Out-of-Distribution Detection on ImageNet-1k Far-OOD (val)
24.77FPR@95Energy Profile Divergence (EPD)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Energy Profile Divergence (EPD)Backbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 24.77 | 92.78 | |
| SHEBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 27.14 | 92.42 | |
| RMDSBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 28.75 | 92.6 | |
| ViMBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 29.2 | 92.84 | |
| MDSBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 29.97 | 92.6 | |
| KNNBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 31.93 | 90.81 | |
| GENBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 32.23 | 91.35 | |
| MSPBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 51.66 | 86.04 | |
| TempScaleBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 53.77 | 86.32 | |
| ReActBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 53.92 | 85.69 | |
| OpenMaxBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 55.49 | 89.27 | |
| GradNormBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 92.64 | 41.74 | |
| ASHBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 96.77 | 51.56 |