For your convenience, we provide a download link for the whole dataset, as well as links for each object category. For each object, the data consist of four folders:
'train', which contains the anomaly-free training images.
'validation', which contains the anomaly-free validation images.
'test', which contains the test images in three subdirectories: 'good', logical_anomalies', and 'structural_anomalies'.
'ground_truth', which contains the pixel-precise annotations of anomalous regions which are represented as a 1-channel PNG image.
The file 'defects_config.json' contains the mapping between defect names and pixel values as well as parameters used for evaluation.
