An Accurate Assessment of Diabetic Retinopathy using a Deep Learning based Approach

Diabetic retinopathy is a disorder that damages the eyes and is an outcome of diabetes. High blood sugar levels damage the blood vessels in the retina, which is the lightsensitive tissue in the posterior of the eye. If not treated, this injury might result in visual issues and perhaps blindness. The datasets used in this investigation is from the IDRiD to investigate the efficacy of Convolutional Neural Networks (CNNs) in diagnosing Diabetic Retinopathy (DR) in fundus pictures. As the accuracy of CNN-based detection relies heavily on the availability of large datasets, leveraging this data becomes crucial owing to high prevalence of diabetes and its associated risk of DR. By employing CNNs, this research aims to optimise the efficiency and accuracy of DR detection compared to manual diagnostics, thereby saving both time and resources. The implementation utilizes Python programming language along with libraries such as Keras, OpenCV, and Numpy, with a focus on image-classification to distinguish between DR and non-DR cases.