Semantic segmentation on S3DIS (val)
73.4mIoUPTv3
Evaluation results
| Method | Links | |
|---|---|---|
| PTv3Backbone=PTv3, Pre-training=None2025.06 | 73.4 | |
| PonderV2Backbone=SparseUNet, Pre-training=PonderV22025.06 | 73.2 | |
| CSCBackbone=SparseUNet, Pre-training=CSC2025.06 | 72.2 | |
| STBackbone=ST, Pre-training=None2025.06 | 72 | |
| PPT(Unsup.)Backbone=SparseUNet, Pre-training=PPT(Unsup.)2025.06 | 71.9 | |
| PTv2Backbone=PTv2, Pre-training=None2025.06 | 71.6 | |
| UniPre3DBackbone=SparseUNet, Pre-training=UniPre3D2025.06 | 71.5 | |
| PointNextBackbone=PointNext, Pre-training=None2025.06 | 70.5 | |
| PTv1Backbone=PTv1, Pre-training=None2025.06 | 70.4 | |
| PCBackbone=SparseUNet, Pre-training=PC2025.06 | 70.3 | |
| PonderV2+Backbone=SparseUNet, Pre-training=PonderV2† (Single-dataset)2025.06 | 70.2 | |
| MSCBackbone=SparseUNet, Pre-training=MSC2025.06 | 70.1 | |
| SparseUNetBackbone=SparseUNet, Pre-training=None2025.06 | 65.4 | |
| UniPre3DBackbone=Standard Transformer, Pre-training=UniPre3D2025.06 | 62 | |
| PCP-MAEBackbone=Standard Transformer, Pre-training=PCP-MAE2025.06 | 61.3 | |
| ACTBackbone=Standard Transformer, Pre-training=ACT2025.06 | 61.2 | |
| Point-MAEBackbone=Standard Transformer, Pre-training=Point-MAE2025.06 | 60.8 | |
| Standard TransformerBackbone=Standard Transformer, Pre-training=None2025.06 | 60 | |
| PointNet++Backbone=PointNet++, Pre-training=None2025.06 | 53.5 | |
| PointNetBackbone=PointNet, Pre-training=None2025.06 | 41.1 |