Banana and Orange Classification Detection Using Convolutional Neural Network

Fruits play a crucial role in human health, with an average consumption of 81.14 grams per capita per day in Indonesia, where bananas and oranges are the most consumed fruits. Inconsistent fruit quality, typically evaluated manually by farmers, can influence consumer decisions. Artificial intelligence (AI) and computer vision can enhance efficiency and consistency in analyzing fruit quality. Convolutional Neural Networks (CNN) are particularly effective in image recognition. This research uses CNN to classify the quality of bananas and oranges from a dataset of 4000 images, divided into 10% test data, 80% training data, and 10% validation data. Among three models tested, Model 2 performed best with an accuracy of 96.75% and balanced high F1-scores across all categories. The results demonstrate that the CNN model is capable of classifying the quality of bananas and oranges with high accuracy and good evaluation results.