Login / Signup

A strategy to digitise natural history collections with limited resources.

Joaquim SantosPaulo Rupino da CunhaFátima Sales
Published in: Biodiversity data journal (2020)
The present work is a contribution towards accelerating the digitisation process of natural history collections, usually a slow process. A two-stage process was developed at the herbarium of the University of Coimbra: (i) a new workflow was established to automatically create records in the herbarium master database with minimum information, while capturing digital images; (ii) these records are then used to populate a web-based crowdsourcing platform where citizens are involved in the transcription of specimen labels from the digital images. This approach simplifies and accelerates databasing, reduces specimen manipulation and promotes the involvement of citizens in the scientific goals of the herbarium. The novel features of this process are: (i) the validation method of the crowdsourcing contribution that ensures quality control, enabling the data to integrate the master database directly and (ii) the field-by-field integration in the master database enables immediate corrections to any record in the catalogue.
Keyphrases
  • quality control
  • deep learning
  • convolutional neural network
  • adverse drug
  • optical coherence tomography
  • machine learning
  • healthcare
  • transcription factor
  • health information