Classification of Fresh and Rotten Fruits and Vegetables Using a Customized Convolutional Neural Network

M SYAFIQ UBAIDILLAH (Program Studi Sistem Informasi, Universitas Islam Negeri Sunan Ampel, Surabaya.)
REYHAN DANY INDRARTO (Program Studi Sistem Informasi, Universitas Islam Negeri Sunan Ampel, Surabaya.)
DWI ROLLIAWATI (Program Studi Sistem Informasi, Universitas Islam Negeri Sunan Ampel, Surabaya.)
AHMAD YUSUF (Program Studi Sistem Informasi, Universitas Islam Negeri Sunan Ampel, Surabaya.)

MIND (Multimedia Artificial Intelligent Networking Database) Journal
Vol 10, No 1 (2025): MIND Journal

This research aims to develop a customized Convolutional Neural Network (CNN) model based on ResNet-18 for classifying fruit and vegetable types and freshness. The Fresh and Rotten dataset was used to train and test the model, consisting of 30,357 images across nine fruit and vegetable categories. The model employs three additional blocks to enhance classification capabilities. The study results indicate an average accuracy of 98% for freshness classification and 99% for fruit and vegetable type classification, with consistent training and validation performance. Data augmentation and normalization methods also improved the model's generalization capabilities. These findings highlight CNN's potential as a reliable tool for agricultural product management, supporting efficient distribution and maintaining product quality.