Cataract Detection using Deep Learning Model on Digital Camera Images

Cataracts are one of the most prevalent visual diseases that people get as they gets older. A cataract is a fog that forms on the lenses of our eyes. The main symptoms of this illness include dim view, colorless, and difficulties in watching a daylight. Slit lamps and fundus cameras are routinely used to detect cataracts, although they are both expensive and require domain knowledge. As a result, the shortage of skilled ophthalmologists may cause cataract identification to be delayed, necessitating medical treatment. Consequently, early detection and prohibition of cataracts might assist to reduce the frequency of occurrence of blindness. Hence the goal of this study is to utilize Convolutional Neural Networks (CNN) to diagnose cataract pathology using a publicly available Digital Camera Image dataset. The CNN cycle takes a considerable amount of time and expense. As a result, optimization will take place. It can increase accuracy while also reducing processing time. In this study the proposed model consist of three Convolutional layers, three pooling layers, one flatten layer, and two dense layers with an ADAM optimizer. The proposed CNN model can detect cataracts with a testing accuracy of 0.9925 with a loss of 0.0475, and a training accuracy of 0.9980 with loss of 0.0038, for the selected Digital Camera Images Dataset.