Open Marien Archief | Vlaams Instituut voor de Zee
 

Vlaams Instituut voor de Zee

Platform voor marien onderzoek

Open Marien Archief

Het Open Marien Archief van België (OMA) biedt vrije toegang tot de digitale publicaties over de Vlaamse kust en het Belgisch deel van de Noordzee, en alle andere mariene, estuariene en kustgebonden publicaties van Belgische auteurs en wetenschappers en van buitenlandse wetenschappers geaffilieerd aan een Belgische instelling.

Meer info

Nieuwe zoekopdracht
[ meld een fout in dit record ]mandje (0): toevoegen | toon Print deze pagina

Otolith identification using a deep hierarchical classification model
Stock, M.; Nguyen, B.; Courtens, W.; Verstraete, H.; Stienen, E.; De Baets, B. (2021). Otolith identification using a deep hierarchical classification model. Comput. Electron. Agric. 180: 105883. https://dx.doi.org/10.1016/j.compag.2020.105883
In: Computers and Electronics in Agriculture. Elsevier: Amsterdam. ISSN 0168-1699; e-ISSN 1872-7107, meer
Peer reviewed article  

Beschikbaar in  Auteurs 

Author keywords
    Otolith identification, Seabird diet, Deep learning, Hierarchical softmax

Auteurs  Top 
  • Stock, M., meer
  • Nguyen, B.
  • Courtens, W., meer

Abstract
    The diet of seabirds can yield important insights into the status of economically and ecologically important fish. By analyzing the otoliths found in the birds’ droppings, researchers can observe which fish the birds eat in which abundances. However, identifying the species based on an otolith image is quite labor-intensive and requires particular expertise. In this work, we show that a deep convolutional neural network can identify six fish species with high accuracy. We show that this deep learning approach outperforms more traditional methods and is also more accessible to set up in practice. By exploiting the hierarchy in the species labels, we impose a structure on the prediction probabilities, leading to a remarkable improvement compared to a conventional artificial neural network. Importantly, we can attain good results using only a modest dataset, demonstrating that such approaches are feasible for small-scale and specialized projects.

Alle informatie in het Integrated Marine Information System (IMIS) valt onder het VLIZ Privacy beleid Top | Auteurs