• Title: “Automated detection and analysis of behaviors among piglets using artificial intelligence” ; INRAE.
  • PhD student: Adèle Denis.
  • Affiliated unit: INRAE, PEGASE.
  • Co-supervision:
    • Céline Tallet (INRAE).
    • Ewa Kijak (University of Rennes).
    • Caroline Clouard (INRAE).
  • Doctoral school: École doctorale Mathématiques, télécommunications, informatique, signal, systèmes, électronique.
  • Project Duration: 2023–2026.

Project Summary:

In promoting practices aimed at enhancing animal welfare in farming, it is important to study social interactions while distinguishing between aggressive contacts (fighting) and positive social contacts, such as social play. Current tools for behavior quantification mainly rely on tracking body parts’ positions and movements. However, this approach is unsuitable for identifying more subtle behaviors like affectionate contacts. Moreover, social interactions inherently involve multiple animals, adding complexity compared to tracking a single animal or a pair of animals. Another challenge is focusing on short-duration and infrequent behavioral events during the day. An original aspect of this thesis is the integration of audio information in addition to video recordings, to determine if piglet vocalizations can be associated with their behaviors and if sound characteristics can enrich visual analysis.

Objectives:

• Define and distinguish different types of behaviors in piglets before and after weaning through automated classification of individual and group behaviors.

Methodologies:

• Use models developed for other contexts (humans or other species) to determine which properties can be extracted or learned from continuously recorded video images.
• Apply knowledge transfer and learning techniques.