Improving Detection of Eye Cataract Disease by Transfer Learning Technique with Convolutional Neural Network

One of the most usual eye diseases is cataracts and individuals get during their aged and it happens whenever the eye lens forms a fog. Major symptoms and signs of the disorder include blurry vision, difficulty in looking bright light and fading colors. These symptoms are commonly associated with difficulty accomplishing a variety of tasks. Thus, early cataract detection and treatment may help to reduce the incidence of blindness. The present research employs Convolutional Neural Networks (CNN) to classify cataract illness using a publicly accessible image dataset. The suggested CNN combines the MobileNet as Transfer Learning (TL) benefits with CNN as pre-trained for obtaining better feature understanding of data that correlate among the patches. When compared to traditional diagnosis techniques, image classification using CNN is a potentially high-performing and cost-effective solution. Thus, the current research aims to develop a technique for cataracts prediction. Normal human being may have difficulty in detecting early and slight modifications in the optic disc. This research work has access in identifying the cataract using fundus images are identified with respect to a complete ophthalmologic examination. Hence, the encoder of Deep Learning (DL) may be trained basic features in fundus images for individuals with early stage cataracts.