[data request] MVTec AD : Image Anomaly Detection

[innat]

• Name of dataset: MVTec AD
• URL of dataset: https://www.mvtec.com/company/research/datasets/mvtec-ad
• License of dataset: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0)
• Dataset Paper: CVPR-LINK.
• Short description of the dataset and use case(s): MVTec AD is a dataset for benchmarking anomaly detection methods focusing on industrial inspection. It contains over 5000 high-resolution images divided into fifteen object and texture categories. Each category comprises a set of defect-free training images and a test set of images with various kinds of defects and images without defects.

---

Folks who would also like to see this dataset in tensorflow/datasets, please thumbs up so the developers can know which requests to prioritize.

(above gif from here, inference output of a model on this dataset.)

Note

• AFAIK, this dataset is one of the most well-used and high-quality datasets for this task.
• AFAIK, there is no similar or equivalent dataset available in tensorflow/dataset.
• AFAIK, the overall goal (image anomaly detection) is less practiced with tensorflow due to limited resources. ( Mostly contained auto-encoder modeling stuff. )
• AFAIK, all SOTA models (in PyTorch) are primarily benchmarked on this dataset. SOURCE.