3D Semantic Segmentation on ScanNet++ 100-class benchmark 51
20.9mIoURelGraphOV
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RelGraphOVEvaluation Mode=Zero-Shot, Training Dataset=ScanNetV2, Fine-tuned=false2026.07 | 20.9 | 33.2 | |
| OpenScene-2DEvaluation Mode=Zero-Shot, Training Dataset=ScanNetV2, Fine-tuned=false2026.07 | 13.3 | 20 | |
| OpenScene-2D3DEvaluation Mode=Zero-Shot, Training Dataset=ScanNetV2, Fine-tuned=false2026.07 | 11.7 | 15.3 | |
| CUA-O3DEvaluation Mode=Zero-Shot, Training Dataset=ScanNetV2, Fine-tuned=false2026.07 | 9.3 | 13.7 | |
| OpenScene-3DEvaluation Mode=Zero-Shot, Training Dataset=ScanNetV2, Fine-tuned=false2026.07 | 8.9 | 13.6 |