Moo-ving research: Student projects combine agriculture and AI

“Six engineering capstone projects this year had one big, bovine thing in common: they all focused on cows. Specifically, dairy cows – and the practical applications of AI to improve their health and welfare while streamlining farm management. The projects came about through a partnership with CATTLEytics , an agri-tech company that was founded by a McMaster Engineering grad and veterinarian Shari van de Pol . CATTLEytics provided pre-approved capstone projects that students could sign up for at the start of the year. Van de Pol said they expected one group, maybe two. Instead, they got six. “ Dairy cows are one of the most underrated AI opportunities in agriculture — biologically rich, economically meaningful, and tracked at individual-animal scale for decades. The students figured that out faster than most of the tech world has ,” van de Pol said. Through their capstones, students got to work on real-life tech solutions in the ag-tech industry. And CATTLEytics got a chance to meet and nurture some of the new graduating talent. What is CATTLEytics? Van de Pol graduated with a degree in Computer Engineering and worked as an engineer for 10 years before earning her Doctor of Veterinary Medicine (DVM) from the University of Guelph and becoming a large animal vet. McMaster Engineering grad and veterinarian Shari van de Pol. Working with cows, she realized there were gaps in existing software that served dairy farms. She launched CATTLEytics, a data platform and software system, in 2014. The platform deploys cutting-edge analytical tools to support improvements in nearly every aspect of dairy farming. Now, there are 200,000 cows on the platform, which is used by hundreds of farms across Canada, the United States, Europe and Australia. CATTLEytics can do a lot, from managing tasks and staffing on the farm, to predicting how much high-quality milk a cow will produce in the coming days and years, to providing their genetic information and family tree. “Kind of like a 23andMe for cows.” Next, van de Pol is looking at building two new capabilities: Computer vision (aka cow-puter vision): using cameras to identify cows, monitor their behaviour, and flag early health concerns Pocket AI: creating a voice dictation tool that integrates with their software to allow farmers to log herd health events hands-free on their phones, converting spoken commands into structured records, tasks and updates “ Rural communities are facing labour shortages, and farmers are doing more with fewer hands. Our job is to support them both ways — with technology that works in the barn and on their phone, and with jobs. Many of our employees live in or are deeply connected to rural communities themselves.” van de Pol said. “The farmers we work with aren’t just users — they’re partners. The software exists to back up their judgment, not replace it.” The student ‘ s ’ projects dovetailed into those areas. Ultimately, it’s about helping farmers in the field – whether that’s making dictation-to-task funnels , or enabling basic cameras to monitor cows when farmers can’t be there 24/7. “It’s a way to work smarter not harder,” van de Pol said. The CATTLEytics app. Legen-dairy solutions Nielsen worked on one of the cow capstones. She’s seen the mental load on farmers firsthand. “My grandfather still farms his own land and he is 84 years old. It’s ludicrous. I don’t know how he does it,” she said. “I go over to visit and he’s like, ‘We’re not sitting down, Claire. You’re putting the door back on the barn today.’” Her group’s project (titled Cowgnition) looked at using machine learning and video surveillance to monitor cows. Nielsen’s team worked with an open-source data set they found online, which included 10-second video clips of barn footage. They trained a machine learning model to look at those videos and detect a cow. Then it would classify the behaviour the cow was exhibiting: standing or lying down; and feeding, drinking or ruminating (aka chewing their cud). Software Engineering student Claire Nielsen (far left) with her capstone group. “Our eventual goal was to be able to put in, let’s say, overnight surveillance footage for a couple days,” Nielsen said. “And then you’d get a health summary and you look and see, okay, Bessie is walking around a lot less than normal. Or, this cow hasn’t spent any time feeding.” In Canada, there are over one million cows in the dairy industry, and only around 16,000 people working in dairy farm operations. “If you wanted to have a sense of what’s going on with your cow’s health and what’s generally happening with your cattle, the ratio is unbelievable,” Nielsen said. The machine learning model could help create an early warning system for health or behavioural problems. “It can’t replace a farmer, obviously,” Nielsen said. “A human’s judgment is far and away better than AI and you’re going to know your cows.” But it could create large-scale summaries of what individual cows are doing. “If you could feed 16 hours of overnight barn footage into AI, and they gave you even some of the early issues that they could be detecting, that could make a world of difference.” The project taught Nielsen a lot about machine learning, particularly with image and video tracking. To train the model to work with videos, her group split each video into individual frames , and put a box around the cow to identify it in the frame. But if the cow moved, the computer would sometimes think it was a different cow. To fix this, they applied not just an object detection filter, but also an object tracking filter. “That was all super new to me. That was the biggest learning curve, for sure,” Nielsen said. She thinks most people would probably be surprised to learn about this intersection of agricultural and technology. “A farmer is probably as far from a software engineer as you can go. But when you really think about the applications of what’s available, especially with camera technology, AI technology, and surveillance, and just taking the mental load off of farmers to be able to offload some of that processing and analysis of your farm, it’s a really exciting place to be because there’s so many opportunities,” she said. Her own project has also led to opportunity: She’s just accepted a full-time job with CATTLEytics applying the skills she learned from the capstone onto one of their larger-scale projects. It’s a great opportunity for a new grad. “It’s a super exciting opportunity,” Nielsen said. “And an intersection of stuff that I’m interested in that I didn’t really know existed.” Agtech: More than a moo-t point Building familiarity with the ag industry among new grads is important. Van de Pol points out that one in nine jobs in Canada is food or agriculture related. Canada produces important segments of the food offerings on the world stage, whether that’s the bread basket of the Prairies, or high-quality dairy. According to the Dairy Processors Association of Canada, dairy farms contribute more than $16 billion to Canada’s GDP each year. In 2024, the dairy sector produced 96.61 million hectolitres of milk, 500,000 tonnes of cheese , and 177,000 litres of hard ice cream. “It’s incredibly fun to combine tech with foundational biological and financial systems,” van de Pol said. “Agriculture is one of the foundational industries in Canada, but unless you’re at an ag school, it barely registers. Bringing tech students into dairy means they see what’s actually here — and dairy is full of problems worth solving.” Despite that, people who aren’t familiar with the agricultural industry don’t always realize the level of data and AI that’s being put to use. “People still picture farmers in front of red barns with pitchforks, and when it comes down to it, the barn is still a barn and the cows are still cows. What’s new is how much we can understand about each animal — her cycles, her health, what she’s telling us. The technology hasn’t replaced the natural system. It’s let us see them more clearly.” van de Pol said. The c C ow capstones definitely open students’ eyes to the possibilities in ag tech, van de Pol said. It’s beneficial for CATTLEytics, as well. “Building an agri-tech company in Canada means building the community around it too. We hire students who care about agriculture, we hire locally when we can, and capstones are often where those relationships start.” Herd mentality The student groups were facilitated by Luke Schuurman – a 2025 McMaster engineering grad who now works full-time at CATTLEytics as a full stack developer. Schuurman using the CATTLEytics app in a dairy barn. Schuurman grew up on a dairy farm, and describes himself as “the computer kid who’s on the computer all day, and then you go out and do chores and you’re like, ‘Why do I have to do so much manual labour?’” He met van de Pol at an alumni networking event, connecting over their shared experiences of working on a dairy farm and attending McMaster Engineering. “The Venn diagram of overlap between dairy farmers and engineers is pretty small,” van de Pol said. Just a few years ago, an agriculture project was a hard sell to his engineering classmates. They were interested in working on solar cars, off-road vehicles and bicycles, but when Schuurman suggested tackling the Tractor Student Design Competition , it was like “a foreign concept,” Schuurman said. “It’s interesting how few people are exposed to [not just] dairy, but agriculture as a whole.” And yet dairy is ubiquitous, van de Pol points out. “Dairy is in your fridge, in your coffee, in your favourite ice cream. Most people interact with it three or four times a day without thinking about where it comes from.” “Feeding the country is foundational work, but the industry doing it has gone quiet in the public mind.” van de Pol said. “ The disconnect is real, and it means a lot of talented people never find their way to a field that would suit them.” Both van de Pol and Schuurman attended the engineering Capstone Expo event on April 7, showcasing 250 student projects – including the six cow-focused projects. Van de Pol, who has been to previous Capstone Expos, couldn’t recall seeing any projects before that were agriculture related. Schuurman heard people walking around the expo saying they didn’t realize McMaster was a cow school. Six out of 250 projects isn’t actually a huge segment, Schuurman notes – but “something about cows just sticks with people.” Recently, van de Pol took CATTLEytics to the small screen, appearing on CBC Dragons’ Den . Her pitch included bringing two cows, Pumpkin and Spice, on stage. “Even the response of the dragons seeing cows, they were like, ‘Can I touch them?’” “Projects tied to dairy or animals have an extra layer of interest, connection, meaning. The purpose behind them is real — can we improve animal welfare? Can we improve methane efficiency? When people today are looking for their why, this is a really great why.” That was the case for Mohammed Fuzail, a fourth-year mechatronics engineering student who worked on one of the capstones. He said his group didn’t have any prior experience with agriculture , but were drawn to the cow project because of the chance to make a difference for the community. “A lot of people, when they go into engineering, they go into the consumer tech side of things, like phones, laptops, cars,” Fuzail said. “Those are good fields to work in. But often, we get into this consumption mode on autopilot. So we were looking for something which hit a little bit closer to home for us.” Plus, he said, “Who doesn’t want to look at cows all day?” Hoofing it His group’s project looked at incorporating AI into computer vision to scan cows for signs of lameness, a broad clinical diagnosis that indicates an issue that makes the cow walk differently. These issues can be painful and may result in cows not eating, and not producing milk. Usually, farmers identify lameness during annual or bi-annual hoof-trimming or by watching the cows walk and deciding if they need a more in-depth analysis. Fuzail’s group took the first step in automating that inspection process using AI and computer vision. Mechatronics Engineering student Mohammed Fuzail (second from left) with his capstone group. They mounted four cameras at different angles along a laneway that cows walked through after eating. The software pipeline used three diffe rent AI models to identif y each cow, r u n pose estimation on her gait, and calculate a locomotion score using the indicators a trained eye would use — back arch, head bob, stride length, and how evenly she placed each hoof. High scor ing cows would be flagged , and all analytics displayed on the web dashboard along with the corresponding multi-gait view , for the fa r mer to check . “The whole objective is to aid the farmers in early detection of lameness – a slight shift in how a cow walks can be the first signal that something needs attention. Catching it early is what keeps cows healthy. ” Fuzail said. The next step will be continuing to build data, and cross-referencing the high lameness scores the software finds with an expert’s evaluation. But it’s a big step toward a piece of tech that could help farmers — and cows. The software could also be applied to other four-legged animals like horses, sheep or pigs. Thinking outside the box The experience taught Fuzail just how many fields can benefit from an engineering degree. Agriculture is one of the biggest industries in Canada, representing seven per cent of its GDP. “Getting tech involved in these other industries, that was the key thing that I took away from [the project],” he said. He encourages his peers and fellow engineering graduates to “expand your scope to what you can actually do with your engineering, technological knowledge that you’ve built up over the past four years.” “The knowledge you learn in engineering, once you graduate, that knowledge is yours, and it’s up to you how you use it and where you steer the innovation in technology for the next generation” he said. “It’s essential that we as engineers don’t just go with the flow of things; following what others tell us to do, but rather, make our own decisions on what are the crucial problems facing society and where our skills are most impactful.” The post Moo-ving research: Student projects combine agriculture and AI appeared first on McMaster News .

Summary generated from the RSS feed of McMaster Daily News. All article rights belong to the original publisher. Click through to read the full piece on news.mcmaster.ca.