Advancements in AI-Based Object Detection for Visually Impaired Individuals

This study proposes an AI-driven assistive system aimed at enhancing autonomous navigation and situational awareness for visually impaired individuals. The system integrates real-time object detection using YOLOv5, monocular depth estimation via MiDaS, and heuristic-based pathfinding through the A* algorithm to ensure accurate spatial understanding and obstacle avoidance. Contextual environmental descriptions are generated using BLIP and GenAI, while auditory guidance is provided through the Google Text-to-Speech (gTTS) engine. The proposed solution demonstrates robust performance in both indoor and outdoor settings, offering a scalable, portable, and efficient tool that significantly improves independent mobility, safety, and quality of life for visually impaired users.