Wearable Device for Visually Impaired using Deep Learning

This research introduces a novel wearable technology specifically to assist blind and visually impaired people in navigating their surroundings safely and independently. The proposed device integrates Raspberry Pi 4, cutting-edge object detection algorithms, voice output, and sensor feedback systems to provide comprehensive ambient awareness. Real-time video feeds captured by a Raspberry Pi camera module enable robust item detection and categorization. The Raspberry Pi 4 coupled with efficient deep learning algorithms facilitates real-time inference on resourceconstrained platforms. Additionally, water and ultrasonic sensors have been seamlessly integrated to detect wet surfaces and obstacles, respectively. Upon detecting hazards, the proposed system delivers prompt haptic and audio feedback, allowing users to avoid obstacles and make informed decisions. Extensive field testing and user trials validate the efficacy and versatility of this wearable technology in diverse real-world scenarios. This innovation enables a significant advancement in assistive technology, offering enhanced freedom and safety to visually impaired individuals. Future research directions may focus on algorithm optimization, sensor integration, and user-centered design enhancements to augment user experience and functionality.