YOLOV8 IN REAL-TIME INDOOR OBJECT DETECTION AND AUDIO ASSISTANCE FOR VISUALLY IMPAIRED USERS

In recent years, advancements in Artificial Intelligence have created new opportunities for developing assistive technologies to improve the lives of individuals with disabilities. For the visually impaired, navigating indoor environments and identifying objects presents a significant daily challenge, often hindering their independence. The researchers believe that a crucial step in addressing this is through real-time, intelligent visual aids by developing and evaluating a real-time object detection system using the YOLOv8 model, designed to provide auditory feedback for visually impaired users. The study rigorously assesses the performance of the model under various real-world conditions, including different lighting levels, image sizes, and object densities. The researchers found that while the model achieved an overall accuracy of 60%, its performance is significantly impacted by these factors. Accuracy dropped sharply in dim lighting, an optimal image resolution of 640x640 was identified, and performance degraded substantially in cluttered, high-density scenes. The study concludes that there is a significant difference in the effectiveness of the model across these conditions, providing critical insights for the development and deployment of robust, reliable assistive technologies.