Publication

Abstract : In the food sector, ensuring the safety and quality of banana products is crucial. The classification of bananas into “fresh banana” and “rotten banana” categories is the main objective of this study. The studied banana varieties are cavendish, lady fingers, and red bananas. We use the VGG16 deep convolutional neural network with a dataset of high-resolution banana images. Our method includes training-testing ratios, picture enhancement, and rigorous data preprocessing. The results indicate how well the VGG16 model performs in classifying freshness, with good recall, accuracy, precision, and F1-score. Additionally, the model successfully differentiates between cavendish, lady finger, and red bananas, highlighting its capacity to handle minute variations. This work expands the application of image classification to other fruit varieties by offering a dependable technique for quality control and automated evaluation of banana freshness.