Research And Application Of Classroom Personal Behavior Recognition Based On Bone Data

As the main platform to educate students,classroom plays a vital role.Classroom is the student accept knowledge,understanding knowledge,the important link of knowledge and mastery of knowledge and growth way,the behavior of students in the classroom to a large extent affects their willingness to accept knowledge,so the classroom behavior recognition is of great significance to teaching,the teacher can timely and fully understand students’ learning behavior in the classroom,both neither can cause hysteresis and one-sidedness of student behavior information,also can improve the students’ self-consciousness and concentration,it can not only improve the efficiency of teaching,can improve the quality of teaching.Compared with the traditional teaching evaluation,which costs both manpower and time,the evaluation of classroom students’ behavior by combining computer vision and artificial intelligence will be more objective and diversified.In order to solve the problem of limited amount of collected classroom behavior videos and low recognition rate of behavior recognition methods,a multi-kernel SVM behavior recognition method based on manual features was proposed.First,use the device sensor extract students in a classroom setting bone 3D coordinate data collection and RGB image,the fusion of the angle and distance between skeleton coordinates to describe the information of bones,defines the key points and the euclidean distance and the cosine value of the hip,key points of each frame with the first frame of the key points of euclidean distance and the cosine value of between the bones of quaternions are characterized,and then use PCA algorithm for high-dimensional data dimension reduction processing,finally,use the nonlinear classifier based on kernel function of SVM classification.In the experimental part of this paper,the classroom data set and the open test data set are used to verify and analyze the proposed method.The experimental results show that the proposed method has a better recognition rate than the comparison method on the small and medium-sized data set.On class data sets,they were 43.4%,46% and 48.2% more accurate than Joint angles,Joint positions and ST-GCN methods.According to the method proposed in this paper design and implement a class student behavior recognition based on skeleton data of prototype system,can realize to the human body skeleton data records,behavior identification and analysis,and other functions,through the sensor device as an input device to get the students classroom environment behavior of data information,and then collect data at the classroom behavior of information processing,identification and storage,into the classroom behavior recognition in the prototype system for identification and analysis,behavior identification conclusion,finally record and an interface to display.