This document describes a study that used deep learning and the VGG16 model to develop an artificial intelligence system for detecting glaucoma from fundus images. The system first preprocessed images then extracted features using VGG16. It was trained on labeled images to classify images as glaucoma or normal. The proposed system achieved 99.8% accuracy, outperforming previous methods. It provides an effective way to diagnose glaucoma from fundus images using deep learning.