Mobile indoor navigation with object recognition for visually impaired

Gooi, Yong Shen (2025) Mobile indoor navigation with object recognition for visually impaired. Final Year Project, UTAR.

PDF Download (5Mb)

Abstract

Navigation is an important aspect in daily life, but visually impaired individuals might struggle to navigate by themselves safely and independently. Nowadays, the advancement of mobile solutions with artificial intelligence (AI) and computer vision (CV) technology has encouraged the development of many innovative solutions to solve problems. In this project, a standalone mobile application called Visiovigate is developed for indoor navigation assistance of the visually impaired communities. It is designed to support visually impaired individuals by integrating real-time object recognition, and indoor navigation using mobile sensors. Existing assistive technologies often lack comprehensive indoor navigation abilities or are reliant on expensive hardware. It might limit their accessibility and effectiveness. Thus, Visiovigate addresses these gaps by leveraging deep learning and computer vision techniques by using the You Only Look Once (YOLO) model for efficient object detection on mobile devices and utilizing the mobile pedestrian dead reckoning mobile sensors like magnetometer and accelerometer for indoor navigation without relying on global positioning system (GPS) and internet connection. The application will also offer real-time audio and haptic feedback for ensuring the visually impaired users receive immediate environmental awareness and directional guidance. Therefore, this mobile application is best to use with a traditional solution like cane that will further increase efficiency as this application can inform users about the incoming obstacles that are not reachable by the traditional solution. This system operates entirely on standard mobile sensors which have been commonly built into smartphones nowadays. It aims to run in a stable condition at any mobile device without internet connection. Hence, this project also will provide a more cost-effective and accessible solution that enhances the safety and independence of its users in indoor environments.

Actions (login required)