Published Paper Details:

Safe and independent navigation remains a major challenge for visually impaired individuals, particularly in crowded and unfamiliar environments. Traditional assistive tools such as white canes provide limited environmental awareness and are unable to detect obstacles beyond short distances. To overcome these limitations, this research proposes a phone camera-based deep learning navigation assistance system designed to support visually impaired people through real-time obstacle detection and intelligent audio guidance. The proposed system utilizes the built-in smartphone camera to continuously capture live environmental data. A YOLOv8-based deep learning model is employed to detect navigation-related objects such as pedestrians, vehicles, doors, stairs, and other obstacles in real time. The detected objects are analyzed according to their position and relative distance from the user. Based on this analysis, the system generates immediate voice instructions using text-to-speech technology, enabling users to navigate safely without relying on visual input. The smartphone-based implementation eliminates the need for expensive external hardware, making the system portable, affordable, and suitable for everyday use. Additional features such as GPS support, low-light enhancement, and emergency assistance further improve user safety and usability in different environments. Experimental evaluation demonstrates that the proposed system achieves reliable object detection accuracy with low response time and real-time processing performance on smartphone devices. The study highlights the effectiveness of combining deep learning, computer vision, and smartphone technology to improve mobility, independence, and environmental awareness for visually impaired individuals.