Quantifying Conversational Facial Gestures

EasyChair Preprint 10644

Abstract

Characterisation of facial gestures has so far relied on qualitative methods with manual coding of facial action units. Depth cameras provide an opportunity to automatically capture the structure and dynamics of conversational facial expressions across large datasets. Combined with cluster analysis, this methodology can enable a systematic characterisation of facial gestures in large corpora, facilitating more accurate, frame-by-frame multimodal analysis of gesture-speech combinations.

Keyphrases: Pragmatic facial gestures, computer vision, corpus data, depth camera, dialogue, facial action units

@booklet{EasyChair:10644,
  author    = {Ella Cullen and Patrick Healey and Paraskevi Argyriou},
  title     = {Quantifying Conversational Facial Gestures},
  howpublished = {EasyChair Preprint 10644},
  year      = {EasyChair, 2023}}