Article Per Year (5 Year)
p-Index From 2021 - 2026
0.23
P-Index
This Author published in this journals
All Journal Jurnal ELTIKOM : Jurnal Teknik Elektro, Teknologi Informasi dan Komputer
Aulya Rachma Dewi
Universitas Ahmad Dahlan
Aerospace Engineering Computer Science & IT Control & Systems Engineering Electrical & Electronics Engineering Engineering

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

Found 1 Documents
Search

 1 
Hybrid ResNet50 with Convolutional Block Attention Module (CBAM) for Image Classification using Fine-Tuning Aulya Rachma Dewi; Aris Thobirin; Sugiyarto Surono
Jurnal ELTIKOM : Jurnal Teknik Elektro, Teknologi Informasi dan Komputer Vol. 10 No. 1 (2026)
Publisher : P3M Politeknik Negeri Banjarmasin

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.31961/eltikom.v10i1.2089

Abstract

Image classification is a crucial area in digital image processing that requires models capable of robust and stable feature representation. The main challenges in this study include variations between visual classes, di-verse image quality, and limited labeled data, which often hinder the model’s ability to generalize optimally. This study proposes a hybrid ResNet50-CBAM approach, which integrates the strengths of the ResNet50 archi-tecture in deep feature extraction with the Convolutional Block Attention Module (CBAM) attention mecha-nism to improve the model’s focus on the most informative areas of the image. The training process was carried out in two phases, namely transfer learning to utilize the initial representation from the ImageNet dataset, fol-lowed by fine-tuning to adjust the network weights to the image characteristics of the research dataset. The datasets were reorganized and split into 70% training, 15% validation, and 15% testing subsets to ensure a balanced distribution of samples. In addition, various augmentation techniques were applied to increase data diversity and improve the model’s generalization capability. The evaluation results showed that this hybrid approach achieved an overall accuracy of 99%, indicating very high and consistent performance across the entire dataset. The integration of CBAM into the ResNet50 architecture was proven to strengthen the feature extraction process by highlighting the most relevant areas, resulting in a more accurate, stable, and effective image classification model for a wide range of artificial intelligence image processing applications.
Co-Authors Aris Thobirin Sugiyarto Surono