Fine-Grained Visual Classification on Jekyll (test)
59.63AccuracyEKDC-Net (HERBS)
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| EKDC-Net (HERBS)Backbone=HERBS, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 59.63 | 82.14 | 60.11 | 67.01 | 81.69 | 59.63 | 66.48 | |
| EKDC-Net (CGL)Backbone=CGL, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 58.49 | 83.34 | 60.09 | 65.07 | 81.54 | 58.49 | 62.66 | |
| HERBSPre-trained=CU-Tree102, Fine-tuning=none2026.01 | 52.98 | 80.56 | 52.79 | 60.04 | 80.14 | 52.98 | 59.75 | |
| CGLPre-trained=CU-Tree102, Fine-tuning=none2026.01 | 51.83 | 81.08 | 50.22 | 58.04 | 79.84 | 51.83 | 59.13 | |
| EKDC-Net (Swin-Base)Backbone=Swin-Base, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 44.27 | 69.79 | 44.43 | 49.28 | 68.37 | 44.27 | 48.4 | |
| EKDC-Net (ResNet-50)Backbone=ResNet-50, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 39.22 | 66.81 | 39.08 | 42.83 | 63.65 | 39.22 | 41.56 | |
| EKDC-Net (ViT-Base)Backbone=ViT-Base, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 35.32 | 62.79 | 35.27 | 38.02 | 62.37 | 35.32 | 36.74 | |
| Swin-BaseBackbone=Swin-Base, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 30.05 | 56.09 | 28.39 | 30.8 | 55.52 | 30.05 | 31.4 | |
| ResNet-50Backbone=ResNet-50, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 26.83 | 58.84 | 27.17 | 31.03 | 58.48 | 26.83 | 29.99 | |
| ViT-BaseBackbone=ViT-Base, Pre-trained=CU-Tree102, Fine-tuning=none2026.01 | 26.38 | 46.41 | 25.8 | 28.88 | 44.63 | 26.38 | 28.43 |