An overhead scanning system combined with artificial intelligence is automatically assessing cowsâ health status twice a day on dozens of âsmartâ dairy farms across the UK.
3D cameras film the animalsâ backs as they leave the milking barn, while sensors read their individual identity tags. The associated computers then use machine learning to process the data, providing critical daily information about each cowâs weight, body condition and mobility, says at the University of the West of England (UWE) in Bristol, UK.
âWe can detect certain health conditions, including lameness, just by scanning the cows walking past, without even having to see their hooves,â he says.
Advertisement
Poor body condition and lameness in cattle reflect pain, and affected cows generally produce less milk. Delays in veterinary care can lead to poor welfare, more complex treatments, higher expenses and deaths of animals.
While farmers are well trained to recognise such issues as early as possible, they often miss subtle, day-to-day changes in individual cows. Because they are a prey species, cows will also instinctively hide signs of discomfort, such as limping, when they think they are being watched, says , who designed the system with , both at UWE.
In an , Smith and Hansen placed a computer in a waterproof box and linked it to 3D cameras set 2.3 metres high in a milking barn. They aimed the cameras downwards over a narrow walkway that cows pass through when they are finished milking. An ear tag reader, set at the cowsâ head level, triggers the cameras to record at a rate of 30 frames per second.
âThe beauty of this equipment is that itâs just passively sitting there, unobserved by the animal,â says Smith. âAnd every time they go past â so every time theyâre milked, which is usually twice a day â itâs gathering data.â
The system estimated body weight with an accuracy of 95 per cent, says Smith. Body condition scores, based mainly on flesh and fat measurements over the back and hips, were equivalent to the consensus score given by several trained experts, and more reliable than those given by individual experts.
The system also assessed spine movements to find asymmetries caused by lameness, which were confirmed by physical exams by the researchers. In some cases, the lameness was so subtle it went unrecognised by the farmers themselves. A more in-depth analysis even pointed to the specific leg that was in pain, says Smith.
at the University of Queensland in Australia, who wasnât involved in the project, says the system is âa great innovationâ. âThe dairy industry needs an automated, non-intrusive monitoring system that can aid dairy farmers in observing the health of their herd in a timely and efficient manner, and this technology meets those criteria,â she says. âThe ability to detect poor condition or ill health in dairy cows is paramount to ensuring their welfare needs are met.â
Zhang presented the teamâs research at the in Hartpury, UK, hinting that overhead body scanning may one day be useful for assessing health and welfare in horses as well.
In the near future, the team also hopes to integrate a way to evaluate the animalsâ mental state into the system using simultaneous facial scanning, says Smith.
