Facial expression recognition plays an important role in human-computer interaction. In this paper, we propose the Coarse-to-Fine Cascaded networks with Smooth Predicting (CFC-SP) to improve the performance of facial expression recognition. CFC-SP contains two core components, namely Coarse-to-Fine Cascaded networks (CFC) and Smooth Predicting (SP). For CFC, it first groups several similar emotions to form a rough category, and then employs a network to conduct a coarse but accurate classification. Later, Then, an additional network for these grouped emotions is further used to obtain fine-grained predictions. For SP, it improves the recognition capability of the model by capturing both universal and unique effective features. To be specific, the universal features denote the general characteristic of facial emotions and the unique features denote the specific characteristic of each facial expression. Experiments on Aff-Wild2 show the effectiveness of the proposed CFSP.
View on arXiv