Conceptualization and Construction of ALERT: Alcohol-Related and Eye Closure Risk Technology

The project features a comprehensive alcohol detection and driver monitoring system designed to improve road safety. The system has four different models, each of which addresses different aspects of alcohol impairment. In the static model, an alcohol sensor is used during ignition, allowing the vehicle to be activated only if the driver's alcohol content is within the norm. Simultaneously, the dynamic model includes a wristwatch with a sweat alcohol sensor, providing an alternative measurement method with a nominal variance of 10%. The image processing model employs YOLOv8 to detect alcohol bottles near the driver, causing the vehicle to stop automatically upon detection. Additionally, the sleepiness detection model uses image processing to identify signs of drowsiness, which may be exacerbated by alcohol consumption. The outputs of these models are integrated into a central machine learning model, where the results of the static and dynamic models are prioritized, while sleepiness and object detection serve as supplementary parameters to enhance prediction accuracy. Despite the limitations of individual models, the synergistic integration of these approaches aims to create a robust system that promotes safer road conditions.