A brain tumor is a group development of abnormalities brain cells. There are numerous forms of brain tumors. Some brain tumors are cancerous, whereas others are noncancerous. Brain tumor identification involves several phases, including the capture of an input MRI image, the conversion of the input image to a grayscale image, the application of filters, segmentation, feature extraction, and classification. The detection of a tumor is a difficult process. The position, size, and shape of the tumor differ greatly from patient to patient, making segmentation a difficult process. The detection of a tumour is a difficult process. The position, shape, and structure of the tumour vary significantly from patient to patient, making segmentation a difficult process. A Nine Layer CNN architecture including an input layer, zero padding, Conv2D, Batch Normalization, Re-Lu, Max pooling, Max pooling, Flatten, and Dense layer is designed in this study. TCIA Brain tumor dataset is used to train the Nine Layer CNN. TCIA dataset is augmented to overcome overfitting circumstances. In order to overcome overfitting conditions, the TCIA dataset is augmented. CNN nine layer produced a decent outcome, with a training accuracy of 98.93%. If the classifier determines that the picture is a tumor present image, it will also provide the proportion of the tumor.