Retinal Disease Detection Using Deep Learning and ResNet Model

Retinal diseases like diabetic retinopathy, age-related macular degeneration, and glaucoma cause widespread vision loss, but early detection is limited by specialist shortages. This paper introduces a web-based system using fine-tuned ResNet50 for multi-class retinal disease classification from fundus images. Built with HTML/CSS/JavaScript (Tailwind CSS) frontend, Flask/Python backend, PostgreSQL and TensorFlow with OpenCV preprocessing, it enables secure image uploads and real-time predictions. Experiments on datasets like APTOS achieve 92% accuracy, outperforming baselines in precision, recall, and F1-score. The system supports scalable deployment in underserved areas, reducing diagnostic delays and aiding blindness prevention.