A Multifeatured Based Diabetic Retinopathy Detection Using Hybrid-Net

Diabetic Retinopathy (DR) is a diabetic complication caused by blood vessel changes in the retina, and it can lead to blindness in the developed world. Here for this research authors have used a data set which consists of both diabetic and non-diabetic images. Since normal CNN wouldn’t efficiently identify various underlying patterns, authors have opted for VGG-16, VGG-19, Inception and Hybrid Net model in this paper. In this paper, we consider the input images and pre-process them and convert all the images into an array representing their pixels. With the help of various algorithms we extract the primitive and advance features which help in recognising the disease at an early stage and hence helping ophthalmologist for better diagnosis and vision rescue for diabetic patients. The purpose of this research is to develop an automated and efficient system that can detect DR symptoms in seconds from a retinal scan and expedite the picture evaluation and diagnostic process. The authors have compared various algorithms and Hybrid Net has the highest accuracy among all.