Brain tumors whether benign or malignant can be life-threatening as the Brain is the most crucial organ of the nervous system in a human and it is responsible for controlling all essential processes required for humans to function. This research paper investigates different Convolutional Neural Networks such as VGG19, ResNet152V2, and DenseNet201, then proceeds to apply the concepts of fine-tuning and certain image processing techniques to process the dataset and provide a comparison of how different deep learning models perform. The research categorizes brain tumors into benign, malignant, and three subtypes: Glioma, Meningioma, and Pituitary, with a ‘normal’ class for tumor absence. The research also aims to dive deep into applying certain image processing methods that can enhance the results obtained from training the models. ResNet152V2 has been identified as the proposed model for this research as it achieved an accuracy of 98.18% which is the best result among all the other models.