Title

GinJinn: An object-detection pipeline for feature extraction from herbarium specimens

Stable Identifier

For the dataset:
https://data.bgbm.org/dataset/gfbio/0033/
For version 1 of the dataset:
https://doi.org/10.34656/vnat-x410.1

Citation

Ott, T.; Palm, C.; Vogt, R. and Oberprieler C. (2020). GinJinn: An object-detection pipeline for feature extraction from herbarium specimens. [Dataset]. Data Publisher: Botanic Garden and Botanical Museum Berlin. https://data.bgbm.org/dataset/gfbio/0033/

Data

Version Date Comment Format Download
12019-09-11a zip files, containing 285 images and 243 corresponding XML shape files. Total size: ≈ 890 MiB. (original data)PASCAL VOC XMLdownload icon

Licenses

License LogoCreative Commons Zero (CC0) 1.0 (for the XML shape files)
License LogoCreative Commons Attribution Share-Alike (CC-BY-SA) 3.0 (for the herbarium scans)

Details

The collection of morphological data in plant evolutionary, taxonomic and ecological studies based on herbarium material has traditionally been a labor-intensive task. We establish GinJinn as a deep-learning object-detection tool for the automatic recognition and extraction of individual leaves or other structures from herbarium specimens. As an example, GinJinn is applied to herbarium specimens of two species of ox-eye daisies of the genus Leucanthemum Mill., namely the diploid L. vulgare Lam. and the tetraploid L. ircutianum DC.

Additional Info

Creators Tankred Ott, Dr. Robert Vogt
Contributors Christoph Oberprieler , Christoph Palm
Technical Contact biodiversitydata@bgbm.org
Last updated 2019-09-11
Created 2019-09-11
Record Basis Multimedia Object
Keyword(s) object detection, deep learning, herbarium specimen