Brain tumour is one of the most fatal diseases and for cure early diagnosis and treatment are required. Manual classification of MRI is complex. It takes lot of time. The problem here is to automate classification of brain tumours using different machine learning algorithms and image processing techniques. This paper presents automated method using the VGG19 model for feature extraction. GANs are used for data augmentation. They are also used for enhancement. LightGBM serves as classifier. This achieves overall accuracy of 98.89%. Without GANs accuracy was 94.32%. Without classifier it was 94.76% We also reference state-of-the-art techniques and architectures. Validate the efficacy of our approach against existing methods.