Hazardous Object Detection for Visually Impaired People Using Edge Device

• Published in: SN computer science 2025-01, Vol.6 (1), p.7
• Main Authors: Kadam, Ujjwal, Kushwaha, Riti, Meena, Arvind, Abuzar, Chaudhary, Ujjwal, Singal, Gaurav, Verma, Madhushi
• Format: Article
• Language: English
• Subjects: Accident prevention; Accuracy; Algorithms; Blindness; Cameras; Computer Imaging; Computer Science; Computer Systems Organization and Communication Networks; Data Structures and Information Theory; Datasets; Deep learning; Hazardous areas; Information Systems and Communication Service; Literature reviews; Machine learning; Original Research; Pattern Recognition and Graphics; People with disabilities; Real time; Sensors; Soft Computing in Engineering Applications; Software Engineering/Programming and Operating Systems; Vision; Visual impairment
• ISSN: 2662-995X; 2661-8907
• DOI: 10.1007/s42979-024-03498-6

The motivation for this research stems from the need to improve the safety and independence of visually impaired individuals in their daily lives. These individuals face significant challenges in navigating their environments, particularly when it comes to identifying and avoiding hazardous objects that can cause physical harm. Existing assistive technologies for visually impaired individuals have focused primarily on mobility aids, such as canes and guide dogs, to help individuals navigate their environments safely. While these aids can be helpful, they are not foolproof, and visually impaired individuals face significant risks when encountering hazardous objects. Additionally, there is a no assistive technologies that specifically address the issue of hazardous object detection for visually impaired individuals. To address this gap, we propose a real-time edge-based hazardous object detection system that leverages light-weight deep learning model to classify objects captured by a camera mounted on a Raspberry Pi edge device. By identifying and alerting visually impaired individuals to the presence of hazardous objects in their environments, our system has the potential to significantly improve their safety and independence. Additionally, our research contributes to the growing body of literature on deep learning-based object detection systems, which have the potential to revolutionize many fields beyond assistive technology.