Open-vocabulary Semantic Segmentation on ScanNet Source View (40 unseen scenes)
73.44mIoULangFlash
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LangFlashPer-Scene Reconstruction Time=0.180s2026.05 | 73.44 | 87.46 | |
| LangFlash (zero-shot)Per-Scene Reconstruction Time=0.180s, Evaluation Protocol=zero-shot2026.05 | 62.17 | 78.78 | |
| Uni3RPer-Scene Recon. Time=0.162s2025.08 | 54.03 | 82.55 | |
| LSegSfM Reconstruction Time=N/A, Per-Scene Reconstruction Time=N/A2026.05 | 52.78 | 76.54 | |
| LSM*Per-Scene Recon. Time=0.108s2025.08 | 50.34 | 77.4 | |
| LSMPer-Scene Reconstruction Time=0.108s2026.05 | 50.34 | 77.4 | |
| LSegSfM Recon. Time=N/A, Per-Scene Recon. Time=N/A2025.08 | 47.01 | 78.91 | |
| NeRF-DFFSfM Recon. Time=20.52s, Per-Scene Recon. Time=1min2025.08 | 45.4 | 71.73 | |
| NeRF-DFFSfM Reconstruction Time=20.52s, Per-Scene Reconstruction Time=1min2s2026.05 | 45.4 | 71.73 | |
| Feature-3DGSSfM Recon. Time=20.52s, Per-Scene Recon. Time=18mins2025.08 | 44.53 | 72.76 | |
| Feature-3DGSSfM Reconstruction Time=20.52s, Per-Scene Reconstruction Time=18min36s2026.05 | 44.53 | 72.76 |