An Enhanced Approach for Detecting and Classifying the Diabetic Retinopathy by Utilizing an Optimized Convolutional Neural Network

Diabetic Retinopathy (DR) is a condition that develops due to diabetes-related harm to the retina. It affects a significant ratio, around 80 percent, of individuals who have had diabetes for more than one decade. Regrettably, the areas needing Diabetic Retinopathy detection lack the essential expertise and equipment. Manual diagnosis of DR can be tedious and error-prone. As a result, numerous artificial neural network based systems have been developed to automate the detection of DR and its various stages using retina images. However, these methods often need help to capture the complex underlying features and exhibit biased accuracy, especially when classifying the early stages of DR. we have implemented an optimized Convolutional Neural Network model. We trained our model with various hyperparameters and finalized the best-performing model with an accuracy of 0.91.