Fruit Ripe Detection Using Deep Learning and Transfer Learning

Abstract:

This project presents a deep learning–based fruit classification system to automatically detect and classify the ripeness and freshness of fruits using image analysis. Leveraging transfer learning with MobileNetV2, the model is trained to categorize six classes: Fresh Apples, Rotten Apples, Fresh Bananas, Rotten Bananas, Fresh Oranges, and Rotten Oranges. The system incorporates a GUI developed in PyQt5, allowing users to train the model, evaluate its performance using classification metrics (like confusion matrix and ROC-AUC), and classify new images via a simple interface. The data is augmented using image transformations to improve generalization. The application demonstrates how AI can be effectively used in agriculture and food processing to automate fruit quality inspection, thus reducing human error and ensuring product consistency.