YOLOInsight: Artificial Intelligence-Powered Assistive Device for Visually Impaired Using Internet of Things and Real-Time Object Detection
Gajanan Arsalwad, Saurabh Dabhade, Kabir Shaikh, Sean D’Silva
Abstract
This project presents an innovative artificial intelligence (AI)-powered real-time assistance system specifically designed for visually impaired individuals, aimed at enriching their real-world experiences and fostering inclusivity in various environments. By leveraging state-of-the-art AI algorithms, the system offers personalized assistance, navigation support, and seamless interaction for users with visual challenges. Through intuitive, user-friendly interfaces and adaptive technologies, the solution empowers individuals with visual impairments to navigate spaces independently and with greater confidence. At its core, the system integrates a camera module and YOLO-based deep learning algorithms running on a cost-effective Raspberry Pi 4, enabling real-time object detection and classification. Processed information is converted into accessible audio output, significantly reducing the cost compared to existing solutions without compromising functionality. To further enhance the user experience, the system incorporates language customization via optical character recognition and Google Text-to-speech technology, allowing audio feedback in multiple languages based on user preferences. Additionally, the project includes a built-in natural language processing application programming interface, akin to Siri or Google Assistant, but without relying on third-party services. This approach ensures complete control over the system’s features, enhances user privacy, and further reduces costs. By prioritizing the specific needs of visually impaired individuals, this innovative system aims to improve accessibility, foster independence, and offer a more inclusive experience in environments like shopping centers and public spaces. Overall, the system achieved 96% object detection accuracy, 91-95% intent recognition accuracy, and an average response latency of 650 ms, demonstrating its feasibility as a low-cost assistive technology for visually impaired users.