AI Powered Based Blind and Visually Impaired System for Assistance Process

Abstract

Autonomy in daily mobility and social interaction remains a significant challenge for the visually impaired. Conventional tools like white canes lack high-level environmental context, necessitating advanced technological intervention. This paper presents Smart Vision, an integrated AI-driven wearable system designed to bridge this gap. By leveraging state-of-the-art Deep Learning, the system provides real-time obstacle avoidance, familiar face recognition, and currency identification. We utilize the YOLOv8 framework for rapid object detection and the Grassmann model for robust facial analysis. The system processes visual data locally to provide low-latency audio feedback, empowering users to navigate dynamic environments independently with increased safety.

Introduction

This research focuses on improving independence and safety for visually impaired individuals by addressing the limitations of traditional assistive tools like white canes and guide dogs, which are ineffective in complex and dynamic environments. It proposes an AI-powered wearable assistive system that enhances real-time environmental awareness and reduces reliance on external help.

The proposed SmartVision framework integrates multiple deep learning techniques. It uses YOLO (You Only Look Once) for real-time obstacle detection, a Grassmann-based facial recognition model for identifying known individuals and providing personalized audio greetings, and a CNN-based currency recognition system to assist with financial transactions. These components enable object detection, face recognition, and currency identification in diverse real-world conditions.

The system is implemented using a wearable camera (head-mounted or chest-mounted) connected to a mobile processing unit. Video frames are processed through AI models, and results are converted into speech using a text-to-speech engine, delivered via bone-conduction or Bluetooth earphones so users can still hear ambient sounds.

Experimental results show strong performance, with over 92% accuracy in object detection, reliable face recognition under varying conditions, and accurate currency classification even with worn or partially visible notes. The system maintains low latency (under 500 ms), making it suitable for real-time navigation and assistance.

Conclusion

The SmartVision system effectively addresses the critical need for a unified, high-performance assistive technology for the visually impaired. By integrating real-time YOLOv8-based object detection, Grassmannian face recognition, and CNN-driven currency identification into a single wearable framework, this research provides a comprehensive solution for independent mobility and social interaction. Experimental results demonstrate that the system maintains high accuracy and low latency, delivering vital environmental context through intuitive audio feedback. Unlike traditional aids, SmartVision empowers users with spatial awareness and financial independence, significantly enhancing their quality of life. Future enhancements will focus on integrating GPS-based indoor navigation and cloud-based collaborative data sharing to further extend the system’s capabilities.

References

[1] Rocha, Daniel, et al. \"Using object detection technology to identify defects in clothing for blind people.\" Sensors 23.9 (2023): 4381. [2] Masud, Usman, et al. \"Smart assistive system for visually impaired people obstruction avoidance through object detection and classification.\" IEEE access 10 (2022): 13428-13441. [3] Shen, Junjie, Yiwen Chen, and Hideyuki Sawada. \"A wearable assistive device for blind pedestrians using real-time object detection and tactile presentation.\" Sensors 22.12 (2022): 4537. [4] Srikanteswara, Ramya, et al. \"Object detection and voice guidance for the visually impaired using a smart app.\" Recent Advances in Artificial Intelligence and Data Engineering: Select Proceedings of AIDE 2020. Singapore: Springer Singapore, 2021. 133-144. [5] Atitallah, Ahmed Ben, et al. \"An effective obstacle detection system using deep learning advantages to aid blind and visually impaired navigation.\" Ain Shams Engineering Journal 15.2 (2024): 102387.

Copyright

Copyright © 2026 Sumithra S., Vaishnavi S., Vidhya Sree R., Vishnu Priya K. J., Dr. J. Vaijayanthimala. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.