Multi-Class CNN Models for Banana Ripeness Classification

Abstract

This study develops and evaluates Convolutional Neural Network (CNN) models for classifying banana maturity stages using images, addressing a significant challenge in the banana supply chain. The banana industry represents a major agricultural sector worldwide, with Brazil exporting 56.2 thousand tons in 2023. Accurate maturity classification is essential for optimizing harvest timing, reducing post-harvest losses, and extending shelf life. We utilized a public Brazilian dataset of 1,000 Prata Catarina banana images categorized into eight ripening stages based on peel coloration standards established by the Brazilian Program for Horticulture Modernization. The images were preprocessed to a standardized 200x200 pixel resolution, and we evaluated the effectiveness of data augmentation techniques including horizontal flip, vertical flip, rotation, and zoom. Our CNN architecture consisted of five convolutional blocks with a dropout layer prior to flattening. We conducted six experiments comparing three classification scenarios (8, 5, and 2 ripeness classes) with and without data augmentation. The models achieved test accuracy ranging from 45.3% to 89.5%, with optimal precision and recall of 87.2% and 89.6% respectively in the two-class model without data augmentation. Performance improved as the number of classes decreased, highlighting the challenge of distinguishing between visually similar ripening stages. This research provides a fundamental reference for future banana ripeness classification studies and demonstrates the potential for practical applications using mobile device cameras, supporting increased productivity and sustainability in the banana industry.