Image Classification on EuroSAT (Adversarial Robustness)
42.66Clean AccuracyCLIP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CLIP2025.12 | 42.66 | 0 | 0.01 | 0.17 | |
| FAREFine-tuning dataset=TinyImageNet, Adversarial training=2-step PGD, Epsilon=1/2552025.12 | 24.19 | 0.08 | 0.12 | 0.07 | |
| PMG-AFTFine-tuning dataset=TinyImageNet, Adversarial training=2-step PGD, Epsilon=1/2552025.12 | 21.76 | 11.49 | 8.51 | 7.17 | |
| TGA-ZSRFine-tuning dataset=TinyImageNet, Adversarial training=2-step PGD, Epsilon=1/2552025.12 | 19.7 | 11.75 | 11.56 | 11.26 | |
| UCATFine-tuning dataset=TinyImageNet, Adversarial training=2-step PGD, Epsilon=1/2552025.12 | 19.29 | 2.99 | 2.41 | 2.26 | |
| TeCoAFine-tuning dataset=TinyImageNet, Adversarial training=2-step PGD, Epsilon=1/2552025.12 | 16.13 | 11.88 | 11.63 | 11.42 |