This paper constructs an interdisciplinary, innovative teaching model for environmental design majors based on information fusion technology. Based on sensors to collect students’ expression characteristics, interest characteristics, attention characteristics and other information, and three-dimensional modeling of students’ state, the trajectory multimodal data generated from students’ interactions with the teaching facilities in the process of learning are correlated with the externally manifested learning activities and the implicit learning characteristics. Subsequently, the extracted data was merged and analyzed, and feedback was provided to enhance teaching effectiveness. The results show that the information fusion model has a recall rate of more than 0.9, a checking accuracy rate of 0.87-0.93, and a student behavior recognition rate of 85%-91%, which are all better than other models. After using the innovative teaching method, the coefficient of student feedback evaluation increased from 0.4 to 0.88, which indicates the effectiveness of this paper’s model in student behavior identification and feedback.