Resumen de Involvement measurement through head pose estimation

Cristina Mesquita, Rui Pedro Lopes

• Several decades of research clearly demonstrate that high-quality, developmentally appropriate, early childhood programs produce short and long-term positive effects on children’s cognitive and social de- velopment. The maintenance of high quality education processes and methodologies demand a rigor- ous process for improvement, guided through observation, reflection and consequent action. Typically, teachers will pay attention to the evolution of learning within the child, namely, on how the child is feeling and acting when dealing with a specific activity.

  This information, obtained through a detailed and thorough observation procedure, is based on mea- suring the child’s involvement. In fact, research with the Leuven Involvement Scale has shown that the levels of children’s involvement are a predictor of learning quality. An involved child gets extremely fas- cinated and absorbed by the activity he is performing. There are several indicators that can be used to assess children involvement, such as concentration, energy, complexity and creativity, facial expression and posture, persistence, precision, reaction time, language and satisfaction.

  Traditionally, the teacher will record the observation details in a paper based form. At the same time, he is controlling time and paying attention to the child’s posture when facing a specific activity. This task is complex and requires several observations, on many children, distributed through the day. Usually, this is performed with a video recorder to help the teacher to evaluate the observation.

  The number and diversity of observation tasks makes this a difficult process, with the possibility of many details being lost or disregarded. Moreover, the teacher’s training and background can also influence the subjectivity of the observation, further hindering the exchange of knowledge between teachers and institutions.

  Involvement is directly associated to the gaze direction (focus of attention), since humans face the activity they are performing. The more involved the child is, the more focus and concentration reveal, with less distractions and variations of the head posture. This fact opens the possibility to automatically associate head posture estimation to the measurement of engagement, providing a quantification of the child’s involvement.

  The work described in this paper focus on how modern image processing technology can provide a valu- able aid to kindergarten teachers, helping them in the task of registering observations. In this context, head posture is automatically detected and measured during a specific time period. Although easy for a human to interpret the orientation and movement of the human head, it is a challenge to computers.

  Although the different observation tools and approaches are useful for correlating data, the quantification resulting from this process will contribute to help teachers reduce subjectivity and to make informed and critical judgments about the quality of teaching and learning they offer