Retinal Blindness Detection Due To Diabetes Using MobileNetV2 And SVM

Abstract

International Diabetes Federation estimates put the number of diabetics in India at 50.8 million in 2010. and it is estimated to rise to 87.0 million by 2030. One of the most common problems associated with Type 2 diabetes is Retinopathy. Diabetic Retinopathy is a kind of visual loss that affects persons between the ages of 20 and 64. Diabetic Retinopathy puts pressure on the eyeball by shattering the natural flow of fluid out of the eye, harming nerves and leading to glaucoma. If it is detected and treated early, we can reduce the risk of visual loss. However, diagnoses by ophthalmologists involve time, effort, and money, and if computer-aided diagnosis techniques aren't used, misdiagnosis can occur. In recent times deep learning has become the most popular method for obtaining high performance in various fields, even in medical image analysis and classification. The purpose of this research is to anticipate diabetic Retinopathy beforehand in order to avoid future eye problems. The proposed deep learning architecture is based on the Mobile Net architecture, a mobile-friendly, lightweight design that was trained and tested on retinal fundus pictures from the Aptos 2019 challenge data set.