RECOGNIZING THE EMOTIONS OF PARTICIPANTS IN VIDEO CONFERENCES IN MICROSOFT TEAMS

Abstract

An information technology for recognizing the participants emotions during video conferences in Microsoft Teams using a convolutional neural network and Azure cloud tools is proposed. The structure of information technology and the structure of the Azure platform components are developed. Analogues were analyzed and their advantages and disadvantages were identified. The generalized method of recognizing participants emotions was analyzed, its disadvantages were identified. An improved method of recognizing participants emotions during video conferences in Microsoft Teams was proposed, which made it possible to automate the user's work and obtain data about the emotional reactions of participants in video conferences to certain news, announcements, discussion topics, to assess the abilities of speakers, detection of emotionally negative moments of video conferences and their causes due to the use of the Graph API library and Azure cloud technology. The proposed improved method became the basis of the relevant information technology. Experiments were performed, each of them involving a different number of video conference participants, different scenarios, different combinations of camera on and off, and different emotions. Analysis of the results showed that the developed information technology requires a single launch to handle any number of video conferences, while analog tools require a separate launch for each video conference. The developed technology makes it possible to recognize 94% emotions of video conference participants, unlike similar tools, as it supports a large number of participants at the same time without degrading the quality of their images. The developed technology makes it possible to identify all video conference participants and their emotional state, which is impossible using modern software.