According to a Canadian start-up, the findings of their artificial intelligence-based categorization of dairy cows are within 2% of human results.
Ghader Manafiazar, one of three co-founders of iClassifier, claims that the company’s machine learning-based approach for classification scores of quantitative Holstein traits such as stature (height at rump) and pin width achieved an accuracy rate of 98% after analyzing approximately 30,000 images collected from Canadian dairy farms.
Why it matters: AI-based categorization has the potential to change the way animals are appraised.
Holstein Canada offers in-person categorization services to Canadian dairy farms. A Holstein Canada spokesperson said that the association will have more to say regarding this sort of technology later this year.
The correlation to human classifier ratings for qualitative features like udder texture and dairy capacity was approximately 80%, according to Manafiazar, but “we’re trying to come up with a system that can objectively assess at least some of these qualitative traits.” With near-perfect accuracy on all attributes, the company’s founders are optimistic about commercializing the technology.
Manafiazar, a Dalhousie University in Halifax teaching member who earned his PhD in dairy cow feed efficiency at the University of Alberta in Edmonton, founded the firm alongside Alberta Mechanical Engineering faculty member Reza Sabbagh and data scientist Amir Rahvar.
The firm was created in 2021 and plans to begin operations on four farms in Alberta in early 2024.
The software utilizes cow photographs “and processes them in the cloud using the company’s unique AI algorithm.” For each cow, up to 25 distinct features, including udder texture and depth, bone quality, height, and rump angle, may be measured and analyzed, and full reports with suggestions are delivered to farmers after careful study.”
The AI-based classification tool, which uses images provided by a camera on a mobile app or a customized stall outfitted with special cameras and imaging equipment, can be used at any time on any farm setting, whereas “the human-centred process may not be readily available, and can be costly and inconsistent,” according to the company.
Manafiazar says he got interested in how information technology may be used in the dairy industry while examining dairy farms in Alberta. “I’m blessed with some very smart engineering friends,” he added in a recent interview, smiling.
With the two of them as co-founders, “we had discussions about various topics and ways to bring technology to dairy, but we thought with this one, there is an opportunity.”
Manafiazar met Nelson Jespersen, who milks roughly 300 cows on robots outside Westlock, Alta., and often welcomes university students to his farm, via his previous job at the University of Alberta. Jespersen said he used to categorize his Holstein herd but stopped because it was cumbersome and not worth the effort and money.
As part of its study, the iClassifier team took cameras to Jespersen’s farm, and he believes a more streamlined, non-biased method to categorization may tempt him to repeat the practice.
Sabbagh, the CEO of the firm, has a background in optical imaging. In an interview, he said that the team studied hundreds of photos for categorization qualities and compared them to findings supplied by a human classifier.
“The end outcome was wonderful. “We reasoned that if we could automate this practice, we could help farmers perform these assessments more frequently while also providing an accurate and objective result that they could use to improve the longevity and productivity of their herds,” he adds.
According to Sabbagh, a poll of dairy farmers was conducted roughly two years ago, and “they showed great interest in having iClassifier’s innovative technology on their farms and are even ready to participate in the technology development.”
The business unveiled the iClassifier system in Dubai from October 15 to 18, at Gitex Northstar, a huge digital start-up trade event. It is also looking forward, with aspirations of expanding system capabilities to include horses, swine, and camels.
A research team lead by a mechanical engineering master’s student at the University of Alberta is developing a customized lameness prediction algorithm for the iClassifier app. The team is collaborating with the university’s Optical Diagnostics Group to analyze data from the 150-cow Dairy Research and Technology Centre herd on the university’s south campus, thanks to funding from Mitacs, a not-for-profit organization funded by provincial and territorial governments across Canada.
Early diagnosis of lameness allows cows with foot issues to be treated before they become clinical, as is typically the case currently.
According to Manafiazar, iClassifier has the capacity to quantify certain qualitative features and hence eliminate human bias. Udder texture, for example, might be defined by capturing independent photos of thousands of udders before and after milking and using machine learning to determine an ideal.
Other prospective uses, he thinks, include body condition score and mastitis detection.
