AI-POWERED RETINAL DISEASE DETECTION SYSTEM

Retinal diseases including Cataract, Diabetic Retinopathy, and Glaucoma remain dangerously sight-threatening conditions, particularly in the absence of timely intervention, often resulting in irreparable blindness. Addressing this concern, this study proposes an intelligent system for detecting eye diseases based on the YOLOv8 image classification model. The system was trained to recognize four classes of eye conditions: Cataract, Diabetic Retinopathy, Glaucoma, and Normal using a labeled dataset of retinal images. The training process achieved an accuracy of 89.3%, while validation accuracy was 90.8%, indicating strong performance on new data. Class-wise evaluation showed the model effectiveness in identifying Cataract with precision of 97.4%, recall of 92.7%, F1 score of 94.96% and Diabetic Retinopathy with precision of 97.0%, recall of 90.9%, F1 score of 93.85%. For the Glaucoma category, the model obtained precision of 85.1%, recall of 87.9% and F1 score of 86.51%. For Normal cases, the model achieved precision of 83.3%, recall of 91.3% and F1 score of 87.11%. This performance shows that the model is very capable of recognizing important eye diseases, particularly cataract and diabetic retinopathy, although some enhancement is required in accuracy for glaucoma and normal cases.In summary, this research designed and implemented deep learning models for automatic detection of eye diseases, and the researcher met the aims and objectives outlined at the beginning of the research.