Using a Machine Learning Algorithm and Fundus Images to Treat Diabetic Retinopathy

Diabetes-related complications include diabetic retinopathy (DR), a condition that damages the eyes. Pictures of the retinal fundus can be used to find it. Traditional methods for processing photos have several limitations when it comes to analyzing data and hiding patterns. In recent years, deep neural network have to be effectively used in a variety of computer vision tasks. On the openly avatlable retinal fundus picture available on Kaggle dataset, we have to apply the pre-trained images in Convolutional Neural Network (CNN), MobileNetV1 and VGG16 in this research. In favor of to get larger and clear the of feature extraction, we have used pre-processing and augmentation approaches. 35,126 retinal eye funds images from the Kaggle dataset with five labelsof class are used in the experiment. OnEary trained MobileNet and V1VGG16 with learning are transferred and Tuning clearly, training accuracy was 91.42% and 91.84%, respectively. VGG16 and MobileNetV1 both achieved test accuracy of 91.51% and 91.77%, respectively.