Automated systems for oestrus detection are commonly marketed as a suitable, or in some cases, a higher performing alternative to visual observation. Farmers, particularly those with larger herds relying on less experienced staff, view the perceived benefits of automated systems as both economic and physical, with expectations of improved oestrus detection efficiency with lower labour input. There is little evidence-based information available on the field performance of these systems to support these beliefs and farmer investment decisions. The performances of five systems were evaluated on three large pasture-grazed seasonal-calving dairy farms; four activity monitoring systems using accelerometer or pedometer technologies to provide alerts for cows in oestrus and a camera-based system for automated inspection of heat-patch mounting indicators.

 

Profiles of milk progesterone concentration, sampled twice-weekly, were used as the primary determinate of the timing of ovulations, supported by mating and pregnancy diagnosis records. None of the automated systems performed as well as an experienced operator using manual oestrus detection methods which achieved a sensitivity (SN) of 91% with a Positive Predictive Value (PPV) of 95%. Sensitivity ranged from 70% to 89% and PPV from 33% to 83% for the activity systems, when unfiltered activity alert thresholds were used. The camera system achieved a SN of 91% and PPV of 77%. Results may be improved in practice with the use of activity and mating records to filter out some false positives. Automated systems can play a role in oestrus detection with reduced reliance on skilled labour. However, visual observation is still required to confirm the oestrous status of alerted cows. The challenge for automated systems is to improve SN while maintaining a manageable number of false alerts (high PPV). These studies reinforce the need for consistent, evidence-based performance information and on-farm procedures to maintain and monitor performance of these systems.