Development of a Crooked Road Sign Recognition System

Le résumé fourni par la source

Various signs on roads serve to indicate traffic rules and regulations. The majority of these signs are composed of metal and are structurally sound; however, a significant number of signs throughout the country have been distorted due to the effects of aging, natural disasters, and traffic accidents. These signs have been observed to reduce visibility, which has led to concerns regarding traffic safety. However, the restoration of damaged signage is a time-consuming process that necessitates a thorough assessment of the impairment and the implementation of subsequent repairs. Consequently, some signs remain unattended for extended periods. To address these issues, a bent sign recognition system has been developed. The system captures two consecutive photographs from a dashcam video, calculates the distance to the sign based on changes in the sign's coordinates within the image, and then determines the sign's tilt angle. Furthermore, the system identifies roads merging into the travel lane and utilizes the obtained information to determine whether the sign is targeting the travel lane. This capability facilitates the detection of bent signs. The utilization of this system has the potential to enhance the efficiency of traffic infrastructure management and contribute to the prevention of traffic accidents.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.