Sheep facial recognition technology could soon be used to help with mothering up if an agtech company has its way.
New Zealand-based neXtgen Agri will this month start trials in Victoria as it finetunes the technology, marking the first time it has been tested end to end.
The trial will take place on Mr Ferguson's parents' farm north of Hopetoun, where they run Poll Dorset and White Suffolk studs.
CEO of neXtgen Agri, Mark Ferguson said the last time they trialled their prototype there, they were using a static camera, but now the camera rotates 360 degrees with a zoom function.
"We will put them near water points or congregation points like camps but we don't have to be as finicky about that now because the camera is a lot smarter than the last time we tried it," he said.
"We're confident in having facial recognition models, we're confident in doing the matching, we just haven't run it end to end yet... if it all works like it should, then we're straight into multiplying the number of cameras we've got and starting to offer it as a service fairly short term.
"We're hoping we will be offering it as a mothering up service within a few months if all goes well... we're hoping we can get some done in Australia for this lambing.
"Kiwis lamb a bit later so we definitely want to be organised for that."
The company has been working on the project for about five years, trialling the technology in both New Zealand and Australia.
Mr Ferguson said the idea originated from the need to be able to easily differentiate which lambs belong to which ewes.
"We started off playing around with sensors but quickly came to the realisation that price point for sheep makes it pretty difficult for a commercial scale," he said.
"Unless they're stud sheep, farmers aren't going to pay the sensor type of price."
MORE READING:
A project involving neXtgen Agri, Australian Wool Innovation and Sydney in 2019 saw an early prototype used to photograph about 2500 sheep.
"We were pretty naive back then," Mr Ferguson said.
"We used the wrong models and the wrong systems... it's been completely rebuilt since then but it did show that they were different enough that the computer could tell them apart.
"Now we basically mimic the technology used in humans."
Mr Ferguson said trialling the technology in both countries meant that they could continue work during the different lambing seasons.
"Our aim has always been to make it very accessible, a price point of $4 or $5 per lamb to match and we think that the value we can bring to a farm will make that a very good investment," he said.
"We're hoping that anyone who is running a maternal flock and breeding their own replacements will find it a viable proposition and that we can put it in a package that is user-friendly.
"We envision that over time we can build a database where you can see what a ewe did last year, what she did the next year and if she's good you keep her and her daughters, if she's not so good then her and daughters are gone.
"Then you build your flock based on this data and every year it gets stronger and more complete as you start learning things like maternal behaviour and which ewes stick close to their lambs and the sort of thing we can't tell at the moment because we don't have the time.
"A computer's got all the time in the world as long as the sun's up... whether we crack it or someone else does there's going to be enormous potential in cameras and machine learning for grazing farms."
The trial is expected to run for two or three weeks.