A computer vision-based heuristic framework for estimating traffic sign visibility

This study investigates the visibility of traffic signs from a driver’s perspective using computer vision, with a specific focus on addressing overestimations in visibility distances that arise when using a camera-centred approach. With the proliferation of automated systems in road safety, accurate visibility assessments are crucial, particularly for urban environments where traffic signs play a pivotal role in guiding driver behaviour and ensuring road safety. Leveraging the YOLOv5 (You Only Look Once) algorithm, trained on custom dataset, for traffic sign detection, this research provides a novel approach that combines computer vision with corrections to account for the driver's viewpoint, thus offering a more realistic estimation of visibility distances. The study compares visibility distances of informatory, mandatory, and warning signs with the test vehicle positioned adjacent to the shoulder (ATS) and adjacent to the median (ATM) across varying vehicle speeds. Findings reveal that ATM signs consistently exhibit higher visibility distances compared to ATS signs, highlighting the impact of sign placement on driver perception, particularly at higher speeds. This study's methodology and findings underscore the need for perspective-corrected visibility assessments in designing and positioning traffic signs to optimise safety. Future research directions include expanding the framework to assess visibility in diverse conditions such as low-light or inclement weather, incorporating additional occlusion sources, and exploring enhanced depth estimation techniques. The outcomes of this study contribute valuable insights for policymakers and road safety engineers, reinforcing the importance of human-centred adjustments in the development of effective, technology-enabled road safety strategies.