Data Bite
MVTec AD
A real-world industrial anomaly detection dataset with defect-free training images, anomalous test images, and pixel-level annotations.
Industrial inspection
Unsupervised anomaly detection
CC BY-NC-SA 4.0
MVTec AD is one of the standard datasets for industrial visual anomaly detection. It includes normal training images and test images with different defect types, plus pixel-precise anomaly masks for localization evaluation.
It is especially useful when you want to test whether a model can learn normal appearance from defect-free data and then detect unusual regions at test time. The object and texture categories also make it a practical starting point for comparing anomaly detection methods.
Dataset facts
- Use for: unsupervised anomaly detection, anomaly localization, industrial visual inspection baselines
- Scale: Over 5,000 high-resolution images across 15 object and texture categories
- Watch out: Commercial use is not allowed under the standard dataset license; contact MVTec if the use case may be commercial.