Characterizing lactating sow posture in farrowing crates utilizing automated image capture and wearable sensors

Authors: MACON, ASYA - University Of Nebraska; SHARMA, SUDHENDU - University Of Nebraska; LEE, BERNARD - University Of Nebraska; MARKVICKA, ERIC - University Of Nebraska; Rohrer, Gary; Miles, Jeremy; BROWN-BRANDL, TAMI - University Of Nebraska

Submitted to: Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 7/12/2021
Publication Date: 7/16/2021
Citation: Macon, A., Sharma, S., Lee, B., Markvicka, E., Rohrer, G.A., Miles, J.R., Brown-Brandl, T. 2021. Characterizing lactating sow posture in farrowing crates utilizing automated image capture and wearable sensors. Proceedings of the European Conference on Agricultural Engineering AgEng 2021, July 4-8, 2021, Evora, Portugal. p. 634-642.

Interpretive Summary: Pre-weaning mortality is a major economic and welfare issue in swine production, and one of the major causes is the crushing of the piglets by the sow. This experiment’s objective is to track a sow’s posture changes and activity with a wearable sensor. The sensors used are accelerometers, devices that record movement using the same technology as a pedometer. Accelerometers were placed on the backs of sows to record movement and video cameras recorded images. Video image data was used to assess major movements associated with six posture changes (standing, lying-left, lying-right, lying-other, kneeling, and sitting). Results indicated that the sensors can easily differentiate between standing and lying positions as well as if the sow is lying-left versus lying-right. This system provides a path towards the collection of phenotypic data related to maternal ability, but further work is required.

Technical Abstract: Pre-weaning mortality is a major economic and welfare issue in swine production, and one of the major causes is the crushing of the piglets by the sow. This experiment’s objective is to track lactating sow static posture changes and activity in three different farrowing crate layouts using wearable accelerometer sensors. The accelerometer sensor integrates an electronic sensing circuit with rigid, waterproof enclosure to monitor the posture of the sow. The electronic sensing circuit contains 1) microcontroller for signal processing, 2) non-volatile flash data storage, 3) realtime clock, 4) rechargeable battery, 5) and three-axis accelerometer with integrated temperature sensor. The wearable sensors were placed in denim pockets with a Velcro enclosure affixed to the sow using livestock tag cement. Three Brinno BCC100 time-lapse cameras were utilized for validation of the wearable sensors. The Brinno camera collected data at 1 frame/sec. The image data was used to assess major movements associated with six static posture changes (standing, lying-left, lying-right, lying-other, kneeling, and sitting). The accelerometer data can be used to assess general activity level movement and to determine postures and precise movements. The temperature data can be used to assess the microenvironment surrounding the sow. The system provides a path towards the collection of phenotypic data related to maternal ability. The X-, Y-, and Z-acceleration orientation values were analysed using statistical analysis to determine the significant difference between the six static postures. There was a significant difference between sitting and standing in acceleration X and a significant difference between lying-left and lying-right in accelerations Y and Z. In addition, it was also found that in all three acceleration values, the sow in the offset crate was significantly different from the sows in the expanded and diagonal crates. A combination of the X-, Y-, and Z-values will be used to determine different static postures by comparing the significant differences.