16-06-2025
Limerick brothers to trial facial recognition software in sheep
If you thought all sheep looked the same, it turns out you would be wrong.
In fact, in the world of artificial intelligence and machine learning, the differences are so distinct that two Co Limerick brothers, who are already revolutionising sheep worming, are now hoping to integrate facial recognition into their software, with trials about to begin across the globe, in New Zealand.
The Cotter brothers, Nick and Jack from Abbeyfeale, Co Limerick, have already picked up an array of awards for their invention, the Cotter Crate, which significantly reduces the labour involved in drenching lambs.
The pair are hoping to begin trials for the facial recognition software for sheep in New Zealand alongside Beef+Lamb NZ in February.
Nick and Jack have already picked up an array of awards for their invention, the Cotter Crate, which significantly reduces the labour involved in drenching lambs.
Previously, they have also developed software that allows stockkeepers to reduce wormer use on sheep by up to 60% without hampering productivity.
The brother's SmartWorm software, which has recently been validated in trials with UCD, QUB, and Beef+LambNZ, uses data on how the animals are growing, local weather conditions, and pasture availability and quality to predict whether stock should be treated for worms with 95% accuracy, with scientific studies on the project expected to be published imminently.
Later this month, the pair hope to take it a step further and will be partnering with a US company to launch an automatic drenching gun for sheep in Ireland and the UK — which will work with their software and select doses based on weight — and with that comes huge sustainability benefits.
Meanwhile, Nick will be examining the adoption of technology within the sheep sector to improve productivity, profitability and sustainability of the sector as part of his Nuffield scholarship.
'The space we want to be in is technology for sheep farmers — it's a very neglected sector, but it's also a sector we know very well, growing up on a sheep farm,' he said, explaining he was keen to overcome barriers specific to different countries.
'In Australia, some farms are not willing to handle the animals every four weeks [as is currently required by the brothers' technology] — it's too frequent for them,' he said.
Ear tags are not mandatory in sheep in Australia and New Zealand.
'These farms are spread across huge expanses of land — it's unimaginable scale. For those types of farms, we are integrating with remote weighing platforms — these are weighing platforms that operate out in the field. They are set in front of an attractant — it could be a mineral bucket or a water trough. We just need to see the animals once over a four-week period.'
Farmers can also, through the app, make settings that will draft out the ones that are needed into another paddock.
'It means the weighing is already done, and they are only working with the ones that need to be drenched rather than the whole flock and trying to sort through them manually,' Nick added.
'The reason why we started in New Zealand is that our software depends on seeing the average daily liveweight gain of sheep — we need to see each sheep individually, so what we need at the moment is we need to weigh them and we need to scan the eartag with the weight, so we know that animal No 1 weighed 40kg today, and in four weeks' time animal No 1 is 45kg, so we don't need to treat them.
"To do that, we need identification on each animal, so we need eartags.'
It's no problem in Europe, where eartags for sheep have been legally mandated for many years, however, in Australia and New Zealand there is no legislative requirement to individually identify sheep, and as a result, most sheep are not tagged.
It's set to change in Australia from January this year, when new rules were scheduled to come into effect. However, there is no sign of any similar change on the horizon in New Zealand.
It's a problem for farmers who want to be able to reduce their use of wormers there, especially considering resistance has already become a major problem in Australasia.
'At the moment, farmers who want to use our app in New Zealand would have to take on eartagging as an additional cost just to use our software,' Nick explained.
'When you have 10,000 lambs and a set of tags cost roughly €1 each, that's €10,000. We are trying to reduce the barrier to using SmartWorm by using facial recognition technology out in the paddock and on the weigh crates that we use. This is a far lower-cost technology — in fact, it's about a quarter or a fifth of the cost of ear tags.'
Sheep at Tawharanui Regional Park, New Zealand.
To achieve their aim, the brothers are working on using artificial intelligence to develop a new set of algorithms capable of recognising and identifying individual sheep.
'The machine needs to see the sheep on a regular basis and use machine learning model to analyse videos or pictures of its face,' Nick said.
'We don't use their body because the body is unreliable on sheep — for example, if they are shorn, of course, it will look very different.'
Nick explained early progress suggests there are indeed enough differences on their faces that it can identify each sheep individually. However, work is still under way to test the tech with different breeds.
But with such huge possibilities for the technology, it's a challenge well worth undertaking.
'The potential goes far beyond just individual identification for weighing, we can go much more expansive with this, with cameras out in paddocks and look at automatic detection of lameness in sheep and automatic detection of things like flystrike and go further again and look at mothering up of lambs, to know which lambs came off which mother,' he said.
The trial, which is hoped to begin in February, will also put the Cotter brothers' existing technology to the test and will build on previous research studies.
This time, it will include farms in both the North and South Islands and will also cover a longer period of time over the grazing season.
