Researchers in Taiwan have developed a novel way to detect oestrus in pigs using machine learning techniques.

Detection of oestrus in pigs is time sensitive as sows are only in oestrus (or “heat”) for 50-60 hours (though this can vary from 12-84 hours depending on the individual). Therefore, it is important that farmers quickly and accurately identify when a sow is in heat to optimise production efficiency. 

Traditionally, sows are examined manually for their “standing reflex”. When a sow is in heat and receptive, pressure applied to their back and flank will elicit a standing response. This technique can be stressful to sows if performed when not in oestrus, also presenting opportunities for human error. 

Another indicator of oestrus is changes in vulva size and colour. Though, similarly, eyesight alone can be subjective and allow for human error. 

Chang et al., (2023) have focused on this indicator to develop a non-contact method for the detection of oestrus using artificial intelligence. Their programme YOLO v4 uses markers that detect sow vulvas and monitor changes in their measurements, alerting operators to signs of oestrus. 

This presents a less stressful monitoring technique for pigs that farmers can use to further optimise their breeding cycles.