Semantic Segmentation on ScanNet 2-views (context views)
63.23mIoUEPS3D
Evaluation results
| Method | Links | ||
|---|---|---|---|
| EPS3DPerception Model=End-to-end 3D, FF.=true, Sem.=true, Ins.=true, Time=0.73s2026.06 | 63.23 | 84.65 | |
| Uni3RPerception Model=Two-stage 3D, FF.=true, Sem.=true, Ins.=true, Time=0.85s2026.06 | 52.33 | 81.88 | |
| LSMPerception Model=Two-stage 3D, FF.=true, Sem.=true, Ins.=true, Time=0.43s2026.06 | 50.34 | 77.4 | |
| LSegPerception Model=2D, FF.=true, Sem.=true, Ins.=false, Time=0.2s2026.06 | 47.01 | 78.91 | |
| NeRF-DFFPerception Model=Two-stage 3D, FF.=false, Sem.=true, Ins.=false, Time=1min2026.06 | 45.4 | 71.73 | |
| Feature-3DGSPerception Model=Two-stage 3D, FF.=false, Sem.=true, Ins.=false, Time=18min2026.06 | 44.53 | 72.76 |