A Mobile Lens: Voice-Assisted Smartphone Solutions for the Sightless to Assist Indoor Object Identification

Abstract

Every aspect of life is organized around sight. For visually impaired individuals, accidents often occur while walking due to collisions with people or walls. To navigate and perform daily tasks, visually impaired people typically rely on white cane sticks, assistive trained guide dogs, or volunteer individuals. However, guide dogs are expensive, making them unaffordable for many, especially since 90% of fully blind individuals live in low-income countries. Vision is crucial for participating in school, reading, walking, and working. Without it, people struggle with independent mobility and quality of life. While numerous applications are developed for the general public, there is a significant gap in mobile on-device intelligent assistance for visually challenged people. Our custom mobile deep learning model shows object classification accuracy of 99.63%. This study explores voice-assisted smartphone solutions as a cost-effective and efficient approach to enhance the independent mobility, navigation, and overall quality of life for visually impaired or blind individuals.