Development of a Convolutional Neural Network Model for Automated Ripeness Classification of Palm Oil Fresh Fruit Bunches

The palm oil industry relies heavily on accurate ripeness classification of fresh fruit bunches (FFB) to optimize oil quality and production efficiency. Traditional ripeness assessment, often performed manually, is subjective, labor-intensive, and inconsistent, leading to suboptimal harvest decisions. This study aims to develop an artificial intelligence (AI)-based system that automates FFB ripeness classification using machine learning and computer vision techniques. The objective is to create a model that reliably classifies ripeness stages, thereby improving the consistency, efficiency, and accuracy of FFB assessments in real-time. This research introduces a novel approach by employing deep learning, specifically convolutional neural networks (CNNs), to recognize complex visual patterns in FFB images that correspond to various ripeness levels. Unlike conventional methods that rely on thresholding or simple color-based analysis, our approach leverages advanced image processing capabilities to enhance classification accuracy across diverse environmental conditions. The model was trained on a comprehensive dataset of FFB images, captured under different lighting conditions, to ensure adaptability and generalizability in real-world applications. Additionally, the model is designed for use on mobile devices, facilitating real-time, on-field classification accessible to workers in the palm oil industry.