Classroom learning behavior statistics can provide a data-driven mechanism for classroom evaluation and pedagogy improvement to improve teaching quality. Traditional classroom learning behavior identification methods are time-consuming, inefficient, and subjective, and deep learning-based automatic classroom learning behavior identification techniques are in the stage of continuous improvement. Existing classroom learning behavior recognition studies use a coding framework that lacks an effective mapping between behaviors and cognitive engagement in learning, which does not facilitate the provision of feedback and classroom evaluation for teachers. Therefore, this study refers to the ICAP framework to identify students' classroom learning behaviors based on YOLOv8n automatically and migrates the application to analyze statistical features, time-series plots, and scatter plots of classroom learning behaviors to characterize the classroom learning process visually. The results show that the Yolov8n-based automatic classroom learning behavior recognition model performs well and can be applied to the in-depth analysis of classroom learning behaviors; the migration of the model can provide more intuitive and directly usable visual feedback on the learning behaviors of individual students and student groups, which is conducive to the teacher's intervention and improvement of teaching and intelligent classroom evaluation.