Deakin University
Browse

File(s) under embargo

Automatic facial expression recognition in an image sequence using conditional random field

conference contribution
posted on 2024-04-09, 06:27 authored by M Roshanzamir, A Mirzaei, MT Darbandy, A Shoeibi, Roohallah AlizadehsaniRoohallah Alizadehsani, Fahime KhozeimehFahime Khozeimeh, Abbas KhosraviAbbas Khosravi
Facial expression recognition is one of the fields that nowadays has attracted the attention of many researchers. It is possible to automate facial expression recognition using artificial intelligence methods. This will be of great help to researchers, especially in areas such as psychology. Automatic facial recognition can be derived from a static image of facial expression, but a better and more efficient way to do this is through a sequence of images. In this paper, a new method is proposed to automatically detect facial expressions from a sequence of images. Each sequence of facial images begins with a face neutral state and ends with one of the six main emotions. Motion vectors are extracted from the sequence using optical flow algorithm. These vectors are then used to train the conditional random field and finally to automatically determine the emotion. In this paper, in addition to the basic conditional random field, the hidden dynamic conditional random field is also investigated. Additionally, the effect of changing some parameters of these algorithms such as different optimization methods has been investigated. Given that a facial expression is recognized during a sequence of images, random field-based methods can be used for efficient classification of facial expressions reaching accuracy (more than 90%) competitive with the best existing methods for facial expression recognition.

History

Volume

00

Pagination

271-278

Location

Budapest, Hungary

Start date

2022-11-21

End date

2022-11-22

ISSN

2380-8586

eISSN

2471-9269

ISBN-13

9798350398823

Language

eng

Publication classification

E1 Full written paper - refereed

Title of proceedings

CINTI-MACRo 2022 : Proceedings of the IEEE Joint 22nd International Symposium on Computational Intelligence and Informatics and 8th International Conference on Recent Achievements in Mechatronics, Automation, Computer Science and Robotics 2022

Event

IEEE Computational Intelligence and Informatics & IEEE Mechatronics, Automation, Computer Science and Robotics. Combined Conference (2022 : Budapest, Hungary)

Publisher

IEEE

Place of publication

Piscataway, NJ.

Series

International Symposium on Computational Intelligence and Informatics