Deep Learning Based Edge Device for Diabetic Retinopathy Detection

Diabetic Retinopathy is the major problem in Diabetic affected people, as it causes blindness to the people who are affected by this particular disease. Recently many people are being affected by Diabetes due to changes in food habits and the quality of food people take. Early and continuous testing of the eye in regular intervals leads to the identification of this disease by which the blindness problem can be overcome, but due to lack of availability of resources like optholomists and the machinery for monitoring the eyes the early detection is not that easy. The early machines and methods adopted will not provide good results in any cases because of many constraints, As the technology is advancing, with the help of neural networks and Edge Devices regular monitoring the diabetic people can be done easily and effectively in any part of the globe with much ease. At present image processing techniques are being used. In this study, we propose Deep Learning algorithms that process the date very accurately in very less time because of which we could good results in performance evaluation. In this study we have also proposed Edge Device which are end term devices, where these devices perform analytics on images of the eye and predict the condition of the eye and also it gives the information of stages of Diabetic Retinopathy, we can also store the data which is processed by edge device on cloud servers via wifi, From the cloud server the concerned people can obtain the records of that person who is affected with Diabetic Retinopathy.

Keywords: Convolution Neural Networks(CNN), Diabetic Retinopathy, Edge Device, Neural Networks.