ABSTRACT
The novel technique to diabetic retinopathy (DR) detection is the main emphasis of this work. With the use of a specially designed Convolutional Neural Network (CNN) trained on a carefully chosen dataset. It was possible to accurately classify whether DR is present. A well-designed CNN architecture and meticulous preprocessing go a long way toward the model's excellent performance. Real-time DR forecasts are made possible by the research's innovative creation of an intuitive online application. The model's efficacy is demonstrated by evaluation on a different test set, underscoring its potential for early DR detection in clinical settings.
