Authors: P.Sreesudha, CV.Kiranmaiee, P.Santhoshini, Nadia Shareen, Megha Chandana, Harini Vadla
Abstract: Real-time object detection and environmental awareness are essential components in assistive technologies for visually impaired individuals. Traditional mobility aids provide limited information about surrounding objects and their proximity, making independent navigation difficult in complex environments. In this work, an AI-based assistive vision system is proposed that integrates the YOLOv8 deep learning model for real-time object detection, distance estimation techniques for proximity awareness, and text-to-speech output for auditory feedback. The system captures input from a camera, detects and classifies multiple objects in the environment, estimates their distance from the user, and converts the detected object labels along with distance information into speech output. This enables visually impaired users to understand nearby obstacles and objects more effectively while moving in indoor and outdoor environments. The proposed approach offers a practical, low-cost, and efficient assistive solution by combining computer vision and artificial intelligence to enhance user safety, independence, and confidence. Experimental observations indicate that the system performs effectively for common object categories and provides meaningful audio guidance in real time
DOI: https://doi.org/10.5281/zenodo.19280231
