Quick look
A team of animal science professors at Iowa State University is studying how automated monitoring and advanced data science - the same digital tools that power self-driving cars and facial recognition - could modernize livestock production, especially in swine barns.
AMES, Iowa - Commercial pigs are bred for consistency and housed in large barns, which makes production efficient but brings challenges for identification and precision management. In a sea of hogs, which one is which?
"You might think that if we can identify a person in a crowd, surely we can identify a pig. Well, we've been trying," said Juan Steibel, a professor of animal science and the Lush Chair for Animal Breeding and Genetics at Iowa State University.
Steibel is part of a group of Iowa State researchers studying how digitally enabled barns powered by sensors, cameras and data science can transform livestock monitoring, unlocking new opportunities for precision management, welfare assessment and genetic improvement- especially in swine. The ultimate goal is continuous long-term tracking of individual pigs by computer vision, the same sort of advanced image and video analysis that unlocks a cell phone with its owner's glance, keeps autonomous vehicles safely on the road and enhances medical scanning.
"Eventually, we want to know what specific pigs to sell at specific times," said David Rosero, an assistant professor of animal science.
In the meantime, more incremental uses of computer vision, along with sensors and other automated technologies, already offer hog farmers additional insight into their herds, Rosero said. However, adoption remains limited due to uncertainty about the return on investment the systems will deliver and how reliable they'll be in commercial settings.
With a three-year grant from the U.S. Department of Agriculture, Rosero is leading a team of researchers working to close that gap by testing digital technologies and providing the pork industry with evidence-based guidelines for evaluation and implementation.
"These are impressive technologies, but they usually require significant investment. Producers need to know what the value will be, and they need to be sure they work," he said.
Testing new tech
A portion of the $294,000 grant will go toward outfitting the Swine Nutrition Farm a few miles northwest of Ames with cameras, environmental sensors and an automatic feeding system. Once that work is completed this summer, one of the main goals is to test and validate how well cameras can estimate the weight of pigs.
Rosero, who spent 10 years working in the pork industry before coming to Iowa State in 2023, said it remains common in swine finishing barns to manually select animals to sell. Using a computer vision system to assess pig weights would make the process easier and more accurate, reducing labor demands and improving the precision of market timing. Camera-powered weight evaluations could boost producer revenue by an estimated 10-20%, he said.
Iowa State researchers also will test using computer vision systems to identify struggling post-weaning pigs in need of intervention, while collaborators at the University of Missouri will study wearable sensors to augment monitoring of sows and their litters.
Tests will run through at least four production cycles, informing extension resources and training programs. Other than cost-benefit calculations, one of producers' main concerns is how new technology works under commercial conditions, said Edison Magalhaes, an assistant professor of animal science and an ISU Extension and Outreach state specialist in innovative swine systems. Digital systems in barns need to be durable and compatible with other management needs, said Magalhaes, who will help develop extension materials for the USDA smart barn grant in coordination with ISU's Iowa Pork Industry Center.
"Is it actually designed to work in a barn?" he said. "Producers can't afford to slow down their operations to accommodate new technology."
Continuous ID is 'holy grail'
The federal grant helps lay the groundwork for ongoing research into precision livestock farming at Iowa State, as did earlier funding from Iowa State's Agriculture Experiment Station to install cameras at some ISU swine and sheep barns.
That includes work by Steibel to develop computer vision systems capable of continuously identifying individual pigs. Existing pig-tracking technology performs well over limited time periods for specific purposes such as estimating weight and gauging activity levels. But maintaining identity across the full production cycle remains a major technical barrier.
"The challenge is identifying them throughout their life," Steibel said. "Animal identification is the holy grail for using computer vision on livestock. It could have so many applications."
Steibel's research group studies whether AI-enabled image analysis of key points of pigs' bodies - their heads or their joints, for instance - can support sustained tracking. They also work on classifying pig behavior with computer vision, which could inform both welfare assessment and selection for socially compatible animals.
"Is an animal in front of a feeder eating or just blocking it? Is an interaction with biting just play or is it aggression? And if it's aggression, which animal started it? It's a big question, determining what amounts to good behavior in pigs," he said.
Busy barns ahead
Tech-enhanced ISU barns are bound to host a variety of programming, including research projects, teaching opportunities and industry collaboration, researchers said.
Rosero is working with a team led by Santosh Pandey, an associate professor of electrical and computer engineering, on an automated scale system to rapidly capture precise weights that they plan to test in the swine research barn.
If pigs can be individually identified on a long-term basis, Magalhaes envisions building a comprehensive precision health assessment index for pigs that integrates food and water consumption, body and environmental temperatures, weight, behavior and any other available data.
"That could trigger an alert when something's off with an individual animal, and we can then measure the return on investment of intervening early," he said.
Private partnerships may drive some projects because the combination of precision livestock farming expertise and updated facilities is expected to attract industry interest, Steibel said.
"Part of the vision for this initiative is that as results are shared, companies building these technologies will want us to test their products and provide feedback," he said.
Students in a new upper-level undergraduate class in the animal science department - Digital Technologies in Animal Production Systems - will also have many opportunities for hands-on learning at the upgraded barns, said Steibel and Rosero, who co-taught the first-ever section of the course this spring.
"The teaching component is really important for this effort because we have to be training students to use these digital systems," Rosero said.
