Automated detection of parturition - lessons from the lambing shed

Abstract

We aim to predict imminent birth events in livestock through detection of periparturient behaviours such as circling and restlessness using remote Wireless Sensor Network (WSN) technology (Hurley et aI, 2006). The algorithms to classify these behaviours will be achieved via the fusion of high-rate, low power inertial sensor data collected on numerous animals up to 10 days prior to giving birth during a CSIRO FD McMaster lambing trial in late 2012. We believe this to be the first attempt at developing algorithms to remotely predict birth events in pregnant merino ewes using WSN technology. Many challenges were encountered during the data collection phase of the experiment. These insights will aid development of future WSN technologies deployed on livestock.