An Emotion Recognition System Based on Edge Computing and Cloud Computing Applied in the Kindergarten

Ruizhi Zhang,Chujie Tian,Yang Ji

DOI: https://doi.org/10.1145/3321408.3321596

2019-01-01

Abstract:As much attention is paid to the safety and education of children in the kindergarten, the importance of analyzing emotions of children is realized. With the adventure of the Internet of Things, more and more visual sensors are connected to the network, which makes the emotion analysis available through the recognition of images. It is necessary to response quickly when negative emotions of children are recognized, which means the processing of images containing negative emotions is delay-sensitive. The processing of an enormous amount of images brings challenge for emotion recognition based on traditional cloud computing. Cloud computing is of high scalability and sufficient capability, but cannot meet the need of real-time detection due to the time delay of transmission and computing of large amounts of images. Emerging edge computing can help to solve the problem by offloading computing from cloud center to edge nodes. However, because the resource of edge nodes is limited, only part of the computing can be placed at the edge nodes. In our work, an emotion recognition system combined with edge computing and cloud computing is proposed to apply in the kindergarten. Due to delay-sensitive negative emotions in this case, a priority scheduling algorithm is also proposed to improve the computing effectiveness of the edge. In this way, the low latency of the edge and the high scalability of the cloud are utilized to reduce the time delay and improve the capacity of the emotion recognition system. The experiments in the real scene of the kindergarten verify that the system performs well.