UniNet

Reported on 23 benchmarks across 5 tasks

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology13 results

Anomaly DetectiononMVTec 3D-AD (RGB)
Detection AUROC
95.76
Anomaly DetectiononMVTec 3D-AD (RGB)
Segmentation AUPRO
95.55
best: 96.9 (CPR)
Anomaly DetectiononUCSD Ped2
AUC
97.9
best: 99.1 (AI-VAD)
Anomaly DetectiononBTAD
Detection AUROC
97.73
best: 98.3 (CPR)
Anomaly DetectiononBTAD
Segmentation AUPRO
80.01
best: 97.5 (ReConPatch WRN-50)
Anomaly DetectiononBTAD
Segmentation AUROC
97.7
best: 99.3 (WeakREST-Block)
Anomaly DetectiononMVTec AD
Detection AUROC
99.9
Anomaly DetectiononMVTec AD
Segmentation AUPRO
96
best: 98.4 (WeakREST-Block)
Anomaly DetectiononMVTec AD
Segmentation AUROC
98.81
best: 99.7 (WeakREST-Block)
Anomaly DetectiononVisA
Detection AUROC
99.8
Anomaly DetectiononVisA
Segmentation AUPRO
93.9
best: 96 (DiffusionAD)
Anomaly DetectiononVisA
Segmentation AUPRO (until 30% FPR)
93.9
best: 96.1 (AnomalyDINO-S (full-shot))
Anomaly DetectiononVisA
Segmentation AUROC
98.8
best: 99.1 (Dinomaly ViT-L (model-unified multi-class))