Online education has been expected to be the future of learning; it will never replace the value of traditional classroom experiences fully. Technical problems have less impact on offline education, which gives students more freedom to plan their time and stick to it. In addition, teachers cannot observe their students' behavior and activities during offline education, and they can intervene when necessary. The offline education helps to know the way of behavior analysis of students. Depending upon the analysis student's characteristics and classroom performance can be evaluated by the teachers. The classroom analysis of the students helps in framing the lesson plan easier. The student's activity freedom is also focused on the offline education. The student's behavior and the study characteristics are clearly noticed by offline education classes. The complete educational sector performance is centered on the behavior analysis of the students. As long as students need offline education, it would be a critical component of their overall growth. As educational resources have grown, it has become more crucial to examine and evaluate offline classroom teaching behavior to indicate overall institution performance. A deep learning-student attention recognition framework (DL-SARF) for offline classroom assessment is developed in this article. There are three approaches to professional classroom behavior analysis: the student's intense focus on their side face, head down, and eyes. As the experiments demonstrate, the proposed deep learning-student attention recognition framework can accurately assess student behavior in the classroom and make the administration and implementation of the lesson plan easier.