Artificial Intelligence - Taking the Sleep out of Counting Sheep
While counting sheep is recognised for being a tedious chore that can put even the most agitated person to sleep, returning an accurate number without making costly mistakes appears to be practically impossible for humans.
As a result, some processors have turned to artificial intelligence to achieve a more precise count of the number of sheep entering a plant. Similarly, work is being done in cattle feedlots.
The trials are being conducted by animalEYEQ, a Singapore-based business that utilises camera vision and artificial intelligence to monitor cattle in intensive settings.
Andrew Grey, the company's Australian director, recently spoke at the global protein conference in Dalby. He stated that there were numerous applications for the technology, one of which was ensuring that animals were tallied correctly.
“We started talking to feedlots, piggeries and other intensive operations to see what opportunities there were in tracking and monitoring animals, mainly with the goal of detecting early symptoms of disease,” Mr Gray said.
“Most of them came back to us and said ‘that’s great you can do that, but can you count them first?’ So, we’re working with receivals at feedlots and abattoirs, tracking entry numbers and validating them against National Vendor Declarations.
“We have heard of processors having quite a big discrepancy at the end of the day, which ends up being quite costly. If they’re killing more than 100,000 head of sheep per week and they are 1pc out, at $250 per lamb, that adds up quickly.”
Mr Grey stated that the company is now testing items in abattoirs and feedlots and hopes to commercialise the technology after its accuracy and effectiveness have been proven.
“At the moment, we are focusing wholly and solely on accounting – we want to start turning a profit,” he said.
“We see counting livestock as the low hanging fruit and if we can turn a profit then we will be able to expand into other areas.
“So far, we have been able to demonstrate a 100pc accuracy in good conditions. Now we can go to processors if we are having issues with accuracy and recommend some changes to the flow of animals to get that count right.”
Mr Gray said if the company can land contracts with processors, it could use the technology in other parts of the plant.
“Cameras can very easily measure carcases, which I think is an area we could really value add with our product,” he said.
The company is running a trial at a big feedlot doing similar work to abattoirs – counting animals on arrival to validate NVDs. Mr Gray said it had the potential to move into other areas.
“We have done work tracking animals in feedlots to better detect diseases,” he said.
“There is a lot of potential uses in other parts of a feedlot, or any intensive livestock system, which will be able to do more research on when we make some more sales.”
(Source: Sheep Central)