Datamodules#
Anomalib provides various datamodules for different types of data modalities. These datamodules are organized into three main categories:
Image Datamodules#
BMAD dataset datamodule for medical anomaly detection.
BTech dataset datamodule for surface defect detection.
Datumaro format datamodule (compatible with Intel Geti™).
Custom folder-based datamodule for organizing your own image dataset.
Kolektor Surface-Defect dataset datamodule.
MVTec AD dataset datamodule for unsupervised anomaly detection.
MVTec AD 2 dataset datamodule for anomaly detection with natural images.
MVTec LOCO dataset datamodule for logical and structural anomaly detection.
Real-IAD dataset datamodule for industrial anomaly detection with real-world scenarios.
Custom tabular datamodule for datasets with image paths and labels in tabular format.
Valeo Anomaly Detection dataset datamodule for automotive anomaly detection.
Visual Anomaly (VisA) dataset datamodule.
Video Datamodules#
CUHK Avenue dataset datamodule for video anomaly detection.
ShanghaiTech dataset datamodule for video anomaly detection.
UCSD Pedestrian dataset datamodule for video anomaly detection.
Depth Datamodules#
MVTec 3D-AD dataset datamodule for unsupervised 3D anomaly detection and localization.
Custom folder-based 3D datamodule for organizing your own depth-based anomaly detection dataset.