iClassifier, an artificial intelligence invention, has demonstrated it can classify Holsteins as well as trained experts.
In a trial involving about 30,000 images, its results matched expert classifiers 98 per cent of the time measuring traits such as stature and pin width.
The team of three inventors doesn’t say whether it thinks the two per cent difference was mistakes by classifiers or its iClassifier technology.
But there was only 80 per cent agreement on quality traits such as udder texture and dairy capacity.
Ghader Manafiazar of Dalhousie University said the team is working on improving evaluation of quality traits.
The team will run trials on four Alberta dairy farms next year.
Reza Sabbagh, the chief executive officer, is an expert in optical imaging.
Another team member is at University of Alberta studying for a master’s degree in mechanical engineering and is working on technology to provide early detection of lameness.
