IMPLEMENTASI FEATURE FUSION RESNET 50 DAN DENSENET 121 DENGAN SCAM MECHANISM UNTUK KLASIFIKASI PENYAKIT M ATA PADA CITRA FUNDUS

Kurniawan, Muh. Irsyad Dwi (2025) IMPLEMENTASI FEATURE FUSION RESNET 50 DAN DENSENET 121 DENGAN SCAM MECHANISM UNTUK KLASIFIKASI PENYAKIT M ATA PADA CITRA FUNDUS. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21081010127.-cover.pdf Download (2MB)
	Text (Bab 1) 21081010127.-bab1.pdf Download (82kB)
	Text (Bab 2) 21081010127.-bab2.pdf Restricted to Repository staff only until 5 December 2027. Download (889kB)
	Text (Bab 3) 21081010127.-bab3.pdf Restricted to Repository staff only until 5 December 2027. Download (658kB)
	Text (Bab 4) 21081010127.-bab4.pdf Restricted to Repository staff only until 5 December 2027. Download (2MB)
	Text (Bab 5) 21081010127.-bab5.pdf Download (12kB)
	Text (Daftar Pustaka) 21081010127.-daftarpustaka.pdf Download (77kB)
	Text (Lampiran) 21081010127.-lampiran.pdf Restricted to Repository staff only Download (135kB)

Abstract

This study aims to improve the performance of eye disease image classification through the implementation of the Feature Fusion method, which combines two pretrained architectures, ResNet-50 and DenseNet-121, along with the application of the SCAM (Spatial and Channel Attention Module). This approach is designed to leverage the strengths of each model in extracting deep features and to enhance the representation of important spatial and channel features, thereby producing more accurate and stable predictions. The dataset used in this study was obtained from the open-source Eye Diseases Classification dataset on the Kaggle platform, consisting of four image classes: Cataract, Diabetic Retinopathy, Glaucoma, and Normal, with a total of 4,184 images. All data underwent a preprocessing stage that included resizing, augmentation (rotation and flipping), and normalization to ensure consistency and increase the diversity of the training data. The model was evaluated using the accuracy, precision, recall, and F1-score metrics to assess its classification performance. The experimental results show that the combined ResNet-50 and DenseNet-121 model without SCAM achieved a maximum accuracy of 94.52%, while the model with SCAM improved the accuracy to 95.48% with an average F1-score of approximately 0.94. These findings indicate that the application of the SCAM module significantly enhances the model’s ability to highlight important features in fundus images, resulting in a more effective, accurate, and reliable eye disease classification system to support early detection in medical imaging.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Sari, Anggraini Puspita	NIDN0716088605	anggraini.puspita.if@upnjatim.ac.id
Thesis advisor	Junaidi, Achmad	NIDN0710117803	achmadjunaidi.if@upnjatim.ac.id.

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Mr. Muh. Irsyad Dwi Kurniawan

Date Deposited:

05 Dec 2025 08:40

Last Modified:

05 Dec 2025 08:49

URI:

https://repository.upnjatim.ac.id/id/eprint/48091