EUVIMEDEUVIMED entdecken
Open Access

Recognizability bias in citizen science photographs

Wouter Koch iD, Laurens Hogeweg iD, Erlend B Nilsen iD, Robert B O'Hara iD, Anders G Finstad iD

Royal Society Open Science · 2023-02 · 10 (2)

DOI: 10.1098/rsos.221063 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Royal Society Open Science, 2023-02. Verfasst von Wouter Koch und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Natural History, Norwegian University of Science and Technology, 7491 Trondheim, Norway. Die Arbeit wurde bisher 19-mal zitiert (Crossref) und nennt 35 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: Norges Forskningsråd.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Citizen science and automated collection methods increasingly depend on image recognition to provide the amounts of observational data research and management needs. Recognition models, meanwhile, also require large amounts of data from these sources, creating a feedback loop between the methods and tools. Species that are harder to recognize, both for humans and machine learning algorithms, are likely to be under-reported, and thus be less prevalent in the training data. As a result, the feedback loop may hamper training mostly for species that already pose the greatest challenge. In this study, we trained recognition models for various taxa, and found evidence for a ‘recognizability bias’ , where species that are more readily identified by humans and recognition models alike are more prevalent in the available image data. This pattern is present across multiple taxa, and does not appear to relate to differences in picture quality, biological traits or data collection metrics other than recognizability. This has implications for the expected performance of future models trained with more data, including such challenging species.

Schlagwörter

Stichwörter der Autor:innen

citizen scienceimage recognitionmachine learningrecognizability

Autor:innen und Institutionen

  • Department of Natural History, Norwegian University of Science and Technology, 7491 Trondheim, Norway.
  • Norwegian Biodiversity Information Centre, Havnegata 9, 7010 Trondheim, Norway.
  • Intel Benelux, High Tech Campus 83, 5656 AE Eindhoven, The Netherlands.
  • Naturalis Biodiversity Center, PO Box 9517, 2300 RA Leiden, The Netherlands.
  • Norwegian Institute for Nature Research, Postboks 5685 Torgarden, 7485 Trondheim, Norway.
  • Faculty of Biosciences and Aquaculture, Nord University, Steinkjer, Norway.

ORCID-Profile: Wouter Koch, Laurens Hogeweg, Erlend B Nilsen, Robert B O'Hara, Anders G Finstad

Förderung

  • Norges Forskningsråd

Zitieren

APA

Koch, W., Hogeweg, L., Nilsen, E. B., O'Hara, R. B., & Finstad, A. G. (2023). Recognizability bias in citizen science photographs. Royal Society Open Science, 10(2). https://doi.org/10.1098/rsos.221063

Vancouver

Koch W, Hogeweg L, Nilsen EB, O'Hara RB, Finstad AG. Recognizability bias in citizen science photographs. Royal Society Open Science. 2023;10(2). doi:10.1098/rsos.221063

BibTeX
@article{Koch2023,
  author = {Koch, Wouter and Hogeweg, Laurens and Nilsen, Erlend B and O'Hara, Robert B and Finstad, Anders G},
  title = {Recognizability bias in citizen science photographs},
  journal = {Royal Society Open Science},
  year = {2023},
  volume = {10},
  number = {2},
  doi = {10.1098/rsos.221063}
}
Kostenlos starten →