Artificial Intelligence (AI) has revolutionized assistive technologies for the visually impaired by enabling real-time environmental perception through smart glasses. This paper presents an AI-powered smart glass system that integrates computer vision, deep learning (YOLOv5), ultrasonic obstacle detection, and text-to-speech (TTS) feedback. The proposed solution achieves 92.5% object recognition accuracy with 150ms latency, outperforming existing devices like OrCam MyEye and Microsoft Seeing AI.

The system consists of a wearable glass frame equipped with a high-resolution camera, ultrasonic sensors, and an AI processing unit. It provides real-time auditory feedback to users, describing objects, reading text, and warning of obstacles. Extensive testing with 50 visually impaired participants demonstrated 86% user satisfaction, with particular improvements in navigation safety and independence.

Key challenges such as power consumption, low-light performance, and cost are addressed, with future directions focusing on augmented reality (AR) integration and energy-efficient hardware optimization.