Rove-Tree-11: The not-so-Wild Rover, A hierarchically structured image dataset for deep metric learning research

Roberta Hunt, Kim Steenstrup Pedersen; Proceedings of the Asian Conference on Computer Vision (ACCV), 2022, pp. 2967-2983

Abstract


We present a new dataset of images of pinned insects from museum collections along with a ground truth phylogeny (a graph representing the relative evolutionary distance between species). The images include segmentations, and can be used for clustering and deep hierarchical metric learning. As far as we know, this is the first dataset released specifically for generating phylogenetic trees. We provide several benchmarks for deep metric learning using a selection of state-of-the-art methods.

Related Material


[pdf] [supp] [code]
[bibtex]
@InProceedings{Hunt_2022_ACCV, author = {Hunt, Roberta and Pedersen, Kim Steenstrup}, title = {Rove-Tree-11: The not-so-Wild Rover, A hierarchically structured image dataset for deep metric learning research}, booktitle = {Proceedings of the Asian Conference on Computer Vision (ACCV)}, month = {December}, year = {2022}, pages = {2967-2983} }