GCAD

Reported on 6 benchmarks across 1 task

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

Methodology6 results

Anomaly DetectiononVisA
Detection AUROC
89.1
best: 99.8 (UniNet)
Anomaly DetectiononVisA
Segmentation AUPRO (until 30% FPR)
83.7
best: 96.1 (AnomalyDINO-S (full-shot))
Anomaly DetectiononMVTec LOCO AD
Avg. Detection AUROC
83.3
best: 95.3 (CSAD)
Anomaly DetectiononMVTec LOCO AD
Detection AUROC (only logical)
86
best: 98.1 (PSAD)
Anomaly DetectiononMVTec LOCO AD
Detection AUROC (only structural)
80.6
best: 95.9 (PUAD-M)
Anomaly DetectiononMVTec LOCO AD
Segmentation AU-sPRO (until FPR 5%)
70.1
best: 83.2 (SAM-LAD)