Development of CNN-Based Feature Extraction and Multi-layer Perceptron for Eye Disease Detection

Identification of multiple eye disorders utilizing a multi-label categorization methodology is an effective way. The crucial merit of this technique is that it can identify the disorders in earlier times. Ocular disease affects millions of people worldwide, and early detection and treatment of these abnormalities are crucial in preventing avoidable blindness. Diagnosing eye illnesses accurately necessitates the analysis of a diverse array of visually discernible symptoms associated with these conditions. The wide range of symptoms exhibited by various eye illnesses emphasizes the importance of a comprehensive assessment for an accurate diagnosis. Because of the moderate progression, the disorder gives several indications in the initial times, so creating disorder detection is a complex work. To achieve an effective identification system, architecture for a machine learning-assisted method is suggested. Initially, the required eye images are gathered from the standard online data sources. It is then followed by Convolution Neural Network (CNN)-based feature extraction, where the features are extracted by Visual Geometry Group 16 (VGG16). Finally, the attained features are subjected to the Multi-Layer Perceptron (MLP) for detecting the different eye disorders. The performance analysis is conducted contrast with other conventional models to prove the developed model efficacy.