Retinopathy Detection on Diabetic Patients through Transfer Learning based Inceptionv3 Algorithm

Diabetic retinopathy is a most common intricated disease for all peoples, from children to adults facing the eye problems. It damages the blood vessels in retinal region and leads to affect the vision. Based on the diabetic level, the stage of the vision from the eye can be categorised. It can to segregated into five classes, which are No_DR, Mild, Moderate, Severe and Proliferate. By using Deep learning Techniques these classes classify the images on collected dataset of 3724 retinal images as input and producing the retinal disease of diabetic problem gives the output respectively for these classes. The proposed model modified inception v3 been developed to improve the accuracy and lower the loss ratio compared to the existing models. The existing models that are close to the proposed model of modified inception v3 with optimizer, like DenseNet121, VGG16, ResNet50, and InceptionV3 are compared with the proposed model to obtain the accuracy.