MIND.Funga App: images dataset from Neotropical macrofungi used to train an artificial neural network to recognize fungal species

Published: 15 September 2022| Version 1 | DOI: 10.17632/sfrbdjvxcc.1
Contributors:
Elisandro Ricardo Drechsler-Santos,
Fernanda Karstedt,
,
,
,
,
,
,
,
,

Description

Macrofungi produce reproductive structures (spore-bearing), the sporomata, visible to the naked eye. These sporomata have variable shapes, sizes, and colors. They are commonly known as mushrooms, puffballs, earthstars, jelly fungi, and bracket fungi. Informative photos can often be enough to recognize species or taxonomic groups of these macrofungi. Therefore, we built a database with standard and good-quality images for species recognition through the MIND.Funga application (MIND.Funga App). The MIND.Funga App does macrofungi image recognition based on deep learning neural network (DLNN) and suggests species names with confidence ratings. This first version of the database used by MIND.Funga App contains 14,707 images that were manually curated and resized (all images are squared). All images are separated by taxons in individual directories named with the current accepted scientific name. When species identification was not possible, the directory was named with the specimen's genus. The images were gathered and compiled from several sources, including collaborators from Facebook groups, citizen scientists, partner researchers, scientific papers, and from MIND.Funga research projects and personal data from MIND.Funga members. For the proper operation of the MIND.Funga app, it is necessary that the database used to train the DLNN be constantly supplied and composed of an expressive number of good-quality images and a representative number of species. Thus, it is expected that new versions of the database will be published in the near future. This research is supported by FAPESC/CNPq (PRONEM) 2020TR733. For more information on the collaboration net and the project, see https://mindfunga.ufsc.br/.

Files

Institutions

Universidade Federal de Santa Catarina

Categories

Fungus

License