In this paper, we show that artificial neural networks are able to recognize the genus or family of a plant species, although this species has never been seen during the learning process. The network generalizes the visual characteristics of other plant species and conceptualizes the respective genus or family. The best results are obtained if the network generalizes across many different plant species. In the above picture we have shown how our network has recognized each genus or family. We were excited to demonstrate that sometimes the leaves, sometimes the flower are relevant for correct classification. Hence, we use both a picture of the leaf as well as the flower to achieve the most accurate determination of herbaceous plants in our Flora Incognita App.
Seeland, M., Rzanny, M., Boho, D., Wäldchen, J. & Mäder, P. (2019). Image-based classification of plant genus and family for trained and untrained plant species. BMC Bioinformatics 20:4 https://doi.org/10.1186/s12859-018-2474-x