Classifying sows' activity types from acceleration patterns using a Multi-Process Kalman Filter

Monitoring the types of activity that sows exhibit in group housing would help detecting deviations from their ‘normal’ behaviour and inform farmers about the specific state of an individual animal. By mean of an accelerometer fixed on individual sows, activity data can be recorded at any time; this implies that modelling activity patterns can allow monitoring the individual animal during its whole reproductive cycle, i.e. from the mating section to the farrowing house.