Date Full Report Received


Date Abstract Report Received



Primary Investigator:

Ensuring the health and wellbeing of pigs is of the utmost importance to the swine industry. However, one of the biggest challenges to ensure the health and wellbeing of pigs is the rapid and accurate identification of compromised (injured or illness) pigs. However, to date, the only method utilized to identify compromised pigs is visual recognition. While effective, there is substantial room for improvement/enhancement. A major limitation to visual identification is that the pig must exhibiting visual indicators that can be recognized by the producer. Thus, until the pig is displaying recognizable behavior/movement there is no intervention. As such, there is a need for a real-time system that can identify changes in pig activities, as well as activity patterns to accurately identify compromised pigs prior to observance of visible clinical symptoms by facility personnel during daily checks. Therefore, a novel computer vision system that can automatically maintain individual identification and continuously track activities of group housed pigs was developed and evaluated. Thus, the objective of this research trial was to evaluate the viability of the depth-sensing camera coupled with multi-ellipsoid fitting and deep learning detection programing to automatically identify, maintain identity and continuously track the activities of group housed newly weaned nursery pigs. To accomplish this objective, three trials were conducted within a commercial nursery (~1,300 head, Union Farms, Ulysses NE) and one trial within the Animal Science Complex at the University of Nebraska – Lincoln. For each trial, the system was installed above a pen with 14 – 15 newly weaned nursery pigs. At the conclusion of each trial, captured data was analyzed to evaluate the system’s ability to identify, maintain identity, and track the activities/activity patterns of activity of the nursery pigs. Evaluation of 1,020 randomly selected data points indicated an 99.8% accuracy rate for correctly identifying pigs’ location within the pen, body orientation and identity of the pig when standing/walking. Orientation/identity accuracy was reduced to 92.5% when pigs were lying. Classification accuracy for activities was 99.1, 93.6, 97.3, and 80.0% for lying, standing/walking, at the feeder and at the waterer, respectively. The accuracy of the system provides the ability to accurately track the activities of group housed nursery pigs and evaluate changes in these activities over time. Utilizing data, we were able to evaluate the time spent associated with each activity. Activity data generated from the trial conducted at the Animal Science Complex indicated that during the first 15 d of the nursery phase, pigs spent 78.3% of time lying, 17.5% of time standing/walking, 6.5% of time at the feeder, and 0.6% of time at the waterer. The average time associated with each activity also changed over time. On the first day of the nursery phase, pigs spent 72% of the time lying. By day 15, time lying had increased to 81%. On average, pigs traveled 11.3 miles (59,893 feet) during the first 15 days of the nursery turn. The average daily distance traveled was 3,989 ft / day; ranging from 2,874 to 4,718 ft./day. As the time pigs spent in the nursery phase, the distance traveled each day decreased. During the first 5 days, the average distance traveled was 0.90 miles (4744 ft). During the last five days (days 11 – 15), the average distance traveled was 0.67 miles (3526 ft). In addition, the system is capable of functioning in the unique environmental conditions of a commercial nursery facility. Across the trials conducted, the system continuously collected data for 60 days within a nursery facility with no technical failures. Overall, the results of this study indicate that the system (depth-sensing camera coupled with multi-ellipsoid fitting and deep learning detection programing) is capable of accurately for identify individual pigs, maintain identity of these pigs within a nursery pen. The system is also capable of accurately tracking the activities of multiple pigs over an extended period of time within the environmental conditions of commercial nursery facility.