Agreement between passive infrared detector measurements and human observations of animal activity

Abstract

Passive infrared (PIR) detectors are widely used to measure animal activity, which is an increasingly relevant variable for livestock production insofar as activity is an indicator of animal welfare. This paper evaluates the agreement between two animal activity measurement methods: a PIR detector that uses the sensor's digital signal to perform measurements and human observations of activity in a group of 50 weaned piglets on a commercial farm. The location chosen for the sensor allowed for the recording of the main transverse movements with respect to the orientation of the sensor, which maximized its detection capacity. Human observation revealed two types of behavioral activity, feeding (eating or drinking) and playing.Based on the characteristics of both measurement methods, the correlations between PIR detection and human observation of animal activity were strong, with a Spearman's correlation coefficient of 0.90 (p < 0.001) and a concordance correlation coefficient of 0.86. PIR detection overestimated animal activity by 2.59% as compared to human observation, and the precision of PIR measurements decreased with the increase in the level of activity in the pen. In addition, animal weight affected the quality of measurements, which decreased with the increase in the ratio between kg of live weight and area covered by the sensor. The type of activity affected the precision of PIR detectors, which better detected playing activities, which are more intense than feeding activities. It can be concluded that PIR detector measurements of activity in groups of animals provide a good estimation of the relative activity of the group, which is comparable to visual estimation of relative activity.