An abnormal growth of cells or tissues of mass amount growing inside the brain is called brain tumor, which is harmful to humans, as the tumor can be both cancerous and non-cancerous. Early diagnosis and accuracy of brain tumor helps the neurologists to initiate the treatment to their patients at early stages. To provide a detailed view and diagnosis of the brain tumor, MRI is commonly used. Deep learning is the process of teaching how the computer should perform and act in a way the human brain does. In deep learning, the neuron structure, which is this is inspired by a human brain is represented as the artificial neural network, as how the neurons play an important role in the human brain. The study of deep learning is applied in various fields as it gives a high level of accuracy. A Convolutional Neural Network algorithm is proposed to detect and classify the brain images in MRI scan efficiently. In the given MRI image, if the brain image has no tumor, then it is defined as normal, otherwise, if a tumor is predicted, then it is further classified into the tumor types such as Glioma, Meningioma and Pituitary tumor. The model result shows better accuracy of 98% in detecting the tumor from the MR image.