Traditional swine management still depends heavily on caretakers making rapid, subjective assessments of pig health and performance across large populations of pigs, says David Rosero, assistant professor in the Department of Animal Science at Iowa State University. However, outdated processes often delay early detection of diseases, limit timely targeted interventions, and result in inefficient production systems.
Precision Livestock Farming (PLF) tools, such as computer vision and automated environmental and biological sensors, are becoming increasingly important tools as swine management shifts from labor-intensive, manual tasks to automated, digitally enabled systems.
“Despite the rapid development of new PLF technologies, adoption in swine barns remains slow, primarily because swine producers are uncertain about their accuracy, reliability and economic value,” Rosero said at the American Association of Swine Veterinarians Annual Meeting.
Here are three emerging PLF technologies that Rosero and his team at Iowa State University believe offer a transformative opportunity to modernize swine operations.
1. Wearable sensors to monitor pig activity
Today’s technological advancements enable the clustering of multiple sensors into compact, innovative devices for pigs, Rosero says. Previous research has demonstrated this concept using a Bluetooth-enabled electronic sensor board that can record body and ambient temperatures, head tilt, movement and vocalizations, all integrated into an ear-tag form factor. The sensor cluster is now remarkably small and lightweight making it practical for the use in pigs at various stages of their lives, he adds.
“Wearable sensors have also demonstrated strong potential for early detection of infectious diseases,” he adds.
For example, one study evaluated a real-time monitoring system that combined an accelerometer and a thermometer within an ear tag, demonstrating that the model could distinguish between healthy and infected pigs with African swine fever (using an attenuated strain) one to two days before clinical signs became evident.
“Despite these promising results, important limitations remain for the large-scale implementation of ear-tag sensors,” Rosero says. “Costs of assembling electronic sensor boards remains high for commercial operations.”
Computer Vision is an emerging scientific field that seeks to automate tasks beyond the capacity of the human visual system, Rosero says. It integrates edge computing and artificial intelligence systems that extract and process information from images automatically using relatively low-cost equipment. Applications of computer vision technologies include assisting humans in identifying tasks, detecting events from visual surveillance, and analyzing medical images, among others.
“Traditionally, counting pigs at different production phases has been a time-consuming and labor-intensive task, often carried out alongside activities such as weaning, vaccination or sorting,” he says. “Because swine facilities house large numbers of animals, manual counts are frequently inaccurate, which can negatively affect feed and supply planning, health protocols, and marketing accuracy. New computer vision systems provide a superior solution for identifying, tracking and counting animals.”
It is true that the use of cameras in field conditions presents challenges due to variable lighting, diverse backgrounds and occlusion from pen structures. However, researchers have developed robust computer vision methods capable of overcoming these limitations with high accuracy, Rosero points out.
3. Computer Vision to estimate the body weight of pigs
Continuous and automatic monitoring of pig growth performance can provide producers with valuable insights into system efficiency, herd health status and marketing readiness, Rosero explains. In practice, however, caretakers rely on only a few traditional methods, including direct weighing with scales, body tape measurements such as heart girth or flank-to-flank measurements, and visual estimates made by trained technicians.
“The application of computer vision for body-weight estimation has demonstrated strong accuracy in research settings and is now being evaluated within commercial production systems,” he says.
A previous study conducted a direct comparison of methods using 91 individually weighed pigs in a university setting. Results showed that a walk-across scale achieved 98.2% accuracy (with six pigs unregistered), human visual estimation reached only 88.2%, and the PigVision computer-vision system achieved 96.6% accuracy. Researchers noted that PigVision was the least labor-intensive approach and provided continuous weight data throughout the growing period, although it required routine maintenance.
A novel computer vision-based system (Swine Sense Hub Camera) capable of estimating individual body weight of pigs and identifying them through ‘codeflex’ tags to was evaluated in a commercial research finishing barn in Indiana from June to November of 2025. The mean (± standard deviation) absolute percent error (MAPE) was 2.39% (± 2.31%) for Turn 1 and 2.58% (± 2.38%) for Turn 2. Concordance correlation coefficients were measured to evaluate the agreement between the camera and scale weights. Substantial agreement between weights was observed at the individual level in both turns, with estimates of 0.98 (95% confidence interval (CI): 0.975-0.983) in Turn 1 and 0.99 (CI: 0.988-0.991) in Turn 2. Excellent agreement was identified at the pen-level, with correlations of >0.99 for both turns, showing high accuracy of predicting weights, Rosero says.
“Collectively, these findings demonstrate that computer vision can achieve high accuracy while reducing labor requirements,” he says. “Moreover, these studies highlight the need for standardized evaluation protocols to validate the accuracy and reliability of new technologies across diverse production and farm settings.”
The Challenge for Adoption
Rosero says the industry needs a standardized technology evaluation process. One study identified 83 commercially available PLF technologies for pigs; however, despite the large number of devices available for swine producers, only 14% had been evaluated in scientific validation studies.
“The limited number of validated systems is concerning, as rigorous evaluation is a critical step toward commercial adoption,” Rosero says. “Field-based assessments generate essential information on accuracy, reliability and return on investment, along with practical considerations such as barn connectivity, integration with existing controllers, staff training requirements and concerns about data ownership and privacy.”
Rosero believes that for digital tools to provide meaningful value in commercial swine systems, they must consistently capture, process and report data as intended. Scientific evaluation is critical to ensure emerging PLF tools are suitable across production systems, housing environments, growth phases and genetic lines.
To address this knowledge gap, Rosero and his colleagues Sarah Phelps and Nathan Vander Werff at Iowa State University, are developing standardized evaluation tools to assess the accuracy and reliability of emerging digital technologies for commercial swine barns. These science-based assessments are designed to inform technology adoption decisions and enhance swine producers’ confidence in PLF innovations, Rosero says.
“Establishing standardized technology evaluation processes, along with technology testing centers, will be crucial to reducing adoption risk, generating independent performance evidence, and building producer confidence,” he says. “As the swine industry moves toward a more digital, data-driven future, the strategic integration of validated PLF tools will be critical for enhancing animal health, improving labor efficiency, strengthening farm decision-making, and ultimately driving greater profitability across commercial systems.”
