Detection and Classification of Diabetic Retinopathy using Deep Learning

Abstract

The most common reason for blindness in adults in developed nations is Diabetic Retinopathy (DR). Currently, diagnosing DR involves an in-depth arduous examination of digital colour fundus pictures of the retina by a qualified practitioner. By looking for lesions connected to the vascular anomalies brought on by the illness, ophthalmologist can recognise diabetic retinopathy. Although this strategy works, it has substantial resource requirements. It has long been understood that a thorough and automated approach of detecting diabetic retinopathy is necessary, and prior initiatives have achieved excellent strides utilising image classification, pattern recognition, and machine learning. This project seeks for automated detection, grading, and segmentation of Diabetic Retinopathy. In our project we aim to improve image segmentation using UNet and to automise the project using Convolutional Neural Networks and VGG16.