Facial recognition technology isn’t limited to humans. Livestock are now getting their mugs snapped in order to help farmers monitor animal health, increase efficiency and improve traceability.
Researchers from the US to Australia are using drones to handle health assessments of cattle from the air, developing a system to recognize the emotional states of farm animals and capturing the identities of cows based on their unique muzzle prints.
This research aligns with climate-smart agriculture—increasing the sustainability and resilience of agricultural systems in response to climate change. When Congress passed the Inflation Reduction Act last year, it included $19.5 billion to support these efforts in the US. “Agriculture is reactive; it’s not predictive,” says Dr. Sigfredo Fuentes, an associate professor at the University of Melbourne. “[With] AI, we can transform agriculture to be a more predictive enterprise or industry rather than just reacting to climate change.”
Building on current remote sensing technology, Fuentes and his team have developed algorithms that rely on noninvasive cameras and machine learning to help farmers both identify animals and monitor their biometrics (heart rate, body temperature, etc.), gathering critical intel on the volume and quality of milk, heat stress and general well-being. The goal is to help farmers make smarter decisions, detect illnesses earlier and ultimately strengthen herd welfare.
“You need to have tools to manage those more efficiently. The only way we have now to assess animal by animal is having a veterinarian, which is really expensive, even for big countries,” says Fuentes.
Typically, animal welfare tests are invasive (using tools such as blood tests and implanted contact sensors) and stressful for the animals, as well as costly and time-consuming for farmers. Noninvasive visual assessments by trained experts can be hampered by subjectivity. Digital tools—such as remote sensing, infrared thermal energy and AI—automate these exercises, limiting the impact on the animals themselves (which can bias results), reducing potential human error and allowing for more consistent monitoring of larger groups of animals.
A research review Fuentes co-authored in 2022 showed that these technologies have been successful with early detection of respiratory diseases in pigs. The contactless technology was also 96% accurate in predicting metrics such as weight and somatic cells when tested with 102 dairy cows in Victoria, Australia in 2021 and 98% accurate in estimating the animals’ ages based on facial features.
The data collection can be done as animals are moving through troughs or stepping onto transport vehicles and takes less than 10 seconds to capture. It requires basic hardware, such as a camera or a smartphone with a subscription app; it doesn’t even need Wi-Fi, a boon for rural farms. The new technology is an update on ear or RFID tags, which can fall off or be swapped.
The University of Melbourne has licensed its algorithms to commercial businesses such as iTRAKassets, which builds GPS tracking solutions for Australia’s agriculture industry. It will be working with about 85 farms to trial the system on a commercial scale. iTRAK®ID—set to be released in the fall—combines facial recognition with retinal identification to accurately assess the age, body temperature and heart rate of cattle from a quick smartphone scan. “With the retinal [scan], we can identify an animal at birth and that identity is there for life,” says managing director Stewart McConachy.
The company recently partnered with Nebraska-based HerdDogg to bring its biosecurity app BIOPLUS stateside. The technology helps farmers track livestock so, for instance, a disease outbreak doesn’t spread across fences or borders. In combination with HerdDogg’s Bluetooth tags, farmers will be able to access biometrics and tracking all in one place, in real time. The effort will eventually expand to sheep, goats and pigs.
These emerging technologies have broader cultural implications: They can make the transportation of livestock safer and more humane. In 2017, 2,400 sheep died of heat stress en route from Australia to the Middle East. If these tools had been available at the time, they could potentially have alerted veterinarians to the growing heat or informed them of which particular animals were struggling.
Accurate identification of animals also aids in traceability, ensuring farmers are receiving the animal(s) they intended to buy and offering clarity to consumers who want to know where their meals are coming from. (In Australia, the law requires that livestock identifications and movements are recorded; the US is less stringent.)
Interest in these capabilities is only growing. Global Market Estimates projects the worldwide market for AI in livestock farming will grow nearly 26% between 2021 and 2026. In short, the 85 iTRAK farms on which Fuentes’ creations are being piloted is just the beginning.
