Interactive participation, as an important part of e-commerce teaching to check and fill the gaps, still has a large optimization and improvement space in the current e-commerce teaching classroom. This paper takes students’ classroom behavior performance as the entry point, transforms and defines students’ classroom behavior data, and constructs students’ course behavior set. At the same time, in order to effectively improve the recognition accuracy of students’ behavioral actions, an adaptive aggregation method is used to migrate relevant texture features. Dynamic filters and attention enhancement modules are used to capture students’ behavioral action information with high accuracy and build a reference-based super-resolution model. Finally, the student classroom behavior recognition algorithm is utilized to detect student classroom behavior performance and provide effective data reference for teaching interaction optimization. The designed method promotes 33 students to actively participate in classroom interaction in assisting e-commerce classroom teaching, and shows high reliability in optimizing e-commerce teaching interaction.
