Automatic Facial Feature Detection for Facial Expression Recognition

Taner Danisman (1) , Ioan Marius Bilasco (1, 2, 3) , Nacim Ihaddadene (4, 1) , Chaabane Djeraba (1, 2, 5, 6)

1 LIFL - FOX MIIRE
2 LIFL - Laboratoire d'Informatique Fondamentale de Lille
3 Université de Lille, Sciences et Technologies
4 no affiliation
5 IRCICA - Institut de Recherche sur les Composants logiciels et matériels pour l'Information et la Communication Avancé - UAR 3380
6 CRIStAL - Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189

This paper presents a real-time automatic facial feature point detection method for facial expression recognition. The system is capable of detecting seven facial feature points (eyebrows, pupils, nose, and corners of mouth) in grayscale images extracted from a given video. Extracted feature points then used for facial expression recognition. Neutral, happiness and surprise emotions have been studied on the Bosphorus dataset and tested on FG-NET video dataset using OpenCV. We compared our results with previous studies on this dataset. Our experiments showed that proposed method has the advantage of locating facial feature points automatically and accurately in real-time.

Mots clés: Facial feature detection; emotion recognition; eye detection; mouth corner detection

Domaines: Traitement des images [eess.IV]

Taner Danisman, Ioan Marius Bilasco, Nacim Ihaddadene, Chaabane Djeraba. Automatic Facial Feature Detection for Facial Expression Recognition. Fifth International Conference on Computer Vision Theory and Applications (VISAPP) 2010, May 2010, Angers, France. pp.407-412. ⟨hal-00812308⟩