Software Application for Assisting Visually Impaired Persons at Public Transport Stops

Анотація

The paper addresses the urgent problem of social adaptation of visually impaired individuals in Ukraine, where the number of  people with visual disabilities has significantly increased due to the consequences of military actions. The research presents  the development of an intelligent software application designed to assist visually impaired persons in independent orientation  at public transport stops. The system integrates deep learning, computer vision, and text-to-speech synthesis to automatically  detect public transport vehicles (bus, trolleybus, tram) and recognize their route numbers in real time. The YOLOv8 model was  employed for object detection, Tesseract OCR for route number recognition, and pyttsx3 for offline speech synthesis. Data  preprocessing included dataset annotation, augmentation with simulated weather conditions, and the use of OpenCV-based  filters to enhance OCR accuracy. Testing under both real and simulated conditions (fog, rain, snow, blur, low lighting)  demonstrated consistently high detection accuracy (100%) and acceptable classification performance, though recognition  robustness decreased under low-visibility scenarios. The results confirm the practical value of the proposed approach, while  further improvements will focus on expanding realworld datasets, enhancing preprocessing methods, and integrating  stronger deep learning models. The system holds promise as a foundation for wearable assistive technologies aimed at  improving inclusivity and mobility for visually impaired users in urban environments.