Wearable Assistive Device for Enhanced Navigation in Visually Impaired Individuals Using YOLO-Based Object Detection

The proposed study addresses the pressing challenge of enabling visually impaired individuals to navigate their surroundings with enhanced safety and autonomy. To this end, we propose a wearable assistive device designed to improve spatial awareness and facilitate environmental interaction. The primary objectives include developing an accurate object detection system, integrating real-time distance measurement capabilities, and incorporating situational awareness through weather data integration. The study methodology employs cutting-edge computer vision and machine learning techniques, with a YOLO-based object detection model at its core. This model ensures robust identification of objects, achieving an impressive 97% precision and recall. To support safe navigation, real-time distance measurement functionality was implemented. Additionally, the device incorporates real-time weather data via the Open Weather API, delivering comprehensive updates on temperature, precipitation, and wind conditions to assist in user decision-making. Design considerations focused on user-centric aspects such as battery life, portability, and cost-effectiveness to ensure practicality and accessibility. Evaluation results demonstrated an average mAP of 98.34% at IoU 0.5 across all object classes, with some reaching up to 99.50%. User testing validated the system's efficacy, achieving a 95% success rate in object identification and spatial guidance. These findings underscore the device's potential to enhance independence, safety, and quality of life for visually impaired users, promoting inclusivity through advanced assistive technology.