Exploring the Path to Improve the Quality of Student Management Education Based on Knowledge Graph and NB-loT Architecture

Abstract

To research the effects of the smart classroom environment on college students' motivation to learn and engagement in that learning, as well as the link between those two factors in the smart classroom environment, a multi-scene student posture detection method using meta-learning is proposed. Through a combination of offline and online learning, the method develops a posture detection metamodel and a reasonable adaptation optimizer to quickly domain modify the posture detection model for certain teaching scenarios. A small amount of labelled sample data from a single teaching scene is all that is required for the metamodel to quickly adapt to the data distribution of that scene with the help of the adaptation optimizer in the online learning phase. The method simulates the process of the pose detection model in various types of teaching through two-layer training to train the parameters of the pose detection metamodel adaptation optimizer. The experimental findings demonstrate that college students' independent learning level and learning engagement are significantly higher in the smart classroom environment than they are in the traditional classroom environment. Additionally, there is a strong positive correlation between their independent learning level and learning engagement in the smart classroom environment.