Brain tumors pose significant challenges in healthcare due to their complex origins and diverse range of neurological symptoms. It's crucial to quickly and accurately identify and classify these tumors into class such as glioma, meningeal and pituitary tumors to enhance treatment effectiveness and improve patient outcomes. This work investigates the use of sophisticated deep learning algorithms to automate the identification and categorization of brain tumours using 2-D magnetic resonance imaging (MRI) data. The approach includes a two-phase deep learning framework. Initially, the model determines whether a tumor is present. Subsequently, it classifies any detected tumors into their specific types. This work evaluates two advanced deep learning architecture: Convolutional Neural Network (CNN),VGG-16 model, EfficientNetB3 algorithm. Both are trained and tested using a dataset from the National University of Singapore, which includes MRI scans of both healthy subjects and patients diagnosed with glioma, meningioma, or pituitary tumors. Additionally, the research investigates the role of three-dimensional image processing not only in detecting tumors but also in accurately localizing them through the use of bounding boxes. The comparative analysis aims to selecting most ideal algorithm architecture for detecting and classification of brain tumor using 2-D image data. The results of this works deals with could dramatically transform the field of neuro-oncology by enabling faster and more precise diagnoses, improving patient management, and aiding surgeons during tumor removal procedures through accurate tumor localization.