Testing methods

Annotation consistency

Compare class labels and annotation counts on exact duplicate images.

Data quality

Check that annotations can be read and refer to existing images and classes.

Data quality

Find very wide or very tall images.

Data quality

Find images that lack sharp edges.

Data quality

Compare how many images contain each declared class.

Data quality

Find images with a narrow range of tones.

Data quality

Find exact and near-duplicate images, including copies across dataset splits.

Data quality

Find images that are mostly dark or washed out.

Data quality

Confirm that every image file opens and decodes completely.

Data quality

Check that each file's extension matches its decoded image format.

Data quality

Find small images and image sizes that differ from the rest of the dataset.

Data quality

Check whether each populated dataset split contains every declared class.

Data quality

Find images with no annotations or missing label files.

Data quality

Validate that each imported row has a target value for scoring.

Data quality

Validate imported columns against the selected tabular task.

Data quality

1–15 of 15 results