Facial Expression Recognition based on Graph Convolutional Networks with Phase Congruency

Phase congruency extracts features based on phase information. This paper proposes a facial expression recognition method based on phase congruency, which combines the features obtained from the graph convolutional neural network with the phase congruency to achieve high-precision recognition of facial expression images. Experimental result show that the method proposed performs better than the traditional graph convolutional neural network on both CK+ and JAFFE datasets.