Emotion Regulation and Performance Enhancement in College Athletes Based on Emotion Recognition and Deep Learning

Abstract

Different emotions on the field will affect athletes' attention, decision-making and behavior, and exploring the effects of emotion regulation strategies on different emotions will be helpful for athletes to quickly adjust their state on the field and play at their proper competitive level. This paper proposes a novel method for athletes' emotion regulation and performance enhancement based on emotion recognition and deep learning technology. Firstly, a strong attention and residual network model of emotion recognition network is proposed. The model can carry out the whole process of capturing the important features of expressions and form the strong attention function. Specifically, strong attention refers to the capture of expression important features by adding channel and spatial attention mechanism (CBAM) in front of ResNet residual module. The global effective channel attention (G-ECA) is then added to the residual module to enhance the extraction of key features. Finally, CBAM is again embedded into the residual module to play the role of auxiliary extraction to minimize the loss of useful facial information. Secondly, the proposed model is simulated on two public benchmark expression datasets, CK+ and JAFFE, and the experimental results prove the effectiveness of the proposed method. Finally, based on the emotion recognition results, a standardized method for emotion regulation and performance enhancement in college athletes is given