EfficientNet Model for Detecting Glaucoma on the PAPILA Dataset

Glaucoma is the main cause of global vision loss, emphasizing the need for accurate and early detection methods. This study applies EfficientNetB7, a state-of-the-art convolutional neural network (CNN), to classify retinal fundus images from the PAPILA dataset, a challenging real-world dataset with limited size and diverse image quality. The study shows how data balance and region-of-interest (ROI) cropping are two preprocessing methods that can help improve classification performance. The suggested model did better than previous studies, with an AUC of 0.84 and an F1-Score of 0.77 after 10-fold cross-validation. Statistical research shows that combining preprocessing methods can make models more reliable. These findings demonstrate EfficientNetB7’s potential to address challenges in small, imbalanced datasets, paving the way for scalable applications in glaucoma diagnosis and medical imaging tasks.