Feature Extraction Methods on Convolutional Neural Network Models for Objective Classification of Cacao Pods

The classification of cacao pods in the agricultural sector is crucial in determining the quality and validation of the cacao product. This study presents a classification system for healthy and unhealthy cacao pods using Convolutional Neural Networks (CNNs) integrated with three feature extraction methods: Local Binary Pattern (LBP), Edge Feature (EF), and Gabor Filter (GF). Performance of formulated models is then evaluated based on four classification metrics: Accuracy (Acc), Precision (P), Recall (R), Kappa Score (K), and F1-measure (F1). Experimental results showed that the CNN model with EF obtained the highest classification performance with Acc=0.9615, P=0.9615, R=0.9615, K=0.923, and F1=0.9615. The CNN+EF model showed perfect predictive power in classifying healthy and unhealthy cacao pods. The model obtained a high accuracy, considering that the images were trained with their actual environment, making it an advantage over other scholarly works where the pods were captured with a uniform background. 

Keywords:  cacao pods, convolutional neural network, feature extraction, image classification, machine learning 

To cite: Lumacad, G., Galarce, O., Nulo, M., & Luzano, W. (2025). Feature Extraction Methods on Convolutional Neural Network Models for Objective Classification of Cacao Pods.