Brain tumors are a major health hazard, and early detection is important for increasing patient survival. This work proposes a new deep learning-based approach that employs CNNs combined with ensemble learning strategies to achieve enhanced accuracy and efficiency in the classification of brain tumors from MRI scans. The novel method synergistically combines features derived from CNN with conventional machine learning classifiers for maximum detection accuracy. Using publicly accessible datasets, the model is rigorously tested by performance measures of accuracy, sensitivity, specificity, and F1-score. It is found that an ensemble learning strategy involving CNN and Support Vector Machines (SVM) surpasses traditional deep learning structures in terms of performance, reaching a 99.83% accuracy. The current work adds to medical AI in terms of offering a highly scalable and accurate approach to diagnosing brain tumors.