Globally, glaucoma is a leading cause of irreversible visual loss. Due to the lack of symptoms in the early stages of the disease, glaucoma is typically not diagnosed until severe vision loss has occurred. One of the most common ways to diagnose glaucoma is with a comprehensive eye exam. However, a substantial commitment of time, money, and specialist equipment and personnel is necessary to carry out such investigations. Using deep learning optic cup and disc segmentation based on retinal fundus images, this work aims to develop and evaluate the effectiveness of a novel, affordable glaucoma screening tool. The research made use of  ensemble learning technique to enhance semantic segmentation models. A number of recognized deep learning architectures, including Unet 3Plus, Deep Lab V3P, PSPNET and UW-Net, are combined in the proposed semantic ensemble segmentation model. Ensemble combination involves merging predictions from several models using a weighted averaging technique that considers the accuracy and dependability of each individual model. Metrics such as accuracy, specificity, sensitivity, area under the curve, intersection over union, dice coefficient, and f1-score are used to evaluate the performance in the study. The proposed model was validated on three publicly available datasets, namely ORIGA, REFUGE and RIM-ONE DL. The experimental results show that the suggested method can estimate the Cup to Disc Ratio CDR for thorough glaucoma screening, and it is on par with the state-of-the-art architecture utilized for optic disc and cup segmentation.