Open-set Recognition on ImageNet-R and ImageNet-O (test)
43.76AccuracyEATA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| EATA2026.06 | 43.76 | 50.78 | 97.53 | 31.21 | |
| OSTTA2026.06 | 41.33 | 50.86 | 95.81 | 29.36 | |
| Tent2026.06 | 40.96 | 51.41 | 95.21 | 29.23 | |
| SAR(GN)Normalization=GN2026.06 | 40.81 | 26.27 | 99.65 | 18.63 | |
| UniEnt+(Tent)Base Method=Tent, variant=+2026.06 | 40.05 | 43.04 | 98.26 | 26.07 | |
| SAR(BN)Normalization=BN2026.06 | 39.48 | 45.96 | 96.66 | 26.42 | |
| UniEnt(Tent)Base Method=Tent2026.06 | 39.45 | 39.16 | 99.2 | 24.46 | |
| SoTTA2026.06 | 38.77 | 78.01 | 34.4 | 31.76 | |
| SAR(BN+Sigmoid)Normalization=BN, Output Function=Sigmoid2026.06 | 33.66 | 57.03 | 85.8 | 24.83 | |
| Tent(Sigmoid)Output Function=Sigmoid2026.06 | 33.29 | 56.61 | 86.26 | 24.44 |