Brain tumors are the most devastating disease, leading to a very short life expectancy in their highest grade. This paper aims to present a binary as well as a multiclass brain tumor classification algorithm using deep networks such as CNN and ResNet 50 on different datasets of Magnetic Resonance images. Deep learning is a new machine learning field that gained lots of development in the medical field for the last few years. MRI has become an effective tool for medical research in recent years and found itself in applications such as brain tumor classification.