Anomaly Detection on VisA 15% training contamination (test) benchmark leaderboard · SOTA2 Research

Anomaly Detection on VisA 15% training contamination (test)

0.743Capsules AUROCADL

Evaluation Results

Method
ADL2026.01	0.743	0.904	0.899	0.921	0.739	0.596	0.927	0.911	0.816	0.804	0.81	0.863	0.828
Multi-Cue Anomaly Detection framework2026.01	0.733	0.971	0.967	0.964	0.948	0.726	0.952	0.967	0.882	0.886	0.77	0.937	0.892
DestSeg2026.01	0.631	0.909	0.839	0.834	0.737	0.631	0.831	0.911	0.841	0.765	0.895	0.961	0.815
LOE2026.01	0.575	0.845	0.929	0.959	0.658	0.505	0.748	0.809	0.795	0.787	0.759	0.931	0.775
PatchCore2026.01	0.488	0.741	0.755	0.585	0.682	0.537	0.714	0.631	0.729	0.709	0.651	0.88	0.675
DRÆM2026.01	0.441	0.733	0.311	0.753	0.847	0.709	0.665	0.945	0.144	0.195	0.604	0.757	0.592
PaDiM2026.01	0.13	0.786	0.791	0.824	0.632	0.115	0.651	0.798	0.728	0.653	0.633	0.846	0.632