Automated detection of facial expressions of pain and stress in horses

Horses have been used in many areas of our society (agriculture, transportation, sports, healthcare and others). Because of that, there exists a societal responsibility to ensure these animals’ welfare through the different activities where they are present. This PhD project will lean on past works on manual identification of horse facial expressions, and their correlation to pain and stress, to develop computer vision systems that can identify these pain/stress expressions automatically. Using data collected by expert researchers, this project aims to develop both classical and deep learning computer vision techniques to identify horse facial expressions usually associated with pain and stress, and then evaluate and compare developed approaches. One of the project goals is to create new datasets that will expand the research community ability to develop new methods of studying and ensuring animal welfare. The expected outcome of this PhD project is to study and develop an automatic system that detects and differentiates pain and stress in free-roaming horses’ videos.

This project is funded by Danmarks Frie Forskningsfond under the grant number 10.46540/3105-00114B.