Project: Towards an objective measurement method for pig welfare and health

2010-01-01 – 2014-05-31

Abstract

Every year, more than 59 billion animals are slaughtered for food production worldwide. As the demand for meat continues to rise, large-scale intensive livestock farming is becoming increasingly important. To improve animal welfare and health in this situation, automatic monitoring can be applied instead of the time-consuming manual observation by the farmer himself.

This project aims to develop a welfare monitor for pigs based on a limited number of easily measurable animal parameters. In the research, ethological and physiological assessments are used as a reference for the development and validation of a monitor based on automatically measured visual and auditory signals from the animals. The monitor consists of a camera and microphone with corresponding image and sound analysis algorithms. The image analysis algorithms monitored, among other things, drinking and eating behaviour. They estimated body weight, identified pigs in a pen and evaluated the movement and locomotion of the animals. These methods had an accuracy of 92%, 94.6%, 96.2%, 88.7% and 89.8%, respectively. The sound analysis methods determined the pen from which the sound originated with 89% accuracy and counted the number of cries in the compartment with a sensitivity of 72% and a specificity of 92%.

The image analysis algorithms were further validated in another set-up. The results of the validation showed that these algorithms functioned robustly in another set-up.

Furthermore, the combination of image and sound analysis was investigated by correlating the image activity with the sound energy. These first results show the potential of this combination. However, some technical problems hindered further research, such as the synchronisation between the camera and the microphone. Therefore, in the future, the relationship between this combination and health and well-being will be further investigated.

The image and sound analysis algorithms were used to detect a possible edge in animal welfare. This change was stimulated by stressors such as a change in food composition, feed withdrawal and regrouping of the pigs. Both image and sound analysis algorithms detected a change caused by these stressors.

The algorithms developed in this project can be used as part of a management system for the farmer. This system could provide early warning of disruptive events that reduce animal welfare or health. This gives the farmer the opportunity to intervene in time to improve the health and welfare of the animals.