A Dynamic Approach of Eye Disease Classification Using Deep Learning and Machine Learning Model

Abstract

AbstractIn today’s world, machine learning and deep learning models are applied in the diagnosis of eye diseases in an automated manner. Cataract, one of the eye diseases, can be especially dangerous, increasing the risk of eye damage. Early detection of Cataract is a requirement for Cataract patients to take the same medications at the same time. We suggest the use of in-depth reading to detect the presence of eye disease. In this research, Convolution Neural Network (CNN) and Support Vector Machine (SVM) were used for cataract detection on a dataset containing normal eye images and images of eye with cataract. Various techniques like data augmentation as a preprocessing step, label encoding, feature extraction have been applied in the research as a part of the model building process. SVM model has provided an accuracy score of 87.5% with an F1 score of 91.3% and CNN model has given 87.08% training accuracy and 85.42% validation accuracy.KeywordsConvolution neural networkData augmentationSupport vector machineLabel encodingFeature extractionPreprocessingTrainingAccuracy