11 Evaluation and Application of Technology in Swine Production Systems

Abstract

Abstract The rapid development of smart farming technology is improving our real-time and continuous monitoring of animal health status, improvement of animal well-being, and monitoring of growth and performance to maximize revenue capture. Significant innovations are emerging to solve complex tasks in swine barns and to augment the capacity of farmers. Special attention has been given to computer vision technologies because of the low cost of cameras and the advancements in edge computing systems that automatically extract, analyze, and understand complex problems from a single image or a sequence of images. In swine production systems, computer vision technologies have been applied to monitor animal behavior, counting of pigs, estimation of pig body weight, monitoring of feed bin levels, among others. As farmers adopt new technologies, we propose that field evaluation of smart farming tools must include a scientific assessment of accuracy and reliability. Internal validation of the technology PigVision, a smart camera system that estimates the body weight of pigs, used a total of 32 smart cameras installed in a research-commercial finish barn (100 pens, capacity for 2,400 pigs) in Iowa. Estimation of body weights by cameras were compared with pen weights using a built-in pen scale every 2 months for more than 2 years. The initial results of the accuracy of cameras were < 90%. After adjustments on the algorithms related to height of the ceiling and lighting conditions, the accuracy improved to > 96%. The average accuracy during 2022 was 96.7%. Field validation of cameras overtime allowed to assess the reliability of the system by measuring the uptime of cameras. We defined uptime of cameras as the percentage of daily body weight estimations reported to end users relative to the number of cameras placed in the barn. During the first cycle of production (April to July 2020), the uptime averaged 97.2%. The uptime decreased to 85.6% in the next cycle (August to December 2020). Improvements in the hardware, internet reliability and periodicity in lense cleaning resulted in uptimes to > 98.5% during the last 2 cycles tested (February to October 2022). As accuracy and reliability of smart cameras meet expectations, we envision that these can be used to monitor and predict changes in the growth of pigs, early detection of health challenges and support marketing strategies. We anticipate that technology smartness for continuous monitoring of animal growth and health will drive a major shift from our conventional methods of production to more advanced and efficient systems. We expect that the implementation of computer vision technologies in swine barns will accelerate as they continue to prove accuracy, reliability, and cost-effectiveness.