In today’s ever-changing world, many people die due to fatal brain tumors as they are not detected in its early stages. These tumors pose a serious threat to people’s life, if treated early can save a person’s life. To address these problem, this paper introduces a deep learning-based web application named “MRI-Mask”, a solution tailored specifically for Brain Tumor localization. In this work, cutting-edge U-Net model is trained on brain MRI images by considering Dice coefficient Loss as the loss function and evaluated on test images. The developed model effectively localizes brain tumor from MRI images and generates mask on it for visualization. The entire deep framework is seamlessly integrated into a web-application for end-to-end localization of brain tumor. The performance of the framework is measured using standard metrics like mean IoU and dice coefficient. The designed framework has achieved an overall mean IoU score of 0.909 and dice coefficient of 0.9362 on publicly available Kaggle dataset, demonstrating its effectiveness in the field of medical diagnosis. The developed MRI-Mask application is publicly released in https://github.com/Adinp1213/MRI-Mask for academic, research and other non-commercial purposes.