Advanced Eye Disease Classification using the EfficientNetB3 Deep Learning Model

This research study introduces a novel approach for the classification of ocular diseases that employs the EfficientNetB3 architecture. The backbone of our model is EfficientNetB3, which has a total of 42 layers and a well-optimized layer structure. The model's feature extraction capabilities are optimized for medical image data by utilizing pre-trained weights from ImageNet. The model effectively reduces overfitting and enhances generalization across the dataset by incorporating supplementary layers, including Batch Normalization, Dense, and Dropout. The softmax activation function is employed by the final output layer to forecast class probabilities, which is customized to the precise number of eye disease classes. The proposed model's potential for reliable eye disease classification in clinical contexts was demonstrated by its impressive accuracy of 93.8% on the test dataset.