A Fuzzy K-Nearest Neighbour-based Model for detecting Lameness in Cattle

Abstract

In Africa, cattle are reared for meat and milk production and lameness is considered a major problem in modern dairy farming. Several studies have attempted to explicate automatic lameness detection systems using different techniques. However, these detection techniques are easily impacted by the physiological attributes of individual cows, ensuing in imprecise lameness detection. Consequently, this study presents a description and assessment of the performance of a fuzzy k-nearest neighbor (FKNN)-based classification system for automatic lameness detection from sensor data with a view to improving cattle lameness detection accuracy by reducing the rates of false-positive alerts. In order to further improve the model detection accuracy, Principal Component Analysis (PCA) was used in projecting the features on to a lower dimension on which the optimal FKNN model was formulated. The proposed system was tested using Classification Accuracy(Acc), sensitivity, specificity and the area under the receiver operating characteristic curve (ROC) curve (AUC) as performance metrics.