AI-Based Road Safety Audit Automated Detection and Deterioration Assessment of Highway Safety Elements

Ensuring road safety requires continuous inspection and maintenance of critical infrastructure such as lane markings, signboards, and barriers. Traditional manual inspections are time-consuming, expensive, and prone to inconsistencies, leading to delays in identifying deteriorated safety products and increasing accident risks. This study presents an AI-powered solution that automates road safety audits using computer vision[1]. An object detection model identifies road safety elements, and a segmentation model evaluates their deterioration levels by classifying defects such as rust, fading, or structural damage. The deterioration percentage determines the classification: Good (≤30%) – No immediate action required; Moderate (31–70%) – Requires maintenance within a reasonable timeframe; Bad (>70%) – Requires urgent replacement or repair. The implemented system achieves a minimum accuracy rate of 87.5% in detecting and classifying road safety elements, contributing to a 40% reduction in inspection costs and enabling proactive maintenance scheduling. By automating road safety audits, this system enhances detection accuracy, reduces manual inspection costs, and enables scalable, real-time monitoring of highways[11].