Since commencing in December 2023, our research project AI for Cattle Monitoring has made significant progress identifying how technology can benefit animal welfare in feedlots across Australia.
Much of this early success can be attributed to project partners Harmony Agriculture & Food Company and RMIT cultivating a strong and supportive collaboration.
“The most important factor in this project’s success,” says Ali Bab-Hadiashar, Professor and Research Leader at the School of Engineering at RMIT University, “is that we’ve been granted direct access to the paddock.”
Ali’s team consists of two other researchers, Dr Ruwan Tennakoon and Dr Amirali Khodadadian, who have been integral in the progress made to date.
“Without the access to Harmony’s paddocks, we would be working in a laboratory. So often that can be the case in projects like this. A team can spend 1-2 years working on a prototype in a lab, only to discover in the first 10 minutes of testing it in the real world that it is not fit for purpose.
“This project has involved so much development on-site, aided largely by Harmony’s technical team who have really helped this along,” says Ali.
But it’s not just technical assistance that the Harmony team has offered – the physical support has been invaluable too.
“Before we got to the site to install our cameras, the Harmony team emptied and cleaned the entire pen,” he says. “Then when the animals were reintroduced and we returned to our lab, they were on hand to troubleshoot any issues.”
Some of those issues, says Ali, were completely unexpected.
“For instance, when the animals were reintroduced, we found they were interested in eating the fresh soils placed under the cameras. So, the Harmony team had to build protective barriers to stop the cows eating the ground foundation.
“Then we found the solar panels keeping the smart water trough charged were being covered by a fine layer of dirt stirred up by the animals, so we had to continuously clean them to keep them running.
“Perhaps most interestingly though was the challenge of training our cameras to distinguish the difference between an animal and its shadow on new ground. In the afternoon sun, a dark shadow of a cow is easily mistaken as an actual cow!” laughs Ali.
The team also had to learn how to train the AI software to identify subtle variations in a cow’s appearance.
“For Angus cattle, it was straightforward,” he explains. “Each cow looks very different from each other. But when it came to Wagyu, well, they might as well all be the same black cow!
“That meant we had to change our approach, look at bringing in other technology such as RFID tags to be able to identify each cow as an individual, and to ensure that the data on that particular animal matched up.”
Although this project had the funding to bring in one research engineer, Ali says the wide interest in animal welfare meant that many students wanted to engage with the project.
“We’ve had first year students that just wanted to be involved. So, they came along helped with tasks like data labelling – which is the most arduous task in training AI.
“But we also had interest from undergraduate, postgraduate, PhD students – everyone in our community here at RMIT seems to have taken a keen interest in this project and have wanted to contribute, and I think that’s down to the eagerness to work with AI and to improve animal welfare outcomes.”
Optimism ahead
It’s the notion of animal welfare, no matter if you’re a meat eater or a vegan, that Ali says excites people about this project.
“Improving animal welfare is appealing,” he explains. “I understand and empathise with vegans who choose not to consume meat for ethical reasons. But whether you want to or not, we all seem to agree on improving the welfare of animals.
“I am a firm believer that using machines and Artificial Intelligence, when deployed correctly, will improve the welfare outcomes for farmed animals.
“The challenges are enormous - but that’s what makes it so exciting,” he says.
Harmony Agriculture & Food Company’s Chief Operating Officer Patrick Fellows says it was a no-brainer to be involved in this project.
“We all know AI has been around for years, but the recent advancements in LLMs demonstrate the potential for the technology,” he explains.
“Animal welfare is a critical pillar of our business; it’s one of our core values – so the really exciting part of this project is the potential to improve animal welfare outcomes and OHS outcomes at the same time.”
Patrick believes it’s a responsibility for agricultural companies, like Harmony, to offer resources in the development of new tools.
“Helping our team to identify morbidity earlier will have multiple flow on effects – the obvious one being improved animal welfare but also reduced time in pens for the livestock team and improved cost of production,” he says.
“Ali and his team have been stellar research partners to work with. It’s critical that research in agriculture is grounded in practical application, so to be involved in all aspects of the project right from the start was important to us.”
Patrick commended the work of the RMIT team, who he says have been adaptive in their work onsite.
“They’ve worked with what has at times been a challenging environment, from heat waves and installing cameras on 40-degree days to unexpected operational challenges. Ali and his team have been very flexible, willing to work with my team and innovate on the fly to get the project up and running.”
