Early Prediction of Cataract using Convolutional Neural Network

Cataract is one of the most common ocular diseases. It causes cloudy vision and vision loss in a vast number. In this paper, a convolutional neural network model has been proposed to detect cataract in the eyes. This model will help the patient and ophthalmologists with early and accurate cataract detection. This model is built on the VGG19 architecture having roughly three layers. At first, some pre-processing techniques have been applied to the image dataset. After pre-processing, the dataset is split into training and test dataset. The proposed model is trained with the pre-processed training dataset. The datasets for experimentation are downloaded from an online repository. A detailed analysis is done based on the predicted output and actual class label. The overall accuracy of the model is 97.47%.