Geo only2020.06 | 57.7 | — | — | — | — | — | — | — | — | — | — | — | 32.9 | |
| VoteNetLearning Paradigm=Supervised2022.01 | 57.7 | — | — | — | — | — | — | — | — | — | — | — | 32.9 | |
| VoteNetModality=Point Clouds2023.10 | 57.7 | — | — | — | — | — | — | — | — | — | — | — | 35.7 | |
| VoteNetPre-trained=None2024.10 | 57.7 | — | — | — | — | — | — | — | — | — | — | — | 32.9 | |
| VoteNetPre-training Setting=Supervised2025.12 | 57.7 | — | — | — | — | — | — | — | — | — | — | — | 32.9 | |
| Plain Transformer (3DETR)Pre-trained=None, Backbone=3DETR2024.10 | 57.6 | — | — | — | — | — | — | — | — | — | — | — | 31.9 | |
| PointContrastLearning Paradigm=Self-supervised2022.01 | 57.5 | — | — | — | — | — | — | — | — | — | — | — | 34.8 | |
| PointContrastPre-trained=Yes2024.10 | 57.5 | — | — | — | — | — | — | — | — | — | — | — | 34.5 | |
| PointContrastPre-training Setting=Self-supervised2025.12 | 57.5 | — | — | — | — | — | — | — | — | — | — | — | 34.8 | |
| F-PointNetInput=Geo+RGB2020.04 | 54 | 43.3 | 81.1 | 33.3 | 64.2 | 24.7 | 32 | 58.1 | 61.1 | 51.1 | 90.9 | — | — | |
| F-PointNetRGB=true2020.06 | 54 | 43.3 | 81.1 | 33.3 | 64.2 | 24.7 | 32 | 58.1 | 61.1 | 51.1 | 90.9 | — | — | |
| F-PointNetInput=Geo + RGB2020.06 | 54 | — | — | — | — | — | — | — | — | — | — | — | — | |
| BoxNetimplementation=improved2021.09 | 52.4 | — | — | — | — | — | — | — | — | — | — | — | 25.1 | |
| COGInput=Geo+RGB2020.04 | 47.6 | 58.3 | 63.7 | 31.8 | 62.2 | 45.2 | 15.5 | 27.4 | 51 | 51.3 | 70.1 | — | — | |
| COGRGB=true2020.06 | 47.6 | 58.3 | 63.7 | 31.8 | 62.2 | 45.2 | 15.5 | 27.4 | 51 | 51.3 | 70.1 | — | — | |
| COGInput=Geo + RGB2020.06 | 47.6 | — | — | — | — | — | — | — | — | — | — | — | — | |
| 2D-drivenInput=Geo+RGB2020.04 | 45.1 | 43.5 | 64.5 | 31.4 | 48.3 | 27.9 | 25.9 | 41.9 | 50.4 | 37 | 80.4 | — | — | |
| 2D-drivenRGB=true2020.06 | 45.1 | 43.5 | 64.5 | 31.4 | 48.3 | 27.9 | 25.9 | 41.9 | 50.4 | 37 | 80.4 | — | — | |
| 2D-drivenInput=Geo + RGB2020.06 | 45.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| DSSInput=Geo+RGB2020.04 | 42.1 | 44.2 | 78.8 | 11.9 | 61.2 | 20.5 | 6.4 | 15.4 | 53.5 | 50.3 | 78.9 | — | — | |
| DSSRGB=true2020.06 | 42.1 | 44.2 | 78.8 | 11.9 | 61.2 | 20.5 | 6.4 | 15.4 | 53.5 | 50.3 | 78.9 | — | — | |
| DSSInput=Geo + RGB2020.06 | 42.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| H3DNetRGB=false, Refinement=true2020.06 | 39 | 47.6 | 52.9 | 8.6 | 60.1 | 8.4 | 20.6 | 45.6 | 50.4 | 27.1 | 69.1 | — | — | |
| H3DNetRGB=false, Refinement=false2020.06 | 34.2 | 48.9 | 50.6 | 5 | 55.6 | 6.3 | 14.6 | 32.7 | 45.1 | 23.3 | 60.1 | — | — | |
| VoteNetRGB=false2020.06 | 32.9 | 49.9 | 47.3 | 4.6 | 54.1 | 5.2 | 13.6 | 35 | 41.4 | 19.7 | 58.6 | — | — | |
| 2D-drivenRGB=true, room layout context=true2020.01 | — | 43.5 | 64.5 | 31.4 | 48.3 | 27.9 | 25.9 | 41.9 | 50.4 | 37 | 80.4 | 45.1 | — | |
| 4DContrastPre-trained=Yes2024.10 | — | — | — | — | — | — | — | — | — | — | — | — | 38.2 | |
| COGRGB=true, room layout context=true2020.01 | — | 58.3 | 63.7 | 31.8 | 62.2 | 45.2 | 15.5 | 27.4 | 51 | 51.3 | 70.1 | 47.6 | — | |
| DSSRGB=true2020.01 | — | 44.2 | 78.8 | 11.9 | 61.2 | 20.5 | 6.4 | 15.4 | 53.5 | 50.3 | 78.9 | 42.1 | — | |
| F-PointNetRGB=true2020.01 | — | 43.3 | 81.1 | 33.3 | 64.2 | 24.7 | 32 | 58.1 | 61.1 | 51.1 | 90.9 | 54 | — | |
| Hou et al.Pre-trained=Yes2024.10 | — | — | — | — | — | — | — | — | — | — | — | — | 36.4 | |
| IMVOTENETRGB=true2020.01 | — | 75.9 | 87.6 | 41.3 | 76.7 | 28.7 | 41.4 | 69.9 | 70.7 | 51.1 | 90.5 | 63.4 | — | |
| PointFusionRGB=true2020.01 | — | 37.3 | 68.6 | 37.7 | 55.1 | 17.2 | 23.9 | 32.3 | 53.8 | 31 | 83.8 | 45.4 | — | |
| VOTENETRGB=false2020.01 | — | 74.4 | 83 | 28.8 | 75.3 | 22 | 29.8 | 62.2 | 64 | 47.3 | 90.1 | 57.7 | — | |
| VOTENET +region featureRGB=true, mode=concatenate region-level one-hot class confidence2020.01 | — | 71.7 | 86.1 | 34 | 74.7 | 26 | 34.2 | 64.3 | 66.5 | 49.7 | 88.4 | 59.6 | — | |
| VOTENET +RGBRGB=true, mode=append RGB values to point features2020.01 | — | 70 | 82.8 | 27.6 | 73.1 | 23.2 | 27.2 | 60.7 | 63.7 | 48 | 86.9 | 56.3 | — |