Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal Jurnal Algoritma
Rizky Pratama, Muhammad Hafiz
Unknown Affiliation
Computer Science & IT

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Klasifikasi Otomatis Korosi Menggunakan Convolutional Neural Network dan Transfer Learning dengan Model MobileNetV2 Rizky Pratama, Muhammad Hafiz; Akrom, Muhamad; Santosa, Akbar Priyo; Rosyid, Muhammad Reesa; Mawaddah, Lubna
Jurnal Algoritma Vol 22 No 1 (2025): Jurnal Algoritma
Publisher : Institut Teknologi Garut

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33364/algoritma/v.22-1.2182

Abstract

Corrosion is a major problem that causes significant economic losses in various industries, including transportation, energy, and manufacturing. Early detection of corrosion is essential to reduce its negative impact. This research aims to develop an automatic corrosion classification system based on Convolutional Neural Networks (CNN) with a transfer learning approach. Two models were evaluated, namely a simple CNN architecture and the pre-trained MobileNetV2. The dataset consists of corrosion and non-corrosion images divided into training, validation, and testing data. Data augmentation techniques are applied to increase the variety and number of samples in the training process. The experimental results show that MobileNetV2 achieves a testing accuracy of 95%, which is higher than that of a simple CNN that only reaches 82%. In addition, MobileNetV2 showed better performance in identifying both classes (corrosion and non-corrosion). Despite indications of overfitting due to dataset limitations, the transfer learning approach significantly improved the classification performance. This system has the potential to be applied in real-time industrial applications to reduce economic losses due to corrosion. Further research is recommended to improve the generalization of the model by using a larger dataset and applying more robust regularization techniques.
Co-Authors Akrom, Muhamad Mawaddah, Lubna Rosyid, Muhammad Reesa Santosa, Akbar Priyo