Few-shot learning addresses the scarcity of expert-annotated images by enabling segmentation models to generalise from only a handful of labelled examples. Broadly, three families of approaches have ...