Early identification of brain tumors remains essential for better patient results because traditional detection approaches face problems due to their lengthy processing time and human errors. The WHO reports that brain and central nervous system tumors affect 2% of worldwide cancer cases, with India and Bangladesh demonstrating high incidence rates within South Asia. MRI serves as the standard diagnostic tool for medical professionals, although subjectivity in analysis from radiologists can delay and distort diagnostic outcomes. Comprehensive research evaluates brain MRI image detection using Convolutional Neural Networks combined with Transfer Learning through MobileNetV2 to achieve efficient tumor diagnosis. Researchers applied 800 MRI scans to run their experiments, involving CNN model fine-tuning along with MobileNetV2 model pre-training to identify healthy and malignancy-infected images. The model reviewer adopted multiple performance metrics to evaluate the outcomes, such as test accuracy combined with precision, recall, F1-score, and Area Under the Curve (AUC). A test accuracy of 97.50% emerged from MobileNetV2, surpassing the 92.50% test accuracy of CNN. The MobileNetV2 model shows exceptional competency in generalization, which leads to high precision values and low numbers of both false positives and negatives, ensuring that it could effectively serve in clinical brain tumor detection. This study shows how deep learning approaches with transfer learning boost medical image diagnostic precision, thereby establishing new prospects for automatic healthcare systems development.