Glaucoma is a progressive eye condition that causes loss of vision and cannot be restored once it has occurred. The diagnosis of glaucoma is an important issue that has to be resolved in the realm of medicine. There have only been a limited number of research efforts aimed at detecting glaucoma in its earlier stages. Nevertheless, the performance of glaucoma illness identification utilizing the various approaches that are now available was ineffective. In addition to this, the time commitment involved in traditional methods of glaucoma illness identification was significantly higher. Finding glaucoma in color fundus photos is a difficult undertaking that calls both knowledge as well as years of practice. Deep learning techniques have rapidly emerged as a method of choice for the analysis of medical pictures. In this study, we took advantage of the application of various Convolutional Neural Networks (CNN) schemes to demonstrate the influence on performance of relevant factors such as the size of the data set, architecture, and the use of transfer learning versus newly defined architectures at an early stage with higher accuracy and a lower amount of time. Specifically, we compared the two types of architectures in terms of their ability to learn from previous examples. Deep learning techniques have rapidly emerged as a method of choice for the analysis of medical pictures. This study examines the primary deep learning ideas that are applicable to medical image analysis and provides a summary of the contributions to the area. We take a look at the application of deep learning to various image processing tasks such as classification, object identification, segmentation, and registration.