Monitoring students' attention and engagement levels in classrooms are critical in promoting an interactive teaching process. Therefore, in addition to teaching, instructors are constantly making an effort to monitor and maintain the students' attention and engagement levels throughout the lecture session. Recent advancements in computer vision techniques have resulted in tools that empower real-time head-pose estimation for multiple people with reasonable accuracy. This study aims to harness such technology to help instructors determine and evaluate faster and more accurately the level of students' attention, thus supporting them in managing the session appropriately and more beneficial to the students. Our system employs an ordinary webcam, a standard computer, and simple computer vision algorithms. The system determines the students' faces, estimates their level of attention, and displays them through easy color-coded charts to the instructor to take the necessary action during the lecture. To avoid complicating the system, the metric given is a percentage of attention and associated color-coded level of concern so that the instructor can act. Our results show that the proposed system can numerically evaluate students' attention in the classroom, individually and in-group, in real-time. Our scheme was tested a classroom with 8 students, and the instructors' feedbacks were positive. From the results, for instance, a student shows 89% instant attention, and the overall classroom attention was 40% after 11 minutes of starting the lecture. Similar measurements can be obtained from each student and the overall students, at any time during the lecture, which is novel compared to other similar studies that only target either the student or the overall class in mostly offline mode or complicated and unpractical setup ((more detailed comparisons are in the introduction). We assume that this work has potential in significantly enhancing instructors' teaching abilities and students' academic performance.