Semantic Segmentation on ScanNet 5cm voxel size (cross-sequence)
52mIoUIRGI
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| IRGIvoxel_size=5cm, prompting_strategy=grid + informed2026.02 | 52 | — | 7.9 | |
| IRGvoxel_size=5cm, prompting_strategy=grid2026.02 | 51.7 | — | 7 | |
| IRIvoxel_size=5cm, prompting_strategy=informed2026.02 | 51.6 | — | 7.1 | |
| MCvoxel_size=5cm, model_description=multi-view consistency baseline2026.02 | 48.5 | 0.4 | — | |
| PTvoxel_size=5cm, model_description=pre-trained model2026.02 | 48 | — | — |