Detection and Classification of Diabetic Retinopathy Using Pretrained Deep Neural Networks
Abini M. A, S. Sridevi Sathya Priya
Abstract
The retina is harmed by diabetic retinopathy (DR), a consequence of diabetes. Up to, 80% of people who have had diabetes for ten or more years are affected by this type of medical issue. Where the need for diabetic retinopathy identification is greatest, there is frequently a shortage of the necessary knowledge and tools. The majority of research in the area of diabetic retinopathy has relied on manual feature extraction or disease identification. So, the goal of this research is to make a deep learning neural network that can identify the disease in all of its forms. The suggested system enables a DR classification that accounts for normal eyes, mild DR, moderate DR, severe DR, and proliferative DR, which may aid ophthalmologists in making a preliminary decision. Our accuracy rates for estimating the degree of diabetic retinopathy from an image were 90% and 92% respectively, using the pre-trained Convolutional Neural Network (CNN) VGG-16 and MobileNet-V2.