Out-of-Distribution Detection on ImageNet-R (AUROC, AUPR, FPR95)
91.17AUROCSupervised classifier
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Supervised classifierbackbone=ResNet502026.05 | 91.17 | 94.09 | 43.71 | |
| S-PUNAbackbone=ResNet50, classifier_usage=with classifier2026.05 | 88.76 | 91.9 | 56.08 | |
| S-PUNAbackbone=ResNet50, classifier_usage=without classifier2026.05 | 86.21 | 90.87 | 59.57 | |
| LaGAMbackbone=ResNet502026.05 | 80.92 | 84.35 | 82.03 | |
| Dist-PUbackbone=ResNet502026.05 | 74.54 | 80.36 | 86.46 | |
| Dc-PUbackbone=ResNet502026.05 | 70.99 | 78.02 | 88.22 | |
| KNNbackbone=ResNet502026.05 | 69.88 | 77.43 | 88.63 | |
| saPUbackbone=ResNet502026.05 | 68.82 | 73.16 | 95.06 |