MVTec AD
Unknown
5,354 high-resolution images across 15 object and texture categories. The gold standard for industrial anomaly detection with pixel-level annotations.
Benchmark Stats
Models8
Papers8
Metrics1
SOTA History
Coming SoonVisual timeline of state-of-the-art progression over time will appear here.
auroc
auroc
Higher is better
| Rank | Model | Code | Score | Paper / Source |
|---|---|---|---|---|
| 1 | simplenet State-of-the-art on MVTec AD. Simple architecture. | 99.6 | research-paper | |
| 2 | fastflow 2D normalizing flow. Good speed-accuracy tradeoff. | 99.4 | research-paper | |
| 3 | patchcore Image-level AUROC. State-of-the-art with WideResNet-50 backbone. | 99.1 | research-paper | |
| 4 | efficientad Matches PatchCore with 614 FPS inference. | 99.1 | research-paper | |
| 5 | reverse-distillation Knowledge distillation approach. | 98.5 | research-paper | |
| 6 | cflow-ad Conditional normalizing flows. | 98.3 | research-paper | |
| 7 | draem Discriminatively trained reconstruction. | 98 | research-paper | |
| 8 | padim ResNet-18 backbone. Fast inference. | 97.9 | research-paper |