Anomaly Segmentation on SMIYC-RA21 (val)
95.1AuPRC (%)VL-Anomaly
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| VL-Anomalybackbone=ResNet-50, CLIP backbone=ViT-L-16, layers=L=12026.03 | 95.1 | 3.5 | 99.7 | |
| Mask2Anomaly2023.07 | 94.5 | 3.3 | — | |
| ODINreproduced=true, temperature=3.02026.03 | 91.5 | 3.6 | 98.5 | |
| Mask2Anomalyreproduced=true, refine mask stage=excluded2026.03 | 88.7 | 14.6 | 99.7 | |
| Max Entropy2026.03 | 85.5 | 15 | — | |
| Maximized Entropy2023.07 | 80.7 | 17.4 | — | |
| Dense Hybrid2026.03 | 78 | 9.8 | — | |
| Image Resynthesis2023.07 | 76.4 | 20.5 | — | |
| SynBoost2023.07 | 68.8 | 30.9 | — | |
| SynBoost2026.03 | 56.4 | 61.9 | — | |
| Learning Embedding2023.07 | 51.9 | 60 | — | |
| PEBEL2026.03 | 49.1 | 40.8 | — | |
| SML2026.03 | 46.8 | 39.5 | — | |
| ODIN2023.07 | 46.3 | 61.5 | — | |
| Max Softmax2023.07 | 40.4 | 60.2 | — | |
| Void Classifier2023.07 | 39.3 | 66.1 | — | |
| JSRNet2026.03 | 33.6 | 43.9 | — | |
| MC Dropout2023.07 | 29.2 | 77.9 | — | |
| MSP2026.03 | 28 | 72 | — | |
| Mahalanobis2023.07 | 22.5 | 86.4 | — | |
| Mahalanobis2026.03 | 20 | 87 | — | |
| Ensemble2023.07 | 16 | 80 | — |