Driver behavior monitoring system

Abstract

Road safety remains a critical concern in Uganda, especially within the public transport sector where driver distractions contribute significantly to accidents and fatalities. This project presents a Driver Behavior Monitoring System designed to detect and log unsafe driving behaviors such as mobile phone usage, drowsiness, putting on seatbelts, and smoking while driving. The system combines a lightweight machine learning model based on the YOLOv8 architecture, embedded within a mobile application built using Flutter. The app utilizes the smartphone’s front camera to perform real-time behavior detection locally using PyTorch Lite, eliminating the need for constant server communication. Detected behaviors are logged with timestamps and synced to a Firebase backend, enabling fleet managers and safety authorities to access aggregated data via a React-based web dashboard. The dashboard visualizes driver behavior trends, generates alerts for repeated offenses, and offers insights for safety enforcement and driver performance assessment. The project leverages modern mobile and web technologies to deliver an affordable, scalable, and real-time solution to promote road safety and responsible driving behavior in resource-constrained environments.