Learning inter-class optical flow difference using generative adversarial networks for facial expression recognition

Abstract Facial expression recognition is a fine-grained task because different emotions have subtle facial movements. This paper proposes to learn inter-class optical flow difference using generative adversarial networks (GANs) for recognition. Initially, the proposed method employs GAN produce images from between static fully expressive samples and neutral samples. Such used highlight displacement of parts images, which can avoid disadvantage that change adjacent frames same video image not obvious. Then, designs four-channel convolutional neural (CNNs) high-level features produced appearance respectively. Finally, decision-level fusion strategy adopted implement classification. The validated on two public databases, BAUM_1a, SAMM AFEW5.0, demonstrating its promising performance.