Semantic Segmentation on ScanNet v2 (val) (Limited Reconstruction/Annotation mIoU)
31.3mIoU (Limited Reconstruction 1%)PTv3
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| PTv3Training Strategy=Multi-dataset joint training (PPT)2023.12 | 31.3 | 52.6 | 63.3 | 68.2 | 62.4 | 69.1 | 74.3 | 75.5 | |
| MinkUNetTraining Strategy=Multi-dataset joint training (PPT)2023.12 | 26 | 47.8 | 56.7 | 62.9 | 41.9 | 53.9 | 62.2 | 65.5 | |
| PTv3Training Strategy=Trained from scratch2023.12 | 25.8 | 48.9 | 61 | 67 | 60.1 | 67.9 | 71.4 | 72.7 | |
| PTv2Training Strategy=Multi-dataset joint training (PPT)2023.12 | 24.8 | 48.1 | 59.8 | 66.3 | 58.4 | 66.1 | 70.3 | 71.2 |