Instance Segmentation on FOR-instance (SCION, New Zealand) V2
97.4Precision (P)ForestFormer3D
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ForestFormer3D2026.06 | 97.4 | 86.1 | 91.4 | 80 | |
| ForestMamba2026.06 | 96.9 | 79.1 | 87 | 77.2 | |
| ForestFormer3D2026.06 | 92.4 | 75 | 82.8 | 81.2 | |
| SaTv2Training dataset=FOR-inst. train v32026.06 | 90.5 | 80.2 | 85 | 90.7 | |
| OneFormer3D2026.06 | 89.8 | 81.4 | 85.2 | 74.6 | |
| ForAINetV2variant=R162026.06 | 88.1 | 59.2 | 70.8 | 72.7 | |
| SaTv2Training dataset=FOR-inst. train v22026.06 | 87.3 | 79.8 | 83.4 | 89.6 | |
| ForAINetV2variant=R82026.06 | 84.3 | 63.4 | 72.4 | 73.3 | |
| TreeLearn2026.06 | 82 | 36.6 | 50.6 | 52.2 | |
| Oneformer3D2026.06 | 72 | 74.3 | 73.1 | 80 |