DDAD

Reported on 5 benchmarks across 1 task · 1 paper · 1 SOTA

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

Methodology5 results

Anomaly DetectiononVisA
Detection AUROC· 2023-05-25
98.9
best: 99.8 (UniNet)
SOTA
Anomaly Detection with Conditioned Denoising Diffusion Models arXiv:2305.15956
Anomaly DetectiononMVTec AD
Detection AUROC· uses extra data· 2023-05-25
99.8
best: 99.9 (GLASS)
Anomaly Detection with Conditioned Denoising Diffusion Models arXiv:2305.15956
Anomaly DetectiononMVTec AD
Segmentation AUROC· uses extra data· 2023-05-25
98.1
best: 99.7 (WeakREST-Block)
Anomaly Detection with Conditioned Denoising Diffusion Models arXiv:2305.15956
Anomaly DetectiononVisA
Segmentation AUPRO (until 30% FPR)· 2023-05-25
92.7
best: 96.1 (AnomalyDINO-S (full-shot))
Anomaly Detection with Conditioned Denoising Diffusion Models arXiv:2305.15956
Anomaly DetectiononVisA
Segmentation AUROC· 2023-05-25
97.6
best: 99.1 (Dinomaly ViT-L (model-unified multi-class))
Anomaly Detection with Conditioned Denoising Diffusion Models arXiv:2305.15956