EYE DISEASE CLASSIFICATION USING DEEP LEARNING

Abstract

Diabetic retinopathy, glaucoma, and cataract are among the most prevalent eye disorders, often challenging to diagnose accurately during their initial stages. Failure to identify these conditions early can lead to irreversible vision impairment. Manual diagnosis of such diseases may also introduce human errors and inconsistencies. This project focuses on developing a robust deep learning–based model for automatic eye disease classification and detection. The main goal is to determine whether an individual is affected by an eye disease using retinal fundus image data. The proposed framework employs the ConvNeXt-Tiny architecture to perform the classification and enhance diagnostic accuracy across larger datasets. The model successfully achieved an accuracy of 97.16%, showcasing its ability to extract information with high accuracy and differentiate disease-specific visual features among diabetic retinopathy, glaucoma, and cataract. The system contributes to improved early detection, thereby supporting timely medical intervention and better patient care.