Archive Meta Data

Author(s)

Roberta Hunt, Kim Steenstrup Pedersen

Title

Rove-Tree-11

Description

# Rove-Tree-11 Dataset 2022
This folder contains the dataset from the following scientific works：

> Roberta Hunt, Kim Steenstrup Pedersen. Rove-Tree-11: The not-so-Wild Rover. A hierarchically structured image dataset for deep metric learning research.
> Asian Conference on Computer Vision (ACCV), 2022.

The dataset was created in 2022 by Roberta Hunt and Kim 
Steenstrup Pedersen at Department of Computer Science (DIKU), University 
of Copenhagen, Denmark, with significant contributions from Alexey 
Solodovnikov, who assembled the phylogeny, David Gutschenreiter, Søren Bech 
and Andre Fastrup who took the photos and completed the initial segmentations. 
The dataset was published in September 2022.


## License and terms of use
[![Creative Commons License](https://i.creativecommons.org/l/by/4.0/88x31.png "Creative Commons License")](https://creativecommons.org/licenses/by/4.0/)

This dataset is released under a [Creative Commons Attribution 
4.0 International (CC BY 4.0) license](https://creativecommons.org/licenses/by/4.0/).

If you use this dataset in a scientific context please cite the publications listed above.


## Rove-Tree-11
The dataset contains segmented images of 13,877 specimens of rove beetles from the Natural History Museum of Denmark
and a current phylogeny representing the relationships between species. The challenge of this dataset is not
classification, but phylogenetic tree generation - can we make an AI learn how related different species are
based on visual features.

The dataset is split into a training set (8534 images), validation set (4399 images) and test set (954 images).
Each set represents a subfamily of rove beetles (train - staphylininae, validation - paederinae, test - xantholininae).

The phylogeny is provided in phylogeny.nh. This can be read into python using any newick-format reading package.
We used ete3 for this. An example code would be:

```
from ete3 import Tree
phylogeny = Tree('/path/to/phylogeny.nh')
phylogeny.show()
```

Code used in the paper is provided at https://github.com/robertahunt/Rove-Tree-11