3D Instance Segmentation on ScanNet200 (val)
53.7mAPSegVGGT
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| SegVGGTevaluation_setting=class-agnostic, sampled_scenes=10, down_sampling_rate=40x2026.03 | 53.7 | — | — | — | 80.2 | 89.9 | — | — | — | — | |
| GVC-SegEvaluation Protocol=Class-agnostic2026.06 | 49.1 | — | — | — | 67.2 | 72.2 | 70.9 | 90.6 | 95.9 | — | |
| Any3DIS2D Segmenter=3D + SAM2-L2024.11 | 42.5 | — | — | — | 51.2 | 54.5 | — | — | — | — | |
| Open3DIS2D Segmenter=3D + SAM-HQ2024.11 | 41.5 | — | — | — | 51.6 | 56.3 | — | — | — | — | |
| Open3DISEvaluation Protocol=Class-agnostic, Input Modality=2D+3D2026.06 | 41.5 | — | — | — | 51.6 | 56.3 | 74.8 | 90.9 | 97.8 | — | |
| ISBNet2D Segmenter=3D2024.11 | 40.2 | — | — | — | 50 | 54.6 | — | — | — | — | |
| ISBNetEvaluation Protocol=Class-agnostic, Input Modality=3D2026.06 | 40.2 | — | — | — | 50 | 54.6 | 66.8 | 80.4 | 87.4 | — | |
| Any3DIS2D Segmenter=SAM2-L2024.11 | 32.5 | — | — | — | 45.2 | 55 | — | — | — | — | |
| Open3DIS2D Segmenter=SAM-HQ2024.11 | 31.5 | — | — | — | 45.3 | 51.1 | — | — | — | — | |
| LaSSMVariant=•2026.02 | 29.3 | — | — | — | 39.2 | 44.5 | — | — | — | — | |
| Mask-Attention-Free Transformer2023.09 | 29.2 | — | — | — | 38.2 | 43.3 | — | — | — | — | |
| MAFTVenue=ICCV 232026.02 | 29.2 | — | — | — | 38.2 | 43.3 | — | — | — | — | |
| SGIFormerVenue=TCSVT 252026.02 | 29.2 | — | — | — | 39.4 | 44.2 | — | — | — | — | |
| LaSSMVariant=◦2026.02 | 28.8 | — | — | — | 38.2 | 43.7 | — | — | — | — | |
| EmbodiedSAMOnline=true, Zero-Shot=false, Input=Posed RGB-D, Speed (ms)=802025.12 | 28.8 | — | — | — | 42.7 | 54.2 | — | — | — | — | |
| IGGTevaluation_setting=class-agnostic, sampled_scenes=10, down_sampling_rate=40x2026.03 | 28.2 | — | — | — | 42.4 | 55.3 | — | — | — | — | |
| QueryFormerVenue=ICCV 232026.02 | 28.1 | — | — | — | 37.1 | 43.4 | — | — | — | — | |
| Mask3Dbackbone=Res16UNet34C2023.09 | 27.4 | — | — | — | 37 | 42.3 | — | — | — | — | |
| Mask3DVenue=ICRA 232026.02 | 27.4 | — | — | — | 37 | 42.3 | — | — | — | — | |
| Mask3DMask Supervision=Supervised, Semantic Supervision=Supervised2024.01 | 26.9 | — | — | — | 36.2 | 41.4 | — | — | — | — | |
| Open3DIS2D Segmenter=SAM2-L2024.11 | 26.7 | — | — | — | 40.2 | 53.8 | — | — | — | — | |
| MSTA3D2024.11 | 26.2 | — | — | — | 35.2 | 40.1 | — | — | — | — | |
| MSTA3DVenue=MM 242026.02 | 26.2 | — | — | — | 35.2 | 40.1 | — | — | — | — | |
| OpenTrack3DInputs=3D + 2D, Region Proposal Network Name=YOLO-World + SAM2 (2024), No GT RPN=true, Open vocab RPN=true2026.04 | 26 | — | — | — | 37.7 | 45.4 | — | — | — | 356 | |
| Any3DISSegmenter=3D + SAM2-L2024.11 | 25.8 | 27.4 | 23.8 | 26.4 | — | — | — | — | — | — | |
| Any3DISInputs=3D + 2D, Region Proposal Network Name=ISBNet (2023) + SAM2 (2024), No GT RPN=false, Open vocab RPN=false2026.04 | 25.8 | 27.4 | 23.8 | 26.4 | — | — | — | — | — | — | |
| Details MatterInputs=3D + 2D, Region Proposal Network Name=Mask3D (2023) + GSAM (2024), No GT RPN=false, Open vocab RPN=false2026.04 | 25.8 | 26.3 | 23.2 | 28.2 | 32.5 | 36.2 | — | — | — | — | |
| SPFormerreproduced=true2023.09 | 25.2 | — | — | — | 33.8 | 39.6 | — | — | — | — | |
| SPFormerreproduced=true2024.11 | 25.2 | — | — | — | 33.8 | 39.6 | — | — | — | — | |
| SPFormerVenue=AAAI 232026.02 | 25.2 | — | — | — | 33.8 | 39.6 | — | — | — | — | |
| Open-YOLO 3DInputs=3D + 2D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 24.7 | 27.8 | 24.3 | 21.6 | 31.7 | 36.2 | — | — | — | 21.8 | |
| ISBNet2024.11 | 24.5 | — | — | — | 32.7 | — | — | — | — | — | |
| PTv3 + PPTBackbone=PTv3, Pre-training=PPT, Mode=Fine-tuned, Instance Segmentation Framework=PointGroup, Params.=46.3M2023.08 | 24 | — | — | — | 34.1 | 40.8 | — | — | — | — | |
| Open3DISSegmenter=3D + SAM-HQ2024.11 | 23.7 | 27.8 | 21.2 | 21.8 | — | — | — | — | — | — | |
| Open3DISInputs=3D + 2D, Region Proposal Network Name=Superpoints (2004) + ISBNet (2023) + Grounded-SAM (2024), No GT RPN=false, Open vocab RPN=false2026.04 | 23.7 | 27.8 | 21.2 | 21.8 | 29.4 | 32.8 | — | — | — | 33.5 | |
| PTv3Backbone=PTv3, Pre-training=None, Instance Segmentation Framework=PointGroup, Params.=46.2M2023.08 | 23.1 | — | — | — | 33.2 | 40.1 | — | — | — | — | |
| TD3Dreproduced=true2024.11 | 23.1 | — | — | — | 34.8 | 40.4 | — | — | — | — | |
| TD3DVenue=WACV 242026.02 | 23.1 | — | — | — | 34.8 | 40.4 | — | — | — | — | |
| PanSt3Revaluation_setting=class-agnostic, sampled_scenes=10, down_sampling_rate=40x2026.03 | 20.8 | — | — | — | 41.4 | 59 | — | — | — | — | |
| MaskClusteringOnline=false, Zero-Shot=true, Input=Posed RGB-D2025.12 | 19.7 | — | — | — | 36.4 | 51.4 | — | — | — | — | |
| SparseUNet + PPTBackbone=SparseUNet, Pre-training=PPT, Mode=Fine-tuned, Instance Segmentation Framework=PointGroup, Params.=41.0M2023.08 | 19.4 | — | — | — | 29.4 | 36.8 | — | — | — | — | |
| Any3DISSegmenter=SAM2-L2024.11 | 19.1 | 17.5 | 17.2 | 23.3 | — | — | — | — | — | — | |
| OnlineAnySegOnline=true, Zero-Shot=true, Input=Posed RGB-D, Speed (ms)=3000*2025.12 | 18.6 | — | — | — | 36.1 | 53.5 | — | — | — | — | |
| Open3DISSegmenter=SAM-HQ2024.11 | 18.2 | 18.9 | 16.5 | 19.2 | — | — | — | — | — | — | |
| Mask ClusteringEvaluation Protocol=Class-agnostic2026.06 | 17.4 | — | — | — | 33.3 | 46.7 | — | — | — | — | |
| SparseUNet + MSCBackbone=SparseUNet, Pre-training=MSC, Instance Segmentation Framework=PointGroup, Params.=39.2M2023.08 | 17.3 | — | — | — | 26.8 | 34.3 | — | — | — | — | |
| MoonSeg3ROnline=true, Zero-Shot=true, Input=Monocular, Speed (ms)=55 + 662025.12 | 16.7 | — | — | — | 33.3 | 50 | — | — | — | — | |
| SpaCeFormer w/ ProposalInputs=3D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 16.7 | 22.8 | 16.4 | 9.9 | 21.9 | 24.8 | — | — | — | 1.1 | |
| SparseUNetBackbone=SparseUNet, Pre-training=None, Instance Segmentation Framework=PointGroup, Params.=39.2M2023.08 | 15.8 | — | — | — | 24.5 | 32.2 | — | — | — | — | |
| OpenMask3DSegmenter=Mask3D [25]2024.11 | 15.4 | 17.1 | 14.1 | 14.9 | — | — | — | — | — | — | |
| OpenMask3Duse 2D=true2023.09 | 15.4 | 17.1 | 14.1 | 14.9 | 19.9 | 23.1 | — | — | — | — | |
| OpenMask3DInputs=3D + 2D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 15.4 | 17.1 | 14.1 | 14.9 | 19.9 | 23.1 | — | — | — | 553.9 | |
| OpenMask3DMask Supervision=Supervised, Semantic Supervision=Zero-shot2024.01 | 15.1 | — | — | — | 19.6 | 22.6 | — | — | — | — | |
| OVIR-3D2D Segmenter=SAM-HQ2024.11 | 14.4 | — | — | — | 27.5 | 38.8 | — | — | — | — | |
| OVIR-3DOnline=false, Zero-Shot=true, Input=Posed RGB-D2025.12 | 14.4 | — | — | — | 27.5 | 38.8 | — | — | — | — | |
| OVIR-3DEvaluation Protocol=Class-agnostic2026.06 | 14.4 | — | — | — | 27.5 | 38.8 | — | — | — | — | |
| Search3DSegmenter=Mask3D [25]2024.11 | 14.3 | 16.1 | 13.6 | 12.9 | — | — | — | — | — | — | |
| OnlineAnySeg-MOnline=true, Zero-Shot=true, Input=Monocular, Speed (ms)=3000 + 662025.12 | 13.4 | — | — | — | 26.8 | 43.2 | — | — | — | — | |
| SAM2ObjectInputs=3D + 2D, Region Proposal Network Name=SAM2 (2024), No GT RPN=true, Open vocab RPN=true2026.04 | 13.3 | — | — | — | 19 | 23.8 | — | — | — | — | |
| SAI3DSegmenter=SAM-HQ2024.11 | 12.7 | 12.1 | 10.4 | 16.2 | — | — | — | — | — | — | |
| SAI3DInputs=3D + 2D, Region Proposal Network Name=Superpoints (2004) + SAM (2023), No GT RPN=true, Open vocab RPN=true2026.04 | 12.7 | 12.1 | 10.4 | 16.2 | 18.8 | 24.1 | — | — | — | 75.2 | |
| MaskClusteringMask Supervision=Zero-shot, Semantic Supervision=Zero-shot2024.01 | 12 | — | — | — | 23.3 | 30.1 | — | — | — | — | |
| MaskClusteringInputs=3D + 2D, Region Proposal Network Name=CropFormer (2023), No GT RPN=true, Open vocab RPN=true2026.04 | 12 | — | — | — | 23.3 | 30.1 | — | — | — | — | |
| Mosaic3DInputs=3D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 11.8 | 21.8 | 7.2 | 5.4 | 16 | 17.8 | — | — | — | 1 | |
| OpenScene (2D Fusion)use 2D=true2023.09 | 11.7 | 13.4 | 11.6 | 9.9 | 15.2 | 17.8 | — | — | — | — | |
| OpenScene-2DInputs=3D + 2D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 11.7 | 13.4 | 11.6 | 9.9 | 15.2 | 17.8 | — | — | — | — | |
| OVIR-3DSegmenter=SAM-HQ2024.11 | 11.2 | 12.6 | 11.5 | 10.3 | — | — | — | — | — | — | |
| SpaCeFormerInputs=3D, Region Proposal Network Name=Not required (proposal-free), No GT RPN=true, Open vocab RPN=true2026.04 | 11.1 | 13.2 | 9.5 | 10.6 | 18.8 | 24.3 | — | — | — | 0.14 | |
| SAM3DOnline=true, Zero-Shot=true, Input=Posed RGB-D, Speed (ms)=1252025.12 | 9.6 | — | — | — | 24.8 | 49.6 | — | — | — | — | |
| OVIR-3DMask Supervision=Zero-shot, Semantic Supervision=Zero-shot2024.01 | 9.3 | — | — | — | 18.7 | 25 | — | — | — | — | |
| OpenIns3Duse 2D=false2023.09 | 8.8 | 16 | 6.5 | 4.2 | 10.3 | 14.4 | — | — | — | — | |
| OpenIns3DInputs=3D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 8.8 | 16 | 6.5 | 4.2 | 10.3 | 14.4 | — | — | — | 285.2 | |
| RegionPLCInputs=3D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 6.3 | 15.6 | 1 | 1.7 | 8.6 | 9.7 | — | — | — | 1 | |
| OpenScene (2D/3D Ens.)use 2D=true2023.09 | 5.3 | 11 | 3.2 | 1.1 | 6.7 | 8.1 | — | — | — | — | |
| OpenScene-2D/3DInputs=3D + 2D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 5.3 | 11 | 3.2 | 1.1 | 6.7 | 8.1 | — | — | — | — | |
| SuperpointEvaluation Protocol=Class-agnostic2026.06 | 5 | — | — | — | 12.7 | 38.9 | — | — | — | — | |
| OpenScene (3D Distill)use 2D=false2023.09 | 4.8 | 10.6 | 2.6 | 0.7 | 6.2 | 7.2 | — | — | — | — | |
| OpenScene-3DInputs=3D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 4.8 | 10.6 | 2.6 | 0.7 | 6.2 | 7.2 | — | — | — | 1.1 | |
| Mosaic3D w/ DecoderInputs=3D, Region Proposal Network Name=Not required (proposal-free), No GT RPN=true, Open vocab RPN=true2026.04 | 3.9 | 6.6 | 2.1 | 2.8 | 7 | 12.3 | — | — | — | 1.2 | |
| OpenIns3D†Inputs=3D, Region Proposal Network Name=Mask3D (2023), No GT RPN=false, Open vocab RPN=false2026.04 | 3.3 | 7 | 1.4 | 1.2 | 5 | 5.6 | — | — | — | 50 | |
| Mosaic3DInputs=3D, Region Proposal Network Name=Segment3D (2024a), No GT RPN=true, Open vocab RPN=true2026.04 | 2.7 | 3.8 | 2 | 2.4 | 4.2 | 5.7 | — | — | — | 1.9 | |
| OpenIns3D†Inputs=3D, Region Proposal Network Name=Segment3D (2024a), No GT RPN=true, Open vocab RPN=true2026.04 | 1.7 | 3.2 | 0.8 | 1 | 2.7 | 3.7 | — | — | — | 64.8 | |
| DBSCANEvaluation Protocol=Class-agnostic2026.06 | 1.6 | — | — | — | 5.5 | 32.1 | — | — | — | — | |
| RegionPLCInputs=3D, Region Proposal Network Name=Segment3D (2024a), No GT RPN=true, Open vocab RPN=true2026.04 | 1.5 | 2.3 | 0.2 | 1.9 | 2.1 | 2.6 | — | — | — | 1.9 | |
| OpenScene-3DInputs=3D, Region Proposal Network Name=Segment3D (2024a), No GT RPN=true, Open vocab RPN=true2026.04 | 0.6 | 1.4 | 0.4 | 0 | 1 | 1.6 | — | — | — | 2 | |
| CSC2022.10 | — | 22.3 | 8.2 | 4.6 | — | — | — | — | — | — | |
| LGround2022.10 | — | 27.5 | 10.8 | 6 | — | — | — | — | — | — | |
| Mask3D2022.10 | — | 38.3 | 26.3 | 16.8 | — | — | — | — | — | — | |
| Mink34D2022.10 | — | 24.6 | 8.3 | 4.3 | — | — | — | — | — | — | |
| PointGroup2024.11 | — | — | — | — | 24.5 | — | — | — | — | — | |
| PointGroup + LGround2024.11 | — | — | — | — | 26.1 | — | — | — | — | — | |
| SparseUNet + CSCBackbone=SparseUNet, Pre-training=CSC, Instance Segmentation Framework=PointGroup, Params.=39.2M2023.08 | — | — | — | — | 25.2 | — | — | — | — | — | |
| SparseUNet + LGroundBackbone=SparseUNet, Pre-training=LGround, Instance Segmentation Framework=PointGroup, Params.=39.2M2023.08 | — | — | — | — | 26.1 | — | — | — | — | — | |
| SparseUNet + PCBackbone=SparseUNet, Pre-training=PC, Instance Segmentation Framework=PointGroup, Params.=39.2M2023.08 | — | — | — | — | 24.9 | — | — | — | — | — |