Visual speech recognition in sparse data domains

  • Philip McShane

Student thesis: Doctoral ThesisDoctor of Philosophy


This work is focused on the development of Visual Speech recognition systems within sparse data domains. The use of Visual Speech recognition as this basis for a Liveness Verification system is discussed. This thesis also explores the adaptation of models trained within one domain to be used with data from another (e.g. adaptation from RGB to IR).
Date of AwardDec 2022
Original languageEnglish
Awarding Institution
  • Queen's University Belfast
SupervisorDarryl Stewart (Supervisor) & Ming Ji (Supervisor)


  • Lipreading
  • visual speech recognition
  • speech recognition
  • liveness verification
  • biometrics

Cite this