Deep Learning Recognition of Grazing Behaviour in Pasture-Based Cattle

Abstract

Introduction: The grazing behavior is one of the important determinants of cattle health, welfare and productivity in pasture-based farming system. The common behavior analysis approaches have been mostly relying on observation, but it is tedious, time consuming, and not practical for large scale monitoring and assessment. In this research, a framework based on deep learning was proposed for identification of grazing behavior of cattle in pasture environment through movement patterns, feeding behavior and environmental factors. The gathered behavioral data were analyzed and distinct patterns associated with grazing behaviors were identified from those data. Through deep learning approach, automated identification of cattle behaviors was made possible. It has been found that the deep learning algorithms were able to capture the changes of cattle behavior patterns, and classify them accordingly. Thus, the proposed framework offers a good way for precision livestock management through monitoring of behavior patterns of cattle and enhancing welfare and decision making in the field of cattle farming. The experiment shows that behaviour recognition through deep learning contributes towards sustainable pasture management and efficient cattle production systems