Abstract

In urban environments, navigating through construction sites presents significant challenges for individuals with blindness or low vision due to safety hazards and unpredictable changes. Existing navigation apps often fall short in addressing these obstacles. Our previous work introduces a computer vision-based model designed to detect construction sites from a distance. This enhances safety for visually impaired individuals by providing early warnings. In this project, we aim to refine the detection model through technical validation and user trials. Ultimately, our innovative approach seeks to establish a bidirectional flow of information between pedestrians and city agencies, thereby contributing to safer and more accessible urban environments.