aristrops/mvtec-concepts

MVTec AD with concept annotations

Concept-annotated version of MVTec AD, including generated anomalies.

Structure

<category>/
├── train/
├── test/
├── ground_truth/
├── generated_anomalies/
└── concepts/
    ├── <category>_dataset.csv              # concepts extracted on real images
    ├── <category>_dataset_gen.csv          # concepts extracted on generated images
    └── <category>_dataset_gen_to_real.csv  # concepts extracted on generated images and annotated on real ones (semi-supevised training and evaluation of unsupervised)

CSV paths are relative to the CSV's own folder.

Usage

Download the dataset with the Hugging Face CLI:

hf download aristrops/mvtec-concepts --repo-type dataset --local-dir mvtec-concepts

To download a single category only:

hf download aristrops/mvtec-concepts --repo-type dataset --local-dir mvtec-concepts --include "bottle/*"

Citations

If you use this dataset in a scientific work, please cite:

• Paul Bergmann, Michael Fauser, David Sattlegger, and Carsten Steger, "A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection", IEEE Conference on Computer Vision and Pattern Recognition, 2019

• Arianna Stropeni, Valentina Zaccaria, Francesco Borsatti, Davide Dalle Pezze, Manuel Barusco, and Gian Antonio Susto, "Explainable Visual Anomaly Detection via Concept Bottleneck Models." arXiv preprint arXiv:2511.20088, 2025.