VISIONASSIST: Traffic light status detection for the visually impaired

Tan, Pei Shi (2025) VISIONASSIST: Traffic light status detection for the visually impaired. Final Year Project, UTAR.

PDF Download (12Mb)

Abstract

With the advancement of public transport in Malaysia, walking has become a vital component for individuals to navigate to public transport stations, work, school or any desired destination as an alternative to avoid traffic congestion. These walking individuals are known as pedestrians, where pedestrian traffic lights have been an important commander in guiding them to cross the road intersection safely by indicating the colour of red or green. However, these pedestrian lights have been found to provide minimal functionality in terms of accessibility for individuals with visual impairments, who have difficulty in interpreting traffic signal indications. Motivated by the need to enhance intersection safety and mobility for this population, this project aims to develop a mobile application that serves as a real-time pedestrian traffic light detector —effectively acting as the “eyes” for visually impaired users. The application introduces a novel solution that integrates traffic light recognition with accessibility-focused output mechanisms, such as auditory alerts and customizable haptic feedback. Using a YOLOv8-based object detection model and the smartphone’s rear camera, the application identifies and classifies pedestrian traffic light signals (red or green), immediately providing users with intuitive audio and haptic cues based on their personalised settings. In order to have a clear picture of the functionalities and limitations of existing assistive technologies in terms of software applications, existing applications such as Seeing AI, Be My Eyes and Lazarillo have been reviewed and presented in the report. Finally, the expected outcome of this project is a fully functional and accessible mobile application that empowers visually impaired individual to cross intersections safely and independently, thereby improving their mobility and confidence in navigating public environment.

Actions (login required)