Real-Time Object Detection For The Visually Impaired Using Yolov8 And NLP On Iot Devices

Abstract

Object detection is an important development in real-time that integrates and Artificial Intelligence (AI), embedded systems, and Internet of Things (IoT). These innovations aim to address the challenges faced by visually impaired individuals in locating everyday objects independently. Current systems often suffer from limitations like manual tagging, lack of realtime feedback, and poor adaptability in dynamic environments. This paper introduces an IoTbased Voice-Driven Smart Finder that leverages Natural Language Processing (NLP), YOLOv8 object detection, and cloud-based speech recognition for efficient and autonomous object location. The primary objective is to create a voice-interactive, low-cost solution that enhances the independence of visually impaired users by allowing them to locate objects using simple verbal queries. The system’s novelty lies in its integration of real-time object detection, speech-based control, and optimized edge-device deployment without the need for predefined object tagging. The proposed model achieved a detection accuracy of 92% on household objects and a fast response time of approximately 1.8 seconds, highlighting its practical effectiveness. By utilizing affordable hardware like Raspberry Pi 4 and integrating cloud APIs for speech processing, this work contributes a scalable and inclusive solution to the assistive technology landscape. Keywords: YOLOv8, Object Detection, Visually Impaired, Speech Recognition