Construction of an Open/closed Eye Extraction System for Live Capture

Face tracking is a necessary technique for facilitating nonverbal communication between avatars in the metaverse space. In this study, we investigated a dynamic measurement approach for real-time representation of open/closed eye. The conventional method of extracting open/closed eye based on the as...

Full description

Saved in:
Bibliographic Details
Published inInternational Journal of Affective Engineering Vol. 23; no. 3; pp. 259 - 267
Main Authors TODA, Masatoshi, TANAKA, Hisaya
Format Journal Article
LanguageEnglish
Published Japan Society of Kansei Engineering 2024
Subjects
Face tracking
MediaPipe
Open/closed eye extraction
Online AccessGet full text

Cover

More Information
Summary:Face tracking is a necessary technique for facilitating nonverbal communication between avatars in the metaverse space. In this study, we investigated a dynamic measurement approach for real-time representation of open/closed eye. The conventional method of extracting open/closed eye based on the aspect ratio of the eye has been the mainstream method, but it encounters issues with false detection when the neck is tilted or when the user wears glasses during live transmission. To address these limitations, we developed detection method based on the ratio of the distance between the outer and inner eyelid, complemented by a calibration technique. The detection accuracy was verified by comparing the results with camera images. The open/close eye extraction system used a web camera and Google’s MediaPipe. The results demonstrated the accuracy of open/closed eye detection was 78.68%, even when the head was in motion, and 72.96% even when the participant was wearing eyeglasses.
ISSN:2187-5413
2187-5413
DOI:10.5057/ijae.IJAE-D-23-00030