Infrared proximity measurement system development and validation for classifying sow posture
The rapidly progressing field of precision livestock farming is becoming increasingly dependent on the utilization of camera technology. Integration of camera technology involves substantial intellectual input and computational power to acquire, process, and interpret images in real-time. Further, cameras and the necessary computational power can be cost-prohibitive and subsequently, become a constraint for application in a commercial livestock and poultry production systems. The purpose of this study is to develop an infrared proximity sensor based system to serve as a substitute a camera system to perform real-time monitoring of sow posture in farrowing stalls for a potentially lower cost and computational power. Monitoring sow posture can provide producers an indicator of farrowing and aid in evaluating sow demeanor during lactation. During the development of this system the long range infrared (IR) proximity sensors were individually calibrated, a sow posture algorithm was developed, and the IR-Sow Posture Detection System (IR-SoPoDS) system was evaluated in a commercial setting to a Kinect V2® camera for a range of sow postures. Average accuracy of the sow posture algorithm on the training data was found to be 96%. The overall accuracy of the IR-SoPoDS system across the three sow frame sizes were:87% (small), 90% (medium), and 89% (large). This IR-SoPoDS system shows a strong promise for further development for sow posture and behavior detection in the farrowing stall environment.
This presentation is published as Smith, Benjamin C., Brett C. Ramirez, and Steven J. Hoff. "Infrared proximity measurement system development and validation for classifying sow posture." ASABE Annual International Meeting. Boston, MA. July 7-10, 2019. Paper No. 1900327. DOI: 10.13031/aim.201900327. Posted with permission.