Document type: article published in Innovations Agronomiques
Authors: Agathe Cheype, Béatrice Mounaix, Jérôme Manceau, Vincent Gauthier, Quentin Delahaye, Claire Dugue, Laure-Anne Merle, Xavier Boivin
Preview: English abstract provided by the authors. Assessing the welfare of young fattening bulls: refined indicators to improve feasibility without compromising reliability
The routine assessment of welfare in young fattening bulls requires rapid and remote observations. Fourteen indicators, derived from the scientific literature and validated in collaboration with farmers and technicians, were tested on two experimental farms and 31 commercial operations. These indicators proved to be generally reliable and feasible under field conditions. The main challenges relate to the direct observation of certain indicators (injuries, cleanliness, human reactivity). Most farmers found the indicators acceptable, although those based on behavioral assessment raised more concerns. To address this, an automatic behavioral analysis tool using a video-based Deep Learning algorithm was developed to enhance the reliability and ease of behavioral measurements. The algorithm showed promising performance, with specificity exceeding 80% and sensitivity over 78%.
