AI-Based Smart Glasses for Visually Impaired Using Real-Time Object Detection and Voice Assistance

This paper presented the design and implementation of an AI-based smart glasses system for visually impaired individuals, integrating real-time object detection with voice-based feedback in a wearable form factor. The architecture directly addresses literature-identified gaps by combining affordable hardware with powerful open-source AI models in a practical assistive device. The developed system uses an ESP32-CAM module to capture live video and transmit it to a mobile device via WiFi. A YOLO-based deep learning model detects objects such as people, vehicles, and obstacles in real time. Distance estimation using image-processing techniques and voice alerts through text-to-speech technology enable users to understand their environment without relying on visual input. Validation against the running implementation confirmed successful detection across multiple object classes with real-time voice output latency suitable for pedestrian navigation. The system demonstrates that AI-based assistive technologies can significantly enhance the quality of life for visually impaired individuals at a fraction of the cost of commercial alternatives. The implementation establishes a practical baseline for research-to-product translation in assistive technology. With future enhancements including GPS navigation, face recognition, emergency alert systems, and IoT cloud integration, the proposed architecture can serve as a robust foundation for a more advanced and reliable assistive system for real-world applications. Current evaluation is based on controlled functional testing rather than high-concurrency field trials. Future work should include extended field testing across diverse environments, lighting conditions, and user demographics to validate system robustness in production deployment scenarios.

[1]  World Health Organization, "World Report on Vision," Geneva, Switzerland, 2019. [2]  M. A. Williams, A. Hurst, and S. K. Kane, "Pray before you step out: Describing personal and situational blind navigation behaviors," in Proc. 15th Int. ACM SIGACCESS Conf. Computers and Accessibility, 2013. [3]  J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You Only Look Once: Unified, Real-Time Object Detection," in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2016, pp. 779-788. [4]  A. Gupta, A. Verma, and A. Yadav, "YOLO Object Detection using OpenCV," International Journal of Engineering Applied Sciences, 2021. [5]  R. David et al., "TensorFlow Lite Micro: Embedded Machine Learning on TinyML Systems," arXiv preprint arXiv:2010.08678, 2020. [6] H. Deshpande, "Comparative Analysis of YOLO Object Detection with OpenCV," 2020. [7]    A. Rosebrock, "YOLO Object Detection with OpenCV," PyImageSearch, 2018. [8] "Object Detection in Computer Vision," Wikipedia, 2025. [9] "You Only Look Once (YOLO)," Wikipedia, 2025. [10]    J. Redmon and A. Farhadi, "YOLO9000: Better, Faster, Stronger," in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2017, pp. 7263-7271.