Evaluation of CNN And ANN'S Performance In Deep Learning-Based Classification of Fresh And Rotten Fruits

Authors

  • Shafa Nabila Ivana Putri STMIK Amikom Surakarta
  • Suci Muqsithin STMIK Amikom Surakarta
  • Fauzi Diana Pertiwi STMIK Amikom Surakarta
  • Ardi Joko Susilo STMIK Amikom Surakarta
  • Wahyu Maulana Putra STMIK Amikom Surakarta
  • Arif Darmawan STMIK Amikom Surakarta

Keywords:

Artificial Neural Network, Convolutional Neural Network, Deep Learning, Fruit Classification, Image Processing

Abstract

This study aims to develop and evaluate a deep learning-based system for classifying fresh and rotten fruits using Convolutional Neural Network (CNN) and Artificial Neural Network (ANN) methods. The CNN architecture is employed to automatically extract visual features such as color, shape, and texture from fruit images, while the ANN performs the final classification based on the extracted feature maps. The dataset used in this study is the Fresh and Rotten Fruits Dataset obtained from the Kaggle platform, consisting of approximately 10.901 fruit images, which were processed through image resizing, normalization, data augmentation, and an 80:20 train-test split prior to model training. The model was trained for 25 epochs and evaluated using accuracy, precision, recall, F1-score, and confusion matrix. The results show that the CNN-ANN model achieved a test accuracy of 93.75%, outperforming the baseline ANN model trained directly on raw pixel features, which achieved a test accuracy of 86.25%. Unlike previous studies that primarily focused on using CNN or ANN as standalone classifiers, this study specifically evaluates the effectiveness of CNN as a feature extractor before ANN classification, resulting in a 7.50% improvement in classification accuracy over the ANN baseline. These findings demonstrate that integrating CNN-based feature extraction with ANN classification is an effective approach for improving the performance of automated fruit freshness classification systems.

Downloads

Published

2026-08-31

Issue

Section

Articles