Diabetic Retinopathy Using AI

Diabetic retinopathy (DR) is one of the side effects of diabetes. Before the complaint progresses far enough, it largely goes undiagnosed. Because of this, precluding vision loss requires early discovery. This problem can be answered only by continuously covering the glucose position and early discovery of complaint and its timely treatment. Anon-invasive glucose monitoring system has been enforced for the observation of glucose position trends. A deep literacy model to speed up complaint discovery is presented in this paper. The Kaggle Diabetic Retinopathy Detection database was used to examine a dataset made up of thousands of fundus photographs of the retina taken under various imaging settings. After testing and retraining the model will be used to develop the system that have a stoner-friendly hindrance. This provides an automated system that will help ophthalmologists to grade the fundus images as No DR, Mild, Moderate, Severe and Proliferative DR. A system that can efficiently train a deep neural network for semantic segmentation of fundus pictures has been proposed as a prognosis of Microaneurysm and early opinion system for non-proliferative diabetic retinopathy. This system can boost the efficacy and delicacy.