Skripsi

KLASIFIKASI CITRA PUNCTA LACRIMAL NORMAL DAN ABNORMAL MENGGUNAKAN DEEP LEARNING

---

This study aims to classify normal and abnormal puncta lacrimal images using deep learning methods and to analyze the impact of data augmentation strategies on model performance. The dataset consisted of 61 images, including 30 normal and 31 abnormal images, which underwent a preprocessing stage by resizing all images to 256 × 256 pixels. Nine deep learning architectures were evaluated, including Vision Transformer, Swin Transformer, VGG16, ResNet50, ConvNeXt, MobileNetV3, DenseNet121, EfficientNetV2, EfficientNetB0, and a Custom CNN model, under three data processing scenarios: without augmentation, augmentation applied only to the training data, and augmentation performed before data splitting. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics. The results indicate that Convolutional Neural Network (CNN)-based models outperformed transformer-based models on the small-scale dataset. In the scenario without augmentation, VGG16 and DenseNet121 achieved the highest accuracy of 0.90. When augmentation was applied only to the training data, VGG16 and EfficientNetV2 ranked first with an accuracy of 0.90. In the augmentation-before-splitting scenario, most models achieved accuracy scores of up to 0.98, with VGG16, DenseNet121, and EfficientNetV2 demonstrating the best performance. Overall, VGG16 showed the most consistent performance across all experimental scenarios. These findings demonstrate that the selection of an appropriate model architecture and data augmentation strategy significantly influences the success of puncta lacrimal image classification.

---

Availability

Detail Information

Series Title

-

Call Number

T2004162026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xv, 91 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

617.760 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Komputer
Penyakit Kelenjar Air Mata

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment