Gesture Recognition Using Machine Learning for Light Communication Systems

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations

Abstract

Gesture recognition has a wide range of human-computer interface (HCI) applications in the home, commerce or office. However, the most widely used methods for recognizing gestures are computationally expensive and costly. We propose to apply gesture recognition to an existing visible light communication (VLC) system. Different finger motions are detected using a long short-term memory (LSTM) network operating on light transitions between fingers. At the receiver side, the platform utilizes a light-emitting diode and photodiode. The device can distinguish motions from gaps in direct light transmission, making it compatible with high-speed light communication systems. The accuracy of gesture identification was evaluated for five different gestures over a distance of 48 cm and the findings show the method is capable of successfully identifying the motions with 88 percent accuracy.

Original languageEnglish
Title of host publication2022 International Mobile and Embedded Technology Conference, MECON 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages52-56
Number of pages5
ISBN (Electronic)9781665420204
DOIs
StatePublished - 2022
Event2022 International Mobile and Embedded Technology Conference, MECON 2022 - Noida, India
Duration: 10 Mar 202211 Mar 2022

Publication series

Name2022 International Mobile and Embedded Technology Conference, MECON 2022

Conference

Conference2022 International Mobile and Embedded Technology Conference, MECON 2022
Country/TerritoryIndia
CityNoida
Period10/03/2211/03/22

Keywords

  • gesture recognition
  • human activity recognition
  • LSTM
  • machine learning
  • VLC

Funding Agency

  • Kuwait Foundation for the Advancement of Sciences

Fingerprint

Dive into the research topics of 'Gesture Recognition Using Machine Learning for Light Communication Systems'. Together they form a unique fingerprint.

Cite this