Smart Drunk Driving Prevention Using IoT and Computer Vision

Smart Drunk Driving Prevention Using IoT and Computer Vision 1Soumyadip Sahoo, 2Sohan Das, 3Chanchal Kumar De 1,2UG Students, Dept. of Electronics and Communication Engineering, Haldia Institute of Technology, Haldia, Purba Medinipur, West Bengal, India 3Professor & Head, Dept. of Electronics and Communication Engineering, Haldia Institute of Technology, Haldia, Purba Medinipur, West Bengal, India ABSTRACT Drunk driving is a major preventable cause of road fatalities, prompting increased regulatory attention to in-vehicle impairment detection. This paper presents a low-cost Smart Drunk Driving Prevention System using two independent detection channels. An MQ-3 alcohol sensor interfaced with an ESP32 DevKit V1 detects alcohol above a preset threshold and activates an alarm while inhibiting the ignition through a relay. In parallel, an ESP32-CAM streams driver video to a host application using OpenCV and MediaPipe Face Mesh to calculate Eye Aspect Ratio and Mouth Aspect Ratio for drowsiness and yawning detection. A NEO-6M GPS module provides vehicle location, transmitted to a cloud dashboard during impairment events. Bench tests demonstrated reliable engine locking, drowsiness and yawning detection, and location-data transmission. The prototype costs approximately ₹2,475. Key limitations include MQ-3 cross-sensitivity and dependence on Wi-Fi connectivity.