Brain tumors can cause cancer if not detected and diagnosed at early stages. Currently, Brain tumor detection and classification is done by performing Biopsy which is a very time-consuming process. Improvement in technology and Machine learning algorithms can help radiologists in tumor diagnostics in less time and effort. We propose a model that would first segment the MR image and identify the presence of tumor in the brain and if detected then a deep learning-based CNN architecture that would classify the tumors in MRI images into Benign and Malignant tumors and act as a strong base for the staff to decide the curing procedure. The development of the model will be divided into training and testing phases and would be tested using multiple databases and different methods. Having achieved high accuracy, reliability, and execution speed, the developed CNN architecture would act as a strong decision-supportive tool in medical diagnostics for radiologists.