CNN-based Image Classification of Duck Eggs in the Philippine Balut Industry

This study introduces a Convolutional Neural Network (CNN) model crafted to address a critical research gap within the Philippine balut industry: the classification of penoy tuyo and penoy sabaw. By focusing on this nuanced distinction, the model offers a comprehensive solution to this gap that is commonly missed out by the existing studies. Through a rigorous methodology encompassing data acquisition, preprocessing, model development, and evaluation, the CNN model demonstrates exceptional performance, achieving an impressive test accuracy of approximately 95.83% in categorizing duck eggs into balut, penoy sabaw, penoy tuyo, and abnoy. Precision, recall, and F1-score metrics further validate the model’s efficacy, with values reaching 0.9600,0.9596, and 0.9584 respectively, affirming its precision across diverse egg categories. Rigorous 10-fold cross-validation confirms the model’s reliability, highlighting its robustness and consistency in accurate egg classification. The successful deployment of this CNN-based system represents a significant advancement in enhancing efficiency, productivity, and competitiveness within the Philippine balut industry. Future research endeavors may explore additional features or technologies to further augment the model’s capabilities, fostering continual progress in automated image classification techniques for agricultural domains.