3D Semantic Segmentation on ScanNet (mIoU, F-mIoU, mAcc)
65.3mIoUPano3D (MUSt3R)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Pano3D (MUSt3R)Modality=RGB-only2026.06 | 65.3 | — | 76.6 | |
| UniSemNumber of View Inputs=162026.03 | 55.2 | — | 83.2 | |
| UniSemNumber of View Inputs=82026.03 | 53.3 | — | 82.1 | |
| Uni3RNumber of View Inputs=162026.03 | 52.2 | — | 79.5 | |
| Uni3RNumber of View Inputs=82026.03 | 52.1 | — | 80.7 | |
| UNITEModality=RGB-only2026.06 | 48.7 | — | 68.3 | |
| PanSt3RModality=RGB-only, Evaluation source=Evaluated by [21]2026.06 | 42.6 | — | 49.7 | |
| OursCLIP Backbone=OVSeg, Zero-shot=true2024.11 | 39.3 | 50.8 | 60.1 | |
| RADSeg+3D Aggregation Type=Probability Space2025.11 | 38.58 | 48.49 | — | |
| OGScene3DScene=scene0378_00, Setting=Open-vocabulary2026.03 | 37.7 | 38.61 | 58.46 | |
| RADSeg3D Aggregation Type=Probability Space2025.11 | 36.41 | 46.92 | — | |
| RADSeg3D Aggregation Type=Feature Space2025.11 | 36.29 | 45.64 | — | |
| TextRegion3D Aggregation Type=Probability Space2025.11 | 35.83 | 46.09 | — | |
| OursCLIP Backbone=Vit-H-14, Zero-shot=true2024.11 | 35.6 | 47.7 | 57.4 | |
| TextRegion3D Aggregation Type=Feature Space2025.11 | 34.42 | 44.47 | — | |
| Trident3D Aggregation Type=Probability Space2025.11 | 32.78 | 43.67 | — | |
| Trident (w/refine)3D Aggregation Type=Probability Space2025.11 | 32.63 | 44.17 | — | |
| Trident3D Aggregation Type=Feature Space2025.11 | 32.23 | 43.12 | — | |
| LSMModality=RGB-only, Evaluation source=Evaluated by [21]2026.06 | 32.2 | — | 41.1 | |
| OGScene3DScene=scene0011_00, Setting=Open-vocabulary2026.03 | 31.47 | 42.33 | 55.55 | |
| RayFronts3D Aggregation Type=Probability Space2025.11 | 31.05 | 42.3 | — | |
| RayFronts3D Aggregation Type=Feature Space2025.11 | 31.02 | 42.14 | — | |
| OGScene3DScene=Average, Setting=Open-vocabulary2026.03 | 29.44 | 36.34 | 46.41 | |
| OGScene3DTime/scene=10m13s2026.03 | 29.44 | 36.34 | 46.41 | |
| Uni3RModality=RGB-only, Evaluation source=Evaluated by [21]2026.06 | 29.3 | — | 39.4 | |
| Talk2Dino3D Aggregation Type=Probability Space2025.11 | 29.13 | 34.47 | — | |
| SC-CLIP3D Aggregation Type=Probability Space2025.11 | 28.59 | 40.89 | — | |
| Talk2Dino3D Aggregation Type=Feature Space2025.11 | 28.58 | 32.88 | — | |
| OGScene3DScene=scene0050_00, Setting=Open-vocabulary2026.03 | 28.54 | 50.53 | 40.77 | |
| SC-CLIP3D Aggregation Type=Feature Space2025.11 | 28.04 | 40.57 | — | |
| HOV-SGScene=scene0011_00, Setting=Open-vocabulary2026.03 | 27.8 | 42.19 | 44.59 | |
| HOV-SGScene=scene0378_00, Setting=Open-vocabulary2026.03 | 26.67 | 29.15 | 51.28 | |
| OGScene3DScene=scene0231_00, Setting=Open-vocabulary2026.03 | 25.35 | 26.93 | 42.21 | |
| OGScene3DScene=scene0518_00, Setting=Open-vocabulary2026.03 | 24.13 | 23.29 | 35.05 | |
| HOV-SGScene=scene0231_00, Setting=Open-vocabulary2026.03 | 22.72 | 26.39 | 40.5 | |
| HOV-SGCLIP Backbone=Vit-H-14, Zero-shot=true2024.11 | 22.2 | 30.3 | 43.1 | |
| HOV-SGScene=Average, Setting=Open-vocabulary2026.03 | 22.2 | 30.3 | 43.1 | |
| HOV-SGTime/scene=31m56s2026.03 | 22.2 | 30.3 | 43.1 | |
| HOV-SGCLIP Backbone=OVSeg, Zero-shot=true2024.11 | 21.4 | 25.8 | 42 | |
| OVI-MAPOnline=true2026.03 | 17.5 | — | 27.6 | |
| HOV-SGScene=scene0050_00, Setting=Open-vocabulary2026.03 | 17.23 | 36.5 | 41.62 | |
| HOV-SGScene=scene0518_00, Setting=Open-vocabulary2026.03 | 16.58 | 17.27 | 37.51 | |
| OVI-MAPOnline=true, Frame rate=30 fps2026.03 | 16.3 | — | 25.4 | |
| ConceptGraphCLIP Backbone=Vit-H-14, Zero-shot=true2024.11 | 16 | 20 | 28 | |
| ConceptGraphsScene=Average, Setting=Open-vocabulary2026.03 | 16 | 20 | 28 | |
| ConceptGraphsTime/scene=122m5s2026.03 | 16 | 20 | 28 | |
| OVO-SLAMOnline=true, Frame rate=30 fps2026.03 | 15.5 | — | 26.9 | |
| ConceptGraphCLIP Backbone=OVSeg, Zero-shot=true2024.11 | 15 | 18 | 23 | |
| OVO-SLAMOnline=true2026.03 | 14.6 | — | 27.8 | |
| ConceptFusionCLIP Backbone=Vit-H-14, Zero-shot=true2024.11 | 11 | 12 | 21 | |
| Semantic GaussianOnline=false2026.03 | 10.7 | — | 17.8 | |
| Pe3RModality=RGB-only, Evaluation source=Evaluated by [21]2026.06 | 10.7 | — | 19.8 | |
| OpenSceneOnline=false2026.03 | 10.3 | — | 17.3 | |
| Mask3D+OM3DOnline=false2026.03 | 8.6 | — | 17.5 | |
| OpenFusionOnline=true2026.03 | 8.3 | — | 11.5 | |
| ConceptFusionCLIP Backbone=OVSeg, Zero-shot=true2024.11 | 8 | 11 | 15 | |
| Segment3D+OM3DOnline=false2026.03 | 4.5 | — | 13.5 |