Image Classification on ImageNet-C L5 (Imbalanced Labels)
63.3AccuracyDualTTA
Evaluation results
| Method | Links | |
|---|---|---|
| DualTTABackbone=ViTBase-LN2026.04 | 63.3 | |
| SARBackbone=ViTBase-LN2026.04 | 62.1 | |
| DeYOBackbone=ViTBase-LN2026.04 | 58.68 | |
| EATABackbone=ViTBase-LN2026.04 | 53.6 | |
| TentBackbone=ViTBase-LN2026.04 | 52.47 | |
| EATABackbone=ResNet-BN2026.04 | 47.14 | |
| SARBackbone=ResNet-BN2026.04 | 46.27 | |
| SARBackbone=ResNet-GN2026.04 | 45.67 | |
| DualTTABackbone=ResNet-BN2026.04 | 43.83 | |
| TentBackbone=ResNet-BN2026.04 | 42.8 | |
| DeYOBackbone=ResNet-GN2026.04 | 41.54 | |
| DeYOBackbone=ResNet-BN2026.04 | 41.03 | |
| No adaptBackbone=ViTBase-LN2026.04 | 39.83 | |
| DualTTABackbone=ResNet-GN2026.04 | 38.88 | |
| EATABackbone=ResNet-GN2026.04 | 34.86 | |
| No adaptBackbone=ResNet-BN2026.04 | 32.11 | |
| No adaptBackbone=ResNet-GN2026.04 | 31.44 | |
| TentBackbone=ResNet-GN2026.04 | 29.08 |