A major consequence of diabetes, which is the primary cause of vision loss in elderly people, is known as diabetic retinopathy. Early detection and treatment of DR can prevent vision loss, making timely diagnosis crucial. With advancements in deep learning, particularly in deep ensembles, there has been a growing interest in using these techniques for automated DR detection from retinal fundus images. This study explored the use of ensemble learning to improve the accuracy of DR detection and classification. We employ baseline deep classifiers, including MobileNet, ResNet50, VGG16, and Inception V3, and combine their predictions using ensemble learning techniques. The performance of the proposed model can be evaluated using various parameters. The final result of the proposed ensemble model performed better than that of the individual classifiers. The achieved accuracy, precision, recall, and f1-score were 95.00%, 95.12%, 96.23%, and 95.00%, respectively. Ensemble learning has proven to be a favorable approach for enhancing the detection and classification of DR, offering potential for improving patient outcomes in the future.