PENERAPAN COMPUTER VISION UNTUK KLASIFIKASI PENYAKIT MATA MENGGUNAKAN ARSITEKTUR VISION TRANSFORMERS (VITS) PADA CITRA FUNDUS

Eye diseases such as cataracts, glaucoma, and diabetic retinopathy are leading causes of blindness that can be prevented if detected early. This study aims to develop an eye disease classification system using a vision transformer architecture using fundus images. The data used consisted of 4,028 fundus images evenly divided into four classes: cataracts, glaucoma, diabetic retinopathy, and normal. The vision transformer model underwent preprocessing, augmentation, fine-tuning, and evaluation using metrics such as accuracy, precision, recall, and f1-score. Test results showed that the vision transformer model was able to classify eye diseases with high accuracy and stable performance across all classes. This model was also able to effectively recognize the characteristics of each disease, demonstrating the superiority of the vision transformer in understanding the global visual context of medical images.
This study suggests that the vision transformer can be effectively used in an automated eye disease detection system using fundus images, although further optimization is still needed for use with devices with limited resources. This system is expected to facilitate early detection and reduce the workload of medical personnel.
Keywords: vision transformer, fundus image, eye disease classification, deep learning, computer vision