Disaster Damage Classification on Street-view Bi-Temporal Images (val) benchmark leaderboard · SOTA2 Research

Disaster Damage Classification on Street-view Bi-Temporal Images (val)

77.11AccuracyFeature-fusion ConvNext

Evaluation Results

Method
Feature-fusion ConvNext2025.04	77.11	77.51	77.1	76.78	59.39	82.72	84.44
Dual-ConvNeXt+ Cross-Attention2025.04	76.91	77.04	76.91	76.72	71.08	72.36	91.37
Dual-Swin Transformer + Cross-Attention2025.04	76.91	79.7	76.91	76.15	51.47	91.97	74.6
Siamese ConvNeXt2025.04	75.54	78.78	75.54	74.98	53.28	90.79	71.11
Dual-Swin Transformer2025.04	74.17	74.2	74.17	74.08	62.69	75.45	81.33
Baseline-ConvNext2025.04	72.8	73.15	72.8	72.56	56.91	79.12	75.54
Baseline-Swin Transformer2025.04	66.14	68.89	66.14	64.81	39.02	84.34	57.55