Brain tumour refers to an abnormal cell growth in the human brain. For the purposes of plans for therapy and testing, categorization the brain tumours is extremely important. There are many types of tumors are there in that Gliomas, meningioma, pituitary gland tumors are considered as primary brain tumours of the most typical form and these are rated as per the World Health Organization's grading system. In the proposed model uses deep learning techniques, particularly Convolutional Neural Networks (CNNs), have shown promising results in various medical imaging applications. The dataset used for training and evaluation consists of a large collection of brain tumor images obtained from diverse sources, including Magnetic Resonance Imaging (MRI) scans If a brain tumour is anticipated, its location and size can be determined, and the brain tumour is removed.. The images are pre-processed to enhance features and normalize intensities. The CNN models are trained on a high-performance computing system using the labelled dataset. We studied a brain tumor dataset of 3062 images with Meningioma, pituitary, glioma, and no tumour are a few examples of the various forms of brain illnesses. The suggested DenseNet, VGG16, and VGG19 training accuracy were determined to be 94.24%, 92%, and 90%, respectively.