Glaucoma is an ocular disease primarily caused by increased intraocular pressure (IOP), with symptoms observed only in the later stages. The evaluation of intraocular pressure and fundus examination are the main screening tests used to identify glaucoma. Through imaging of the optic disc, signs of glaucoma can be detected, such as widening the cup region and disorganizing blood vessels. Therefore, this study aims to construct models for segmentation and classification to assist healthcare professionals in the early diagnosis of glaucoma. For segmentation, YOLO v5, v7, v8 and v9 models are evaluated to detect the optic disc and cup separately, followed by obtaining ellipses around these regions. Subsequently, geometric information is calculated from the images and classified using machine learning methods to distinguish images of patients with glaucoma from those without it. The best localization and segmentation results are obtained with YOLO v8, achieving Hausdorff, IoU, and DSC scores of 4.092, 0.819, and 0.900 for the optic disc and 3.634, 0.717, and 0.835 for the optic cup, respectively. The classification is achieved with Extreme Gradient Boosting (XGB) achieving accuracy of 84.5%.