Fundus imaging is a valuable tool in the armamentarium of ophthalmologists for diagnosing and managing glaucoma. It provides objective and visual evidence of structural changes in the eye, helping clinicians make more accurate and timely diagnoses. In this work, five datasets were used to detect the Glaucoma such as HRF, DRISHTI-GS1, RIM-ONE, ORIGA, and Private. In order to enhance the classification process, augmentation is applied and for preprocessing steps, spatially weighted Fuzzy C-Means Clustering technique is used to segment the optic disc region from fundus image. Semi-supervised Generative Adversarial network is used to classify the Glaucoma from fundus image. The sensitivity, accuracy, and specificity were analyzed from five proposed datasets. In this work, five different datasets (HRF, DRISHTI-GS1, RIM-ONE, ORIGA, and Private) are used to evaluate certain parameters such as accuracy, sensitivity, and specificity. Specifically, in private dataset, the calculated values for accuracy, sensitivity, and specificity are 97.81, 99.02, and 98.81%, respectively.