Identification of brain tumours is an important issue with medical imaging that needs accurate and efficient classification for timely diagnosis and treatment. CNN often struggles to extract features from complicated medical datasets. This to take characteristics out of paper provides an advanced deep learning method by integrating YOLOv11 with a transformer-based detecting head that uses Swin Transformer as the backbone. Through transfer learning, the model benefits from pre-trained Swin Transformer weights, improving feature extraction while reducing computing costs. The suggested model enhanced a brain cancer dataset with bounding-box labels and effectively classifies pituitary tumours, gliomas, and meningiomas. Multiple studies demonstrate that this design is a powerful tool for medical diagnostics since it improves tumour location, speeds up training, and boosts accuracy.