Out-of-Distribution Detection on ImageNet-1k Near-OOD (val)
65.23FPR@95Energy Profile Divergence (EPD)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Energy Profile Divergence (EPD)Backbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 65.23 | 78.98 | |
| RMDSBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 65.38 | 80.09 | |
| MDSBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 66.11 | 79.04 | |
| KNNBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 70.47 | 74.11 | |
| GENBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 70.78 | 76.3 | |
| SHEBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 70.88 | 76.11 | |
| ViMBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 73.75 | 77.03 | |
| MSPBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 81.88 | 73.53 | |
| ReActBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 84.48 | 69.27 | |
| TempScaleBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 84.63 | 73.17 | |
| OpenMaxBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 88.78 | 73.64 | |
| ASHBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 94.45 | 53.2 | |
| GradNormBackbone=ViT-B/16, ID Accuracy=81.14%2026.04 | 94.72 | 53.68 |