It is true that better patient outcomes and efficient therapy depend on the early identification and categorization of brain tumours. Recent developments in deep learning, especially in object detection frameworks like Faster R-CNN, have a lot of promise for medical categorization through the use of a modified Faster R-CNN model. By improving the region proposal process and fine-tuning the backbone network for the unique characteristics of brain tumour imaging, our method outperforms previously employed approaches in terms of computational efficiency, accuracy, sensitivity, and specificity. To illustrate the advantages of the suggested framework, a comparison with current methods is given. There is also discussion of possible clinical linkages and future possibilities.