High-speed phenotyping of body and carcass composition of the growing cattle by 3D imaging and dual energy X-ray absorptiometry (DXA) coupled with machine learning

My name is Caroline Xavier and I'm a PhD student at UMR PEGASE (INRAE - L'institut Agro (| Agrocampus Ouest, Rennes ) and in the ruminant research group (Agroscope, Posieux, Switzerland). At the start of my studies, I wanted to work "in the field": meeting farmers, advising them. As my studies progressed, and in discussions with researchers and teacher-researchers, I discovered that research was far less abstract than I had imagined. I quickly became interested in the subjects and possibilities of research, which led me towards a thesis. The aim of my thesis is to gain knowledge of body composition for carcass evaluation and payment, choose the right time for animals to leave for slaughter, adapt feed intake to animal needs, understand the composition of animals at different life stages, understand the factors influencing body composition and establish individual models, and understand the determining role of body composition in feed efficiency.

I've always been interested in animal husbandry. During my studies and internships, I discovered data processing and management, as well as the use of new technologies in breeding, and in particular 3D imaging and its possible use in breeding. My thesis will enable me to discover and understand other imaging techniques such as DXA and ultrasound. At the end of the thesis, I hope to work in public or private research, as a researcher or teacher-researcher, in the fields of animal production, and/or data analysis, and/or new technologies.

• Starting date: November 2019
• University:  Institut Agro | Agrocampus Ouest
• PhD school: Écologie, Géosciences, Agronomie, ALimentation (EGAAL), Rennes (France)
• Scientific field: Agronomy (animal sciences) et new technologies
• Thesis management: Yannick Le Cozler, UMR Pegase, Institut Agro et Sylvain Lerch, UR Afpa, Université de Lorraine
• Thesis suprevisors:  Pierre Bressy et Isabelle Morel
• Funding: INRAE – Institut Agro (Agrocampus Ouest)
• #DigitAg : Thèse labeled – Axe 5, Challenge 4

Keywords: 3D imaging, growing cattle, body composition, DXA

Abstract: This French-Swiss thesis project aims to develop innovative methods for estimating the body and carcass compositions of growing cattle. These methods are based on high-speed acquisition and processing i) 3-dimensional images of the external conformation of the vigilant animal and ii) dual energy X-ray absorptiometry data (DXA) of the tissue composition on carcass. Information from these images and from post mortem chemical analyzes will be used to develop predictive models of composition of living animals. This project is based on the mobilization of modern imaging technologies that allow the acquisition of large data sets of morphological traits and chemical composition. Machine learning methods will be used to analyze efficiently these data sets, and continuously improve models. In comparison with existing in vivo approaches, the 3D imaging methods are unequaled in terms of precision / cost / acquisition time / non-invasiveness ratio. This technological breakthrough will improve the fine phenotyping of the body composition of growing animals, for application both in research and in breeding to i) adapt their diet to their individual needs, while reducing their emissions releases, and ii) determine the completion of the fattening phase.

Jury compound:

Marie-Pierre LETOURNEAU-MONTMINY (Université de Laval, Québec), Rapporteur
Jean-François HOCQUETTE (INRAE), Examinateur
Caroline MOLETTE (EURALIS), Examinateur
Mathieu EMILY (Institut Agro), Examinateur 
Yannick LE COZLER et Sylvain LERCH (Agroscope), directeur et co-directeur de thèse.

Contact :  caroline.xavier [AT] agrocampus-ouest.fr

Social network: LinkedIn 

Communications & Papers

Le Cozler, Y., Allain, C., Xavier, C., Depuille, L., Caillot, A., Delouard, J., Delattre, Luginbuhl,  Faverdin, P.2019. Volume and surface area of Holstein dairy cows calculated from complete 3D shapes acquired using a high-precision scanning system: Interest for body weight estimation. Computers and Electronics in Agriculture, 2019, 165 (104977)