Children Activity Recognition (HAR) in an Educational Context.
The objective observation of a child’s behavior in the classroom can be done using motion sensors, which can generate information about the level of agitation, concentration and engagement. The Human Activity Recognition (HAR), together with wearable sensors with accelerometers, have helped in the objective capture and analysis of the movement and behavior of individuals. However, in the case of an applied context of the classroom and with
children, there are still some limitations: the lack of databases in the context, the difference in the expression of activities by adults and children, and between each child.Thus, it is necessary the exploration of HAR in the proposed context, also considering the student’s motor and affective dimensions.Through a case study in the field, a data collection was carried out with inertial wearable sensors of the activities of two classes of the second year of Elementary School, considering the moment of the class, level of attention and activities of each student.It is expected to create a public database for further studies on the subject, classifying students’ activities in class and understanding the school routine of study volunteers in the scenario of pandemic.