CNN Algorithm Approach for Classification of Tomato Fruit Maturity Levels

Abstract

The categorization of tomato maturity is covered in this study, which has important ramifications for the food sector and agriculture. For training efficiency, the approach uses augmentation with adjustments to rescale picture pixel values and shrink image sizes. According to the experiment's findings, accuracy increased by 93% throughout five training epochs. The training and validation graph indicates steady progress, despite the lack of significance in the improvement. Misclassifications that require correction are found during evaluation utilizing the confusion matrix. The study emphasizes that to enhance agricultural production management, flaws in the model must be filled and accuracy must be increased. The amount and diversity of photos in the dataset should be increased, as should the shooting angles and lighting conditions, and hyperparameters should be adjusted for future model performance optimization.