AI-Powered Image Dehazing and Vehicle Detection in Foggy Road Conditions

Abstract

Road safety is often affected by weather conditions such as smog, fog, and smoke, which reduce visibility and make driving risky. Poor visibility increases the chances of accidents, slows down traffic, and creates difficulties for emergency services. In foggy or smoky conditions, drivers struggle to detect vehicles, pedestrians, or obstacles in time. Traditional aids like fog lights and reflectors offer only limited support, leaving a gap for more effective, technology-driven solutions that can work in real time to improve road safety. This project proposes a system that performs real-time haze removal and object detection. The method combines image enhancement with artificial intelligence to provide clearer visuals and better detection accuracy. First, a dehazing algorithm is applied to remove the effect of smog or fog and restore hidden details in the video frames. Once the visibility improves, the YOLO (You Only Look Once) object detection model identifies vehicles, pedestrians, and other objects with greater accuracy. This two-step process ensures that the system works effectively even in low-visibility conditions, providing reliable results for road monitoring and safety.