Showy mistletoes (Loranthaceae, Santalales) are obligate hemi-parasites, unable to survive without connecting to their host. Host use, therefore, is an important determinant of mistletoe diversity. An interesting feature of some showy mistletoes is the striking degree of phenotypic similarity between the mistletoe and its host. The similarity, which involves both vegetative and floral traits, has commonly been viewed as host mimicry. Other explanations, such as coincidence, are plausible, but the processes by which similarity is achieved remain unknown. Further, data needed to study the question have not been available. A limited amount of data have been available to characterize the degree and extent of phenotypic similarity, and species-level phylogenies currently are not available to evaluate this trait in a phylogenetic context. To characterize levels of phenotypic similarity within and among species, we collected image data from leaves of 152 mistletoe-host pairs representing 59 species pairs sampled from diverse habitats across Australia. Scanned images of approximately 9,000 individual leaves were input into a comprehensive phenotypic analysis and comparison that utilized deep learning and computer vision techniques to generate deep embedding vectors of the leaves. These vectors were then compared using cosine similarity, providing an intuitive and robust measure of similarity between mistletoes and their hosts. In this talk we will present results from our initial and current analyses.