In the past, Student Engagements were measured in the form of statistical scales. In previous studies, some scholars divided the bad behaviors of students into 19 categories, covering 22 subcategories. These bad behaviors may represent a lack of either Student Engagements or intention to study the course. With the rise of artificial intelligence, some students’ lousy behavior recognition in the classroom can be used as the judgment standard of Student Engagements. In this work, we try to use image processing technology combined with machine learning and use SVM method to determine whether students have the use of mobile phones in the classroom. We divide the processing stage into several parts, namely pre-processing, segmentation, extract features, and machine learning. In the futures, we may use artificial intelligence to judge the dis-behavior of students during class; it is also possible to assist in the validation of research related to such scales in the past.