Driver fatigue is one of the leading causes of road accidents worldwide, posing a serious threat to transportation safety. In this research, we present a lightweight and effective deep learning-based approach for real-time driver fatigue detection using face-centric transfer learning. Our method utilizes MobileNetV2 as the backbone for feature extraction, leveraging its computational efficiency while maintaining high classification performance. We integrate face detection using MediaPipe to crop and focus solely on facial regions, enhancing the model’s ability to distinguish drowsy states from alert ones. The system is trained and evaluated on a combined dataset sourced from publicly available driver drowsiness datasets, ensuring robust generalization across varied subjects and lighting conditions. Our experiment achieved over 90% accuracy with high precision and recall. It offers good balance between speed and accuracy proving that it can be used in real time drowsiness detection.