Machine Learning (ML) and Deep Learning (DL) techniques have revolutionized the detection and classification of eye diseases, significantly improving diagnostic accuracy and efficiency. This study explores the use of Convolutional Neural Networks (CNNs), specifically the VGG-19 architecture, for multi-class classification of ophthalmological images. The dataset used consists of color fundus images, including various eye conditions such as diabetic retinopathy, glaucoma, cataract, and normal conditions. To enhance model robustness and reduce overfitting, a comprehensive image preprocessing pipeline is applied, which includes resizing, color normalization, and augmentation techniques like random rotations, horizontal flips, and brightness adjustments. The VGG-19 model, renowned for its depth and performance in image classification tasks, is employed for feature extraction, with transfer learning utilized to fine-tune the model on the ophthalmic dataset. In addition, ensemble learning techniques are integrated to further boost classification performance by combining the outputs of multiple models. The model’s performance is evaluated through both standard train-test splits and k-fold cross-validation to assess generalization capabilities. Results indicate that the VGG-19 model achieves an accuracy of 85.05% in standard training and improves to 89.3% with k-fold cross-validation, demonstrating a 15% average increase in accuracy. Class-wise analysis highlights the model’s effectiveness in distinguishing between visually similar conditions, such as cataract and normal images. These findings suggest that deep learning systems, particularly VGG-19, can significantly enhance diagnostic capabilities in ophthalmology.