Fruit Image Classification Using Naive Bayes Algorithm with Histogram of Oriented Gradients (HOG) Feature Extraction

A classification system using Naïve Bayes algorithm was developed to distinguish between fresh and rotten fruits, specifically apples, bananas and oranges. This research utilized a dataset consisting of 13,599 images and applied the Histogram of Oriented Gradients (HOG) technique for feature extraction, followed by model training and evaluation. The results showed that the Naïve Bayes algorithm achieved an accuracy of 87%, with the highest precision in the fresh apple class (0.9792) and the highest recall in the rotten apple class (0.9843). The rotten banana class showed a balanced performance with the highest F1-score of 0.9085. Although there were some misclassifications, especially in the rotten citrus fruit category, this study shows that image processing techniques have great potential and are reliable for assessing fruit quality based on visual characteristics.