This project developed SNORT (Swine Nutritional Observation and Routing Technology), a software application that leverages innovations in visual sensemaking for in situ detection, tracking, and monitoring for individual weaned pig weight growth assessment for feed transition recommendations. Here we addressed the perennial problem of weaning pigs not eating well during transition to finishing, which often results in reductions in weight and increases in finishing time, as well as susceptibility to illnesses. Individual pig identification and traceability are central to optimal production goals, and it is important this is deployed as early in the cycle as possible. Human caretakers are relied upon in current production operations to observe pigs daily to detect feed, water, illness, and air issues, to identify compromised pigs in need for care, and most often there is not satisfactory time and resources to do so. This work emerged from the need for innovative tools for production-scale swine operations to enable consistent and reliable monitoring of weaning pigs, which is one of the biggest challenges faced by the swine industry today. At present, the industry has had few objective tools to identify and/or track individual pigs throughout the production cycle, and these are often invasive, costly, and susceptible to failure. The overarching goal of this research was to minimize yet enhance human efforts through automated visual-only continuous recognition of pig identity, nutritional intake (feed and water), and behavior activity patterns, throughout the weaning process. Visual sensing is a non-invasive method that relies on smart cameras to automatically detect, track and measure objects of interest, often at high frequency. The findings in this work will be of interest to practitioners and researchers interested in improving weaning pig processes in grow-finish animal production systems.

Key Findings

  • A computer vision and machine learning algorithm to determine individual pig activity habits in a pen to assess food intake was validated.
  • First eating time, eating time duration, and a relationship between food intake and movement activities was observed over time from the video data.
  • This work utilizes a primary video data source that could also be collected by producers using standard security camera technology that might already be in place in barns.