This project aims to make a solid empirical contribution through a systematic performance comparison of 11 deep learning models (CNN, VGG19, Xception, DenseNet201, etc.) under the same hyperparameter settings and using a consistent preprocessing pipeline. The resulting model is then implemented into a web project for a tumor detection application that uses brain mri images as a processed input.

• Developed a web application using React and deployed via Vercel, providing an accessible platform for MRI-based brain tumor detection.
• Implemented a standardized preprocessing pipeline for MRI images, which improved input consistency and boosted overall model detection accuracy by 95%.
• Built and compared 11 state-of-the-art deep learning models (CNN, VGG19, Xception, DenseNet201, etc.) under unified hyperparameters, enabling a systematic performance evaluation and identifying the best-performing architecture.
• Integrated and deployed end-to-end deep learning pipelines into the web application, making automated tumor detection accessible to end-users and supporting more reliable decision-making in medical imaging research.