Building Damage Assessment on xBD (test)
80.6F1 Score (Overall)BDANet
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| BDANetmode=full2021.05 | 80.6 | 92.5 | 61.6 | 78.8 | 87.6 | 86.4 | 78.2 | — | |
| Res50 (ours)Backbone=ResNet-50, Fusion=Cross-directional fusion, Data Augmentation=CutMix2020.10 | 80.4 | 92.7 | 61 | 78.1 | 87.3 | 86.4 | 77.8 | — | |
| Res50 (baseline)Backbone=ResNet-50, Fusion=Vanilla U-Net2020.10 | 78.9 | 92.3 | 57.8 | 76 | 86.9 | 86.4 | 75.7 | — | |
| Vanilla network (Ours)mode=vanilla2021.05 | 78.9 | 92.3 | 57.8 | 76 | 86.9 | 86.4 | 75.7 | — | |
| DeepLabv32021.05 | 78.8 | 91.9 | 58 | 76.1 | 85.9 | 86.4 | 75.6 | — | |
| SegNet2021.05 | 78.2 | 92.1 | 56.7 | 74.6 | 85.9 | 86.4 | 74.7 | — | |
| Weber et al.Backbone=ResNet-502020.10 | 77 | 88.5 | 56.3 | 77.1 | 80.8 | 84 | 74 | — | |
| Weber et al.year=20202021.05 | 77 | 88.5 | 56.3 | 77.1 | 80.8 | 84 | 74 | — | |
| FCN2021.05 | 76.5 | 91.9 | 53.2 | 70.8 | 86.1 | 86.4 | 72.2 | — | |
| Our modelprocessing=4 512 by 512 quadrants, temporal_setting=multi-temporal, architecture=R-CNN base with shared weights2020.04 | 74.1 | 90.6 | 49.3 | 72.2 | 83.7 | — | — | — | |
| Gupta et al.Backbone=ResNet-502020.10 | 74.1 | 90.6 | 49.3 | 72.2 | 83.7 | 83.5 | 69.7 | — | |
| RescueNet2021.05 | 74.1 | 90.6 | 49.3 | 72.2 | 83.7 | 83.5 | 69.7 | — | |
| U-Net++2021.05 | 74 | 88.6 | 52.3 | 68.9 | 85.8 | 81.9 | 70.7 | — | |
| WNet2021.05 | 73.7 | 88.4 | 51.8 | 68.4 | 85.5 | 81.7 | 70.3 | — | |
| xBD Baseline2020.04 | 26.5 | 66.3 | 14.4 | 0.9 | 46.6 | — | — | — |