Diabetic Retinopathy Detection Smartphone App Using Tensor Flow

Abstract

Diabetic retinopathy is a medical term for  a disorder that damages the retina and may cause  blindness. The illness is most common in people in their  working years, and it may lead to severe eyesight loss if  left untreated. The continual execution of deep neural  organizations on cell phone stages to distinguish and  organize diabetic retinopathy from eye fundus images is  presented in this study. This execution builds on a  previously announced execution by considering each of  the five phases of diabetic retinopathy. Binary deep neural  organisations are built at principal, one for recognising  four phases and the other for further defining the last step  into additional two stages, using motion learning and  fundus photos from the Eye PACS and APTOS datasets.  Then it is shown how these prepared organizations are  transformed into a mobile phone program, with Android as  well as iOS modifications, to deal with images captured by  cell phone cameras on a continual basis. The application is  planned so that fundus pictures can be caught and handled  progressively by cell phones along with focal point  connections that are industrially accessible. The grew  constant cell phone application gives a cost effective and  generally open methodology for leading first-finish  diabetic retinopathy eye tests in quite a while or regions  with restricted admittance to fundus cameras and  ophthalmologists.