Tomato Maturity Assessment: Using Convolutional Neural Networks and Image Processing-Based Domain Knowledge

Abstract

This study explores the use of Convolutional Neural Networks (CNN) and image processing techniques to automate the assessment of tomato maturity, classifying tomatoes as either "ripe" or "unripe." Traditional manual methods for determining ripeness are subjective and labor-intensive, often resulting in inconsistencies due to environmental variables like lighting and background variations. A CNN model was developed and trained using an augmented dataset with diverse preprocessing steps, including background removal, color space conversion, and image resizing. The model achieved an overall classification accuracy of 95.52%, outperforming traditional methods such as Support Vector Machines (SVM) and decision trees in comparative analyses. Key metrics precision, recall, and F1 score confirmed the model’s robustness in identifying ripeness stages, with verification studies on independent data sets further demonstrating its generalizability. This research highlights the potential of CNN-based maturity assessment in enhancing efficiency in agricultural practices, especially for stakeholders in the tomato supply chain. Recommendations include expanding the dataset for broader environmental conditions and developing a mobile application for practical, field-based usage.