Sample Images

Samples a subset of images to separate sections of a dataset or create a testing split.

plantcv.parallel.sample_images(source, dest_path="./sampled_images", num=100)

returns None

  • Parameters:
    • source - Path to image directory as a string, path to a WorkflowConfig json file, or a WorkflowConfig object. A WorkflowConfig object/json file will only sample from images that the workflow would select.
    • dest_path - Path to write images to, note that the original subdirectory structure will be preserved.
    • num - Number of images to sample, defaults to 100.

Source Code: Here

Note

Sampling from a directory instead of from a configuration file makes some assumptions. If there is a SnapshotInfo.csv file then data is assumed to be from a Phenofront system. If there is a metadata.json file then a Phenodata system is assumed. Otherwise images are randomly sampled from all files with a recognized image extension ('bmp', 'dib', 'jpeg', 'jpg', 'jpe', 'jp2', 'png', 'ppm', 'pgm', 'ppm', 'sr', 'ras', 'tiff', or 'tif'). In general, sampling using a configuration file file is more useful as it will only sample from images that your workflow would run on.