Smartphone-Based AI Detection of Ocular Diseases

Eye diseases are a major global health issue that necessitates prompt detection and treatment to avoid irreversible vision loss. In this research, we describe a unique method for the automatic detection of ocular diseases using a smartphone. Our technology makes advantage of a smartphone’s integrated flash and camera for real-time detection, making it available and practical for consumers. Our strategy intends to address cataracts’ high incidence as Pakistan’s main source of blindness. First, the haar cascade classifier is used in the detection process to identify the eye region from facial images taken by the smartphone. To isolate the important features for further processing, this eye region is then further segmented out. A custom Convolution Neural Network (CNN) is created and trained on a diverse annotated image data set in order to categorize diseases. The CNN model is capable of identifying and classifying ocular conditions like cataracts, conjunctivitis, and healthy eyes. We have created an Android app using Java that incorporates the trained CNN model for remote and real-time applications. Users receive fast feedback on the state of their ocular health thanks to the application’s real-time eye image capturing and instant disease classification. After a thorough examination, our approach was found to be very accurate and effective at detecting eye diseases, addressing the urgent problem of cataract-related blindness in Pakistan.