Semantic Segmentation on S3DIS (Stanford Indoor Dataset)
77.1mIoUOurs
Evaluation results
| Method | Links | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Ours2025.07 | 77.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OmniVec2025.07 | 75.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Swin3D2025.07 | 74.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PTv22025.07 | 72.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Stratified Transformer2025.07 | 72 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PointTransformer+CBL2025.07 | 71.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PCCNlayers=16 continuous conv layers, input=raw xyz-intensity lidar points2021.01 | 58.27 | 67.01 | 92.26 | 96.2 | 75.89 | 0.27 | 5.98 | 69.49 | 63.45 | 66.87 | 65.63 | 47.28 | 68.91 | 59.1 | 46.22 | |
| SEGCloud2021.01 | 48.92 | 57.35 | 90.06 | 96.05 | 69.86 | 0 | 18.37 | 38.35 | 23.12 | 75.89 | 70.4 | 58.42 | 40.88 | 12.96 | 41.6 | |
| 3D-FCNresolution=0.2m, input=3D occupancy grid, backbone=ResNet-502021.01 | 47.46 | 54.91 | 90.17 | 96.48 | 70.16 | 0 | 11.4 | 33.36 | 21.12 | 76.12 | 70.07 | 57.89 | 37.46 | 11.16 | 41.61 | |
| PointNetpoint rotation layer=removed, trained from scratch=true2021.01 | 41.09 | 48.98 | 88.8 | 97.33 | 69.8 | 0.05 | 3.92 | 46.26 | 10.76 | 52.61 | 58.93 | 40.28 | 5.85 | 26.38 | 33.22 |