Abstract

This paper presents an innovative approach for addressing the critical issue of road safety. The proposed solution involves the development of a Driver Monitoring System (DMS) leveraging advanced technologies including Python programming, Raspberry Pi, and YOLOv5 (You only look once) for real-time object detection. The Objective is to consistently monitor vital behaviors such as facial expression, signs of drowsiness and yawning, seat belt usage and compliance with safety guidelines regarding use of mobile phone and earphone while driving. The key features of this paper are Drowsiness and yawning detection, Alcohol detection, Accident detection, Seat belt detection, mobile phone detection and earphone detection. Raspberry Pi is a versatile and affordable SoC (System on chip) used to control the chassis for the movement and GPIO (General-purpose input or output) interface enable the integration of sophisticated sensors like the MQ3 gas sensor for alcohol detection and the MPU6050 gyroscope for accident detection. Utilizing Haar Cascade classifiers, this system performs facial analysis to detect signs of drowsiness and yawning, facilitating real-time monitoring of driver alertness. The YOLOv5 model is trained using a diverse dataset of annotated images to improve detection accuracy for these specific behaviors. Once trained, the YOLOv5 model is deployed to detect seat belt usage, mobile phone presence, and earphone usage in live video streams. Immediate auditory and visual alarms are triggered to alert drivers of unsafe behaviors. This article represents a significant contribution to road safety technology, offering an integrated solution to address key aspects of driver behavior and promote safer driving practices.