A Deep Learning-Based Framework for Early Detection of Diabetic Retinopathy

An adverse effect of diabetes that affects the eyes by harming the blood vessels in the retina is called Diabetic Retinopathy (DR). At first, it either has no symptoms or causes intermittent vision issues. As it worsens, it affects the two eyes and ultimately results in either partially or total blindness. mostly happens when blood sugar levels are too high to control. As a result, the risk of developing diabetes mellitus perpetually high, for those who already have the condition. The chance of perpetual sight loss can be prevented by initial identification. Hence, the important identification approach is needed. The present research examines a Deep Learning (DL) approach, namely the EfficientNet-B7 classification, which is a Convolutional Neural Network (CNN) model used for diabetic retinopathy detection at an early stage. The datasets that are used is Diagnosis of Diabetic Retinopathy Detection via Kaggle. The suggested method was attained the accuracy model of 99.39 %. This work's primary goal is to create a reliable system for automatically identifying DR.