Robustness Evaluation on ImageNet-A (test)
22.4Top-1 AccuracyMixPro-DeiT-S
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MixPro-DeiT-SBackbone=DeiT-S, Augmentation=MixPro2023.04 | 22.4 | 30.3 | 32.4 | |
| TransMix-DeiT-SBackbone=DeiT-S, Augmentation=TransMix2023.04 | 21.1 | 31.2 | 28.8 | |
| DeiT-SBackbone=DeiT-S2023.04 | 19.1 | 32 | 23.8 |