• Title: “Using accelerometers and artificial intelligence methods to assess sheep welfare in pasture” ; INRAE.
  • PhD student: Lucile Riaboff
  • Affiliated unit: INRAE, GenPhySe
  • Supervision:
    • Dominique Hazard (INRAE)

Project Summary:

Animal welfare assessment tools are not well-suited to extensive farming conditions. However, animals' behavioral adaptation capabilities are strongly challenged in extensive environments. Genetic selection on behavioral traits is a strategy considered to contribute to improving welfare by facilitating animals' behavioral adaptation to farming conditions. It is therefore important to complement existing tools, particularly by developing new behavioral indicators of animal welfare in outdoor farming conditions, to eventually study the genetic lever to improve welfare through genetic selection on behavioral criteria. In the long term, large-scale measurable indicators will allow the study of the genetic determinism of welfare indicators and possibly propose welfare indicators for genetic selection programs.

Project objectives:

• Develop welfare indicators from accelerometer sensors in extensive farming conditions.
• Study the impacts of two divergent selections (+/- social; +/- tolerant to humans) on animal welfare.

Methodologies:

• Develop a behavior classification model from accelerometer data coupled with videos collected from sheep.
• Collect accelerometer data from sheep exposed to different environmental conditions.
• Identify welfare indicators or proxies from the time series collected in the previous step by utilizing the behavior classification model and applying signal processing, knowledge discovery, and data mining tools.