AUTOMATED CONJUNCTIVITIS DETECTION USING XCEPTION MODEL WITH TRANSFER LEARNING

Conjunctivitis, an inammation of the conjunctiva, is a common eye condition often diagnosed through visual inspection, a method prone to subjectivity and inefciency. This study explores the potential of deep learning, specically the Xception convolutional neural network (CNN) architecture, for automated conjunctivitis detection. A dataset of eye images, compiled from a published research paper and supplemented with internet sources, was used to train and evaluate the model. The Xception model initialized with pre-trained ImageNet weights and ne-tuned for conjunctivitis classication. Data augmentation techniques were employed to enhance the model's robustness. Results demonstrated promising performance, with the model achieving a test accuracy of 95.16%, an AUC-ROC score of 0.9484, a precision of 0.9773, a recall of 0.9412, and an F1-score of 0.9143. While training and validation metrics suggested potential overtting, the model's robust performance on the test set indicates the potential of deep learning for automated conjunctivitis detection, offering a promising avenue for more objective and efcient diagnosis.