Automatic annotation of facial actions from a video record: The case of eyebrows raising and frowning
Fiche du document
13 juin 2018
- handle: 10670/1.68zslk
- hal: hal-01769684
info:eu-repo/semantics/OpenAccess
Mots-clés
Facial expression machine learning eyebrows movement corpus gold standard automatic annotationSujets proches
Natural science Sciences Science of science Learning, MachineCiter ce document
Stéphane Rauzy et al., « Automatic annotation of facial actions from a video record: The case of eyebrows raising and frowning », HAL-SHS : linguistique, ID : 10670/1.68zslk
Métriques
Partage / Export
Résumé
Recent advances in the field of computer vision and machine learning have given birth to a generation of softwares enable to detect and track a face along a video record and eventually to measure its internal facial movements. We investigate herein on a specific task, the detection of eyebrows raising and frowning facial actions, the potential of these softwares on spontaneous, in-the-wild video corpus. We propose a tool which allow to extract eyebrows raising and frowning from the output of two state-of-the-art facial behavior analysis softwares (OpenFace and IntraFace). The evaluation performed on our manually annotated in-the-wild video corpus suggests that the tool can be used with benefits for automatic annotation purpose.