The motivation behind this research was profoundly simple yet incredibly challenging: how can we leverage cutting-edge artificial intelligence to grant true spatial independence to the visually impaired? My focus was on creating a system that didn't just recognize objects, but understood the environment dynamically in real-time.

To achieve this, I engineered an architecture that fused YOLO-World for open-vocabulary object detection with Monocular Depth Estimation. This combination allowed the system to identify an infinite vocabulary of objects on the fly while simultaneously calculating their precise distance from the user using only a single camera stream.

One of the most significant hurdles was optimizing this dual-model pipeline for real-time inference. Assistive technology must be instantaneous to be safe. Through rigorous architectural tuning, the system processes environmental data rapidly, ensuring that users receive immediate, actionable audio feedback about their surroundings.

The culmination of this extensive development and testing phase was the successful publication of the research in the International Journal of Engineering Research & Technology (IJERT). Seeing the methodology, experimental results, and practical implications peer-reviewed and published was an incredible milestone.

Ultimately, this project represents more than just an academic achievement; it is a testament to the transformative power of AI in accessibility. It reinforces my dedication to building advanced, real-world solutions that directly improve human lives.