Neural network algorithms for analysing the dynamics of the emotional state and involvement of students based on video surveillance data
Project goal
is to investigate solutions to the task of automatically analysing students' engagement in an online class and their emotions based on video surveillance data, which is relevant for the e-learning field.
Project tasks:
To develop new algorithms to classify individual and group emotions of students based on short fragments of video images of faces;
To create a method for predicting student engagement based on analysing lesson video;
To generate an algorithm for visualising video lesson fragments with the most pronounced student emotions;
To train the developed neural network models and classifiers for the set of video images of students' faces, including the protection of personal data;
To create computationally efficient algorithms and neural network models acceptable for implementation on mobile devices.
Project advantages:
-
Identifying the most and least interesting parts of the lesson
-
Identifying the dynamics of student engagement during one session and the average engagement of all students in each session of the course.
-
Assessing the quality of the trainees' perception of the teacher
-
Identification of pedagogical techniques that arouse the greatest interest of students
