Automatically converting in-memory representations of images using a knowledge graph of type description.
