Visual based drowsiness detection using facial features

Quang Nguyen, Le T. Anh Tho, Toi Vo Van, Hui Yu, Nguyen Duc Thang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

157 Downloads (Pure)

Abstract

In this paper, a camera-based system is proposed to detect and monitor drowsiness of a car driver in real time. The system utilizes an RGB image to track the drivers’ face and their eyes to detect sleepy sign. For the face detection and segmentation, a robust method based on Haar features is applied. Within the segmented areas of faces, random forest is utilized to locate eye regions. Once the eyes are located, the local region of eyes is extracted to yield binary images of the eye silhouettes in which the open and close stages of the eyes are revealed. The portion of the close states of the eyes during a certain number of frames is calculated to track the drowsiness signs. If this portion exceeds a predefined threshold, the system concludes that the driver tends to falling asleep and generate alert to the users.
Original languageEnglish
Title of host publicationProceedings of the 6th International Conference on the Development of Biomedical Engineering in Viteman (BME6)
EditorsToi Vo Van, Thanh An Nguyen Le, Nguyen Duc Thang
PublisherSpringer Singapore
Pages723-727
Number of pages5
ISBN (Electronic)978-9811043611
ISBN (Print)978-9811043604
DOIs
Publication statusEarly online - 24 Sep 2017
EventBME International Conference on the Development of Biomedical Engineering in Vietnam - Ho Chi Minh, Vietnam
Duration: 27 Jun 201729 Jun 2017

Publication series

NameIFMBE Proceedings
PublisherSpringer
Volume63
ISSN (Print)1680-0737

Conference

ConferenceBME International Conference on the Development of Biomedical Engineering in Vietnam
Country/TerritoryVietnam
CityHo Chi Minh
Period27/06/1729/06/17

Keywords

  • human-machine interface
  • drowsiness monitoring
  • face detection
  • decision tree

Fingerprint

Dive into the research topics of 'Visual based drowsiness detection using facial features'. Together they form a unique fingerprint.

Cite this