Anomaly Detection on MVTec AD (20% Training Contamination Test) benchmark leaderboard · SOTA2 Research

Anomaly Detection on MVTec AD (20% Training Contamination Test)

0.989AUROC (carpet)Multi-Cue Anomaly Detection framework

Evaluation Results

Method
Multi-Cue Anomaly Detection framework2026.01	0.989	0.998	0.853	0.885	0.999	0.95	0.874	0.838	0.942	0.825	0.917	0.977	0.994	0.979	0.917	0.929
LOE2026.01	0.922	0.979	0.813	0.765	0.871	0.895	0.835	0.889	0.991	0.532	0.714	0.709	0.972	0.941	0.411	0.816
ADL2026.01	0.887	0.999	0.737	0.825	0.952	0.997	0.721	0.767	0.921	0.777	0.869	0.865	0.988	0.972	0.825	0.873
PaDiM2026.01	0.806	0.661	0.344	0.703	0.643	0.485	0.196	0.489	0.367	0.293	0.504	0.409	0.935	0.314	0.303	0.497
DestSeg2026.01	0.758	1	0.418	0.937	0.998	0.895	0.581	0.811	0.871	0.847	0.915	0.858	1	0.976	0.994	0.857
DRÆM2026.01	0.731	0.949	0.497	0.889	0.99	0.641	0.294	0.789	0.805	0.801	0.872	0.825	1	0.875	0.795	0.784
PatchCore2026.01	0.662	0.757	0.737	0.781	0.789	0.567	0.701	0.732	0.741	0.678	0.671	0.679	0.685	0.619	0.521	0.688