Classification of Unhealthy Chicken based on Chromaticity of the Comb

Abstract

Human observation and laboratory tests are the traditional method for identifying bacteria- or virus-infected chicken, but these methods may result in late detection. Major disease outbreaks may occur, leading to significant economic loss and threatening human health. Therefore, this paper reports on the utilization of a supervised machine learning algorithm to provide early detection of bacteria- or virus-infected chickens based on their comb’s colour feature. Current work utilizes a well-established, International Commission on Illumination (CIE) XYZ colour space to investigate the change in the colour of the infected and healthy chicken comb. A logistic regression model was developed and proposed to classify the chickens, and the performance of the model was revealed. The chromaticity analysis shows that the comb chromaticity of the infected chicken was changing from the red to green part, based on the x chromaticity value. The performance of the proposed model indicates that this algorithm can classify between infected and healthy chickens with 100% sensitivity and 83% specificity. This work has demonstrated a new feature that can serve as an indicator for detecting bacteriaor virus-infected chickens, and contributes to the development of modern technology in agriculture applications.