• Manas Negi
• Abdul Aziz Kanoor
• Femenca Noronha

This research conducts a detailed comparative analysis of deep learning models for automated detection of ocular diseases using the Ocular Disease Intelligent Recognition (ODIR) dataset. The dataset includes fundus images from 5,000 patients, categorized into eight diagnostic labels. Early diagnosis of ocular conditions is necessary to avoid permanent vision loss, and automated systems offer a promising complement to clinical diagnostics. The study evaluates four advanced convolutional neural network (CNN) architectures—VGG16, ResNet50, DenseNet121, and InceptionV3—through a standardized preprocessing pipeline. This pipeline ensures uniform image resolution, applies customized data augmentation, and normalizes pixel intensities to address variations in retinal images from diverse clinical and imaging setups. GPU-accelerated training leverages mixed precision and TensorFlow’s tf.data API for efficient data handling. Performance metrics and computational efficiency are assessed across models. DenseNet121 achieves the best diagnostic accuracy and ROC AUC with fewer parameters, while VGG16 balances accuracy and computational efficiency. ResNet50 and InceptionV3 exhibit longer training times and occasionally lower accuracy. These results show trade-offs between model complexity, resource demands, and performance metrics such as accuracy, precision, recall, F1 score, and ROC AUC. This work offers important understanding for choosing appropriate models for clinical settings with limited resources by examining diagnostic performance in conjunction with variables including training time, parameter count, and model depth. The results set a standard for creating effective and trustworthy computer-aided diagnostic instruments in ophthalmology.

Ocular Disease Detection, ODIR Dataset, Fundus Imaging, Deep Learning, Convolutional Neural Networks (CNNs), ResNet50, VGG16, DenseNet121, InceptionV3, Data Augmentation, GPU Acceleration, Model Comparison.

"Comparative Analysis of Deep Neural Networks for Automated Ocular Disease Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 3, page no.a136-a145, March-2025, Available :http://www.jetir.org/papers/JETIR2503018.pdf

"Comparative Analysis of Deep Neural Networks for Automated Ocular Disease Detection", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 3, page no. ppa136-a145, March-2025, Available at : http://www.jetir.org/papers/JETIR2503018.pdf