DEVELOPMENT AND PERFORMANCE ANALYSIS OF FACE AND EMOTION RECOGNITION SYSTEMS IN VIDEO CONFERENCE BASED ON WEB APPLICATION

• Main Author: Hakim, Umar
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/82230
• Institution: Institut Teknologi Bandung

Emotions affect learning outcomes such as grades, motivation, satisfaction, interest and understanding. However, it is difficult for teachers to assess learners' emotions in synchronous online learning due to lack of interaction and limited screen space. This research aims to develop and test the performance of a web application-based dashboard system that serves as a reporting medium for face and emotion recognition results. The proposed system utilizes Next.js web technology, FastAPI, MongoDB, and Google Cloud Platform. In addition, Convolutional Neural Networks (CNNs) models are used to perform real-time face and emotion recognition during synchronous online learning. By capturing the video conference window and analyzing the faces of participants through screenshots, the system displays the recognition results on a dashboard. The research methodology used is the scientific method which includes literature study, development, experimentation, and performance testing of the dashboard system. The results showed that the system was successfully developed with good front-end performance. However, optimization is needed on the back-end to improve the performance, reliability, availability, and resilience of the service so that it can be used by many users at once. In addition, to ensure a more comprehensive and targeted testing approach, a possible follow-up is to test user experience and/or usability of the system.