Anomaly Detection2022en

Visual Anomaly Dataset

10,821 high-resolution images across 12 objects with complex structures. More challenging than MVTec with realistic industrial scenarios.

Metrics:auroc, pixel-auroc, ap, f1-max
Paper / WebsiteDownload
Current State of the Art

SimpleNet

Research

95.5

auroc

auroc Progress Over Time

Showing 2 breakthroughs from Nov 2022 to Mar 2023

91.892.893.894.895.8Nov 2022Mar 2023aurocDate

Key Milestones

Nov 2022
PatchCore

VisA is more challenging than MVTec.

92.1
Source
Mar 2023
SimpleNetCurrent SOTA

Strong generalization to VisA.

95.5
+3.7%
Source
Total Improvement
3.7%
Time Span
4m
Breakthroughs
2
Current SOTA
95.5

Top Models Performance Comparison

Top 3 models ranked by auroc

auroc1SimpleNet95.5100.0%2EfficientAD94.899.3%3PatchCore92.196.4%0%25%50%75%100%% of best
Best Score
95.5
Top Model
SimpleNet
Models Compared
3
Score Range
3.4

aurocPrimary

#ModelScorePaper / CodeDate
1
SimpleNetOpen Source
Research		95.5
research-paper
Dec 2025
2
EfficientADOpen Source
Research		94.8
research-paper
Dec 2025
3
PatchCoreOpen Source
Amazon		92.1
research-paper
Dec 2025

Other Anomaly Detection Datasets

MVTec ADMVTec 3D-ADKolektorSDD2Weld Defect X-RaySeverstal Steel DefectNEU-DET
VisA Benchmark - Anomaly Detection | CodeSOTA