Image Classification on ImageNet-C with Places365-C outliers 1.0 (test)
46.9AccuracySTAMP + Hi-Vec
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| STAMP + Hi-VecBackbone=ResNet-502025.08 | 46.9 | 77.9 | 58.5 | |
| EATABackbone=ResNet-502025.08 | 46.4 | 72.6 | 56 | |
| STAMPBackbone=ResNet-502025.08 | 46.4 | 77.7 | 57.6 | |
| SAR + Hi-VecBackbone=ResNet-502025.08 | 45.3 | 73.8 | 56.1 | |
| SARBackbone=ResNet-502025.08 | 44.9 | 73.3 | 55 | |
| SoTTABackbone=ResNet-502025.08 | 41.7 | 67.8 | 50.7 | |
| RoTTABackbone=ResNet-502025.08 | 36.6 | 68.6 | 46.5 | |
| Tent + Hi-VecBackbone=ResNet-502025.08 | 35.4 | 52 | 42.1 | |
| TentBackbone=ResNet-502025.08 | 34.9 | 51.8 | 39.5 | |
| CoTTABackbone=ResNet-502025.08 | 33.8 | 66.9 | 43.5 | |
| BN StatsBackbone=ResNet-502025.08 | 31.1 | 67.7 | 41.1 | |
| MRLBackbone=ResNet-502025.08 | 18.4 | 61.9 | 28.3 | |
| SourceBackbone=ResNet-502025.08 | 18.2 | 61.6 | 26.1 | |
| OWTTTBackbone=ResNet-502025.08 | 9.1 | 54 | 13.9 |