To-sequence:Multi-label Relation Modeling in Facial Action Units Detection
Facial Action Units Detection (FAUD), one of the main approaches for facial expression measurement, based on the Facial Action Coding System (FACS), makes the subtlety of human emotions available in the various applications, such as micro-expression recognition, expression generation. Therefore, FAUD has recently become a popular research field. Inspired by the recent advance in text multi-label classification task, we adapt the sequence-to-sequence method for multi-label text classification, which directly models the relationship between labels to treat the multiple activated AUs as a sequence in the context of data representation, thus transforming the multi-label classification task into a sequence modeling task. We implement the above algorithm on the data set released by the competition FG-2020 Competition:Affective Behavior Analysis in-the-wild (ABAW).
READ FULL TEXT