OOD Detection on Far-OOD (test)
0.9694AUROCSCALE*
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SCALE*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 0.9694 | 0.1318 | |
| fDBD*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 0.9692 | 0.1263 | |
| ReAct*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 0.9678 | 0.1374 | |
| GEN*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 0.9673 | 0.14 | |
| MSP*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 0.9668 | 0.1373 | |
| KNN*Image Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=true2025.04 | 0.966 | 0.1519 | |
| fDBDImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 0.9319 | 0.2384 | |
| RMDSBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.9298 | 0.2933 | |
| KNNImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 0.9296 | 0.2427 | |
| OursBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.9221 | 0.2697 | |
| MDSBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.9146 | 0.3075 | |
| GENImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 0.9135 | 0.3473 | |
| GENBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.91 | 0.3276 | |
| MSPImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 0.9073 | 0.3172 | |
| ReActImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 0.9042 | 0.449 | |
| SHEBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.8989 | 0.4582 | |
| MSPBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.8965 | 0.4932 | |
| KNNBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.8936 | 0.3441 | |
| ReActBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.8899 | 0.4262 | |
| OpenMaxBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.8854 | 0.5228 | |
| TempScaleBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.8675 | 0.5001 | |
| SCALEImage Encoder=ResNet18, ID Dataset=CIFAR-10, ELogitNorm enhancement=false2025.04 | 0.8646 | 0.6749 | |
| ASHBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.4434 | 0.9478 | |
| GradNormBackbone=Swin Transformer, In-Distribution Accuracy=81.60%2026.04 | 0.3568 | 0.966 |