Modelling Companion Animals' Well-being based on Activity Recognition and Biological Rhythms Monitoring

Human-animal companionship is a particular relationship where the well-being of a companion animal entirely depends on their human partner's care. However, proper care of an animal companion can become challenging with aging due to autonomy loss (i.e., motor or cognitive disabilities). In this regard, this very early-stage study proposes a data collection protocol centered on pet dogs, over several days in the context of an elderly-pet relationship. The goals of this study include dog activity recognition (e.g., resting, running, eating, etc.) and behaviour modelling. In order to do so, machine learning techniques will be used and daily behaviour patterns will be modeled over several days using Markov models. This research consolidates a data base on animal behaviour while bringing insights on how to measure the well-being of animals at home.