Quality control of pikeperch juveniles using digital imaging and AI

The goal is to evaluate the possibilities of applying digital technology and artificial intelligence in fish hatcheries.

Fiche signalétique

  • Département responsable Haute école des sciences agronomiques, forestières et alimentaires
  • Autres départements Technique et informatique
  • Institut Agronomie
  • Unité de recherche Animaux de rente et chevaux
  • Organisation d'encouragement Autres
  • Durée (prévue) 07.07.2020 - 30.04.2021
  • Responsable du projet Thomas Janssens
  • Direction du projet Thomas Janssens
  • Équipe du projet Thomas Janssens
    Philippe Aebischer
  • Partenaire Innosuisse
    Swiss Food Research
    Centre Suisse d'électronique et de Microtechnique SA
    Swifish AG
  • Mots-clés pikeperch, quality control, artificial intelligence, digitalization

Situation

The potential of image processing, deep learning and latest camera technology shall be tested and evaluated on the field. The ultimate goal is improve efficiency of production by facilitating decision taking and improving quality of the juveniles.

Approche

- Selection of the right vision system (camera, lens, illumination, filters). - Classification of swimbladder inflation with deep learning - Evaluation of weaning performance - Classification of quality of the juveniles (no injuiries or abnormalities) and measurement of the size heterogeneity - Potential of automated sorting system according to development, quality and size