Abstract:

In recent years many schools adopted tele-education during the COVID-19 pandemic. However, in several cases the use of cameras is not permitted during tele-education, and as a result student participation may be affected. The work described in this abstract aims to improve the active participation of students during tele-education by employing a privacy preserving human action recognition methodology, so that the educator is informed about student activity during on-line classes without having access to visual input showing the students. The aim of our work aims to evaluate the acceptance of using action recognition system for monitoring student activity in tele-education.

A prototype action recognition application developed as part of our previous work in this area, analyses images captured by a camera using a deep neural network in order to recognize student activity. Information about student activity is transmitted to the educator’s machine, and used to animate student avatars in a 3D virtual class environment. Therefore, the educator can visualize student activity without having access to actual images showing the students.

To investigate the acceptance by educators to use this technology, an evaluation experiment was staged to test the following research questions:
RQ 1: Is it feasible to use privacy protecting methods for monitoring student actions during tele-education?
RQ 2: Is the use of privacy protecting methods for monitoring student actions during tele-education acceptable by educators?
RQ 3: Could monitoring of students' actions improve the efficiency of tele-education ?

The participants taking part in the evaluation were primary, secondary, and higher education educators. All participants completed a first part of an online pre-questionnaire with questions related to their impressions about tele-education. Then the participants were informed about the functionality of the pilot action recognition application by watching a demonstration video. Finally, a post questionnaire, similar to the pre-questionnaire was completed.

According to the preliminary results, the majority of participants consider that it is feasible to use the proposed application for monitoring student actions, while they believe that it does not affect student privacy to a considerable extend. The educators believe that the teacher should be informed about the students’ actions during online courses. Furthermore, educators agree that the application could improve the presence of the students, active participation, teacher-student interaction, evaluation, concentration level, learning outcomes, class control and students’ interest. Statistically significant differences were on observed in several questions before and after the intervention, indicating the positive impact and acceptance of the pilot action recognition application.

Based on the preliminary results, teachers find it feasible to use the proposed action recognition application, while they believe that it could improve the performance of students during distance learning. In the future we plan to cross verify the results obtained by allowing participants to test the application in real scenarios and evaluate their views using interviews. We also plan to evaluate the acceptance of impressions of students and parents and compare results between students, parents, and educators in order to produce more concrete conclusions.

Keywords:

Tele-education, active participation, action recognition.