Portable offline indoor object recognition system for the visually impaired

Mohammed Noman, Vladimir Stankovic, Ayman Tawfik

Research output: Contribution to journalArticlepeer-review

3 Downloads (Pure)

Abstract

This article presents an indoor assistive system that addresses the challenges faced by visually impaired individuals. The proposed system helps the visually impaired individuals to move indoor and make them independent of any external assistance. The proposed system consists of a camera with a processing unit and an accompanying Time-of-Flight sensor providing an efficient, convenient and cost-effective solution. The proposed system achieves average object detection accuracy of 73.34% and a 5% error margin in detecting the distance and length of detected objects. The performance comparison with two existing systems shows that the proposed system provides a very close performance to the benchmarks with advantages of portability easy-to-use and no requirement for cloud services.

Original languageEnglish
Article number1823158
Number of pages19
JournalCogent Engineering
Volume7
Issue number1
DOIs
Publication statusPublished - 23 Sep 2020

Keywords

  • visually impaired
  • indoor
  • object detection
  • time-of-flight
  • tensorflow

Fingerprint Dive into the research topics of 'Portable offline indoor object recognition system for the visually impaired'. Together they form a unique fingerprint.

Cite this