Instance Segmentation on ScanNetV2 (val)
75.9mAP@0.5MAFT
Evaluation results
| Method | Links | |||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MAFTInput Data Type=Mesh Sampled Point Cloud2024.01 | 75.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.4 | — | — | |
| QueryFormerInput Data Type=Mesh Sampled Point Cloud2024.01 | 74.2 | 83.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.5 | — | — | |
| Mask3DInput Data Type=Mesh Sampled Point Cloud2024.01 | 73.7 | 83.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.2 | — | — | |
| ODINInput Data Type=Sensor RGBD Point Cloud, Backbone=Swin-B2024.01 | 71 | 83.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50 | — | — | |
| PBNetInput Data Type=Mesh Sampled Point Cloud2024.01 | 70.5 | 78.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 54.3 | — | — | |
| ODINInput Data Type=Sensor RGBD Point Cloud, Backbone=ResNet502024.01 | 69.8 | 83.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.8 | — | — | |
| SoftGroupInput Data Type=Mesh Sampled Point Cloud2024.01 | 67.6 | 78.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 46 | — | — | |
| PTv3Training=Multi-dataset joint training, Segmentation Framework=PointGroup [35]2023.12 | 63.5 | 78.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 42.1 | — | — | |
| ChorusEvaluation Protocol=fine-tuning2025.12 | 63.4 | 78.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SonataEvaluation Protocol=fine-tuning2025.12 | 63.1 | 77.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MPECFine-tuned=true2025.04 | 62.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GCFine-tuned=true2025.04 | 62.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SonataEvaluation Protocol=decoder probing2025.12 | 62.1 | 77.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PTv3Training=Trained from scratch, Segmentation Framework=PointGroup [35]2023.12 | 61.7 | 77.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 40.9 | — | — | |
| ChorusEvaluation Protocol=decoder probing2025.12 | 60.5 | 76.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PTv2Training=Trained from scratch, Segmentation Framework=PointGroup [35]2023.12 | 60 | 76.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 38.3 | — | — | |
| Mask3DInput Data Type=Sensor RGBD Point Cloud, Trained by this paper=true2024.01 | 60 | 69.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.9 | — | — | |
| Box2MaskSupervision=100% Box2022.08 | 59.7 | 71.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MSCFine-tuned=true2025.04 | 59.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CSCFine-tuned=true2025.04 | 59.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Contrastive Scene ContextsIoU threshold=0.52020.12 | 59.4 | — | 80.5 | 74.1 | 57 | 50.8 | 88.7 | 32.5 | 47.4 | 42.9 | 48 | 54.7 | 33.8 | 49.5 | — | 71.3 | 61.4 | 67.2 | 98.2 | 42 | 68.9 | — | — | — | — | |
| PCFine-tuned=true2025.04 | 58 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PointContrastIoU threshold=0.52020.12 | 58 | — | 80.5 | 72.1 | 57.8 | 49.4 | 87.2 | 24 | 49.9 | 30.2 | 49 | 50.8 | 35.6 | 53 | — | 61.7 | 71.7 | 67 | 98.3 | 40.7 | 65.2 | — | — | — | — | |
| PTv3Evaluation Protocol=supervised2025.12 | 57.9 | 74.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PointGroupSupervision=100%2022.08 | 56.9 | 71.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 34.8 | — | — | |
| MinkUNetTraining=Trained from scratch, Segmentation Framework=PointGroup [35]2023.12 | 56.9 | 72.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 36 | — | — | |
| SCFine-tuned=true2025.04 | 56.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ScratchIoU threshold=0.52020.12 | 56.9 | — | 75.3 | 70 | 53 | 49 | 87.4 | 32.8 | 41.7 | 30.8 | 47.4 | 52.4 | 30.8 | 48.6 | — | 68.4 | 66.5 | 71.1 | 99.9 | 39.6 | 60.1 | — | — | — | — | |
| RWSegSupervision=0.02%2022.08 | 56.4 | 71.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 34.7 | — | — | |
| SonataEvaluation Protocol=linear probing2025.12 | 53.9 | 72.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TWISTSupervision=20%2022.08 | 52.9 | 66.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 32.8 | — | — | |
| TWISTSupervision=10%2022.08 | 49.7 | 63 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 30.6 | — | — | |
| 3D-WSISSupervision=0.02%2022.08 | 47.2 | 67.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 28.1 | — | — | |
| ChorusEvaluation Protocol=linear probing2025.12 | 46.9 | 66.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TWISTSupervision=5%2022.08 | 44.1 | 56.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 27 | — | — | |
| SegGroup (PointGroup)Supervision=0.02%2022.08 | 43.4 | 62.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 23.4 | — | — | |
| Mask3DEvaluation Protocol=Fine-tuning, Pre-training Method=Masked 3D prior learning2023.02 | 41.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 22.7 | 22.8 | — | — | |
| SPIBSupervision=100% Box2022.08 | 38.6 | 61.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MAE-unsupIN→SNEvaluation Protocol=Fine-tuning, Pre-training Method=Unsupervised MAE ImageNet to ScanNet2023.02 | 37.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 20.3 | 20.7 | — | — | |
| CSC-50 (PointGroup)Supervision=50 pts/scene2022.08 | 35.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Pri3DEvaluation Protocol=Fine-tuning, Pre-training Method=3D-based pre-training2023.02 | 34.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 18.7 | 18.3 | — | — | |
| PTv3Training=Multi-dataset joint training, Segmentation Framework=PointGroup [35]2023.12 | 34.1 | 40.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 24 | — | — | |
| MoCoV2-supIN→SNEvaluation Protocol=Fine-tuning, Pre-training Method=MoCoV2 supervised ImageNet to ScanNet2023.02 | 33.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 18.1 | 18.3 | — | — | |
| PTv3Training=Trained from scratch, Segmentation Framework=PointGroup [35]2023.12 | 33.2 | 40.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 23.1 | — | — | |
| ImageNet Pretrain (supIN)Evaluation Protocol=Fine-tuning, Pre-training Method=Supervised ImageNet2023.02 | 32.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 17.8 | 17.6 | — | — | |
| PTv2Training=Trained from scratch, Segmentation Framework=PointGroup [35]2023.12 | 31.9 | 39.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 21.4 | — | — | |
| MPNet2020.01 | 31 | 49.3 | 69.4 | 59.8 | 2.7 | 23.7 | 71.1 | 4.5 | 8.4 | 18.3 | 11.6 | 17.3 | 4.8 | 21.8 | 57 | 13.4 | 27.7 | 41.8 | 87.3 | 18.3 | — | — | — | — | — | |
| CSC-20 (PointGroup)Supervision=20 pts/scene2022.08 | 27.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ScratchEvaluation Protocol=Fine-tuning, Pre-training Method=None2023.02 | 25.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 13.1 | 12.2 | — | — | |
| 3D-BEVISYear=2019, input=multi-view based2020.01 | 24.8 | — | 66.7 | 56.6 | 7.6 | 3.5 | 39.4 | 2.7 | 3.5 | 9.8 | 9.9 | 3 | 2.5 | 9.8 | 37.5 | 12.6 | 60.4 | 18.1 | 85.4 | 17.1 | — | — | — | — | — | |
| MinkUNetTraining=Trained from scratch, Segmentation Framework=PointGroup [35]2023.12 | 24.5 | 32.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 15.8 | — | — | |
| ASISYear=20192020.01 | 24 | 41.5 | 29.9 | 50.5 | 0 | 16.7 | 57.7 | 0 | 18.4 | 7.8 | 14.8 | 12.9 | 1.8 | 12.4 | 38 | 10.2 | 36.9 | 37.4 | 71.7 | 14.5 | — | — | — | — | — | |
| R-PointNetYear=20192020.01 | 23.5 | 40 | 51.3 | 52.3 | 12.5 | 15.2 | 61.8 | 0 | 1.5 | 7.6 | 29 | 11.7 | 14.7 | 25 | 3.7 | 14 | 34.5 | 18.1 | 53 | 16.1 | — | — | — | — | — | |
| TWISTSupervision=1%2022.08 | 17.1 | 26.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 9.6 | — | — | |
| SGPNYear=20192020.01 | 14.3 | 35.1 | 20.8 | 39 | 16.9 | 6.5 | 27.5 | 2.9 | 6.9 | 0 | 8.7 | 4.3 | 1.4 | 2.7 | 0 | 11.2 | 35.1 | 16.8 | 43.8 | 13.8 | — | — | — | — | — | |
| MaskRCNNYear=20172020.01 | 5.8 | 26.1 | 33.3 | 0.2 | 0 | 5.3 | 0.2 | 0.2 | 2.1 | 0 | 4.5 | 2.4 | 23.8 | 6.5 | 0 | 1.4 | 10.7 | 2 | 11 | 0.6 | — | — | — | — | — | |
| MSCEvaluation Protocol=linear probing2025.12 | 5.3 | 13.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| 4DContrasttraining_paradigm=fine-tuning, baseline_comparison=Gain from scratch2021.12 | 4.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BFL2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 20.3 | 444 | |
| HAIS2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 30.9 | 578 | |
| Maft2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 20.1 | 412 | |
| Mask3D2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 39.6 | 578 | |
| QueryFormer2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 42.3 | 487 | |
| SoftGroup2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 30.9 | 588 | |
| SPFormer2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 17.6 | 430 | |
| SSTNet2025.02 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 729 |