Anomaly Detection On Road Anomaly
Métriques
AP
FPR95
Résultats
Résultats de performance de divers modèles sur ce benchmark
Tableau comparatif
| Nom du modèle | AP | FPR95 |
|---|---|---|
| unmasking-anomalies-in-road-scene | 79.70 | 13.45 |
| pixels-together-strong-segmenting-unknown | 90.28 | 4.92 |
| pixel-wise-anomaly-detection-in-complex | 41.83 | 59.72 |
| diffusion-for-out-of-distribution-detection | 89.1 | 8.8 |
| standardized-max-logits-a-simple-yet | 25.82 | 49.74 |
| far-away-in-the-deep-space-nearest-neighbor | 85.6 | 9.8 |
| pixel-wise-energy-biased-abstention-learning | 45.10 | 44.58 |
| synthesize-then-compare-detecting-failures | 24.86 | 64.69 |
| residual-pattern-learning-for-pixel-wise-out | 71.61 | 17.74 |
| Modèle 10 | 95.21 | 2.11 |