Facial expression recognition based on static and dynamic approaches

DANIEL ACEVEDO; PABLO NEGRI; MARÍA ELENA BUEMI; MARTA MEJAIL

Cancún

Conferencia; 23rd International Conference on Pattern Recognition (ICPR); 2016

The identification of facial expressions with humanemotions plays a key role in non-verbal human communicationand has applications in several areas. In this work, we analyzetwo main approaches for expression recognition.One is a dynamic approach introducing a new simple descrip-tor based on the angles formed by the landmarks to capture thedynamic of the facial expression on a sequence. In this case therecognition is performed by a Conditional Random Field (CRF)classifier. An analysis of the most discriminative landmarks forthis approach is presented.The other is a static-based appearance method. In this ap-proach, a binary-based descriptor, denominated Oriented Fastand Rotated BRIEF (ORB), is used on a single frame of asequence of images to extract texture information, and classifiedwith a Support Vector Machine.We compare both methodologies, analyse their similaritiesand differences, and also propose simple combinations of bothapproaches to deal with their limitations.