Drive-Alert: Real-Time Drowsiness Detection Mobile Application
| Title of the Project |
“Drive-Alert”: Real-Time Drowsiness Detection Mobile Application |
| Students Details |
202110929 Lujain Adnan Thouqan |
| Abstract |
Driver drowsiness is a leading cause of road accidents accounting for almost 21% of all fatal collisions worldwide. We suggest Drive Alert, an AI-powered smartphone app made to track driver alertness in real time using just the front-facing camera of the phone in order to solve this increasing safety issue. Unlike conventional solutions depending on specialized sensors or hardware, our application uses deep learning to differentiate between "Alert" and "Drowsy" states. Alert states call for no intervention; if drowsiness is detected, a loud alarm is triggered to wake the driver. Deep learning models trained using TensorFlow and Keras; face detection and segmentation using YOLO; facial landmark extraction via MediaPipe FaceMesh; and image handling via OpenCV form the strong stack of open-source technologies upon which the system is based. Scalable and effective inference is handled on the Google Cloud Platform (GCP). React Native with Expo develops the mobile app in React Native using TypeScript in a modular, component-based architecture together with JavaScript. It combines Expo AV to set off alarms, TensorFlow Lite (TFLite) for on-device inference, react-native-vision-camera for live face capture, and Reload for state management. Emergency warnings are sent via Expo SMS or push notifications. Drive Alert offers a low-cost and easily available solution to decrease accidents related to drowsy driving, improving road safety and saving lives. |