ABSTRACT
The Covid-19 pandemic changed the course of activities, both work and education in the world, migrating to the requirement of virtual platforms and videoconferencing tools, such as Zoom, Google Meet, Jitsi Meet, among others. This generated a globalized and digital culture of learning, activities in congresses, and even business meetings using videoconferences. This new scenario creates uncertainty, especially in educators, due to the level of attention they are receiving from students through virtual classes and other scenarios where they want to evaluate the emotions created in the people who receive them information virtual written description intended to provide factual informationally. For this reason, to support different video conferencing platforms or other audiovisual media, a tool is presented that captures video in real-time. It automatically recognizes the emotions expressed by people using deep learning tools, happiness, sadness, surprise, anger, fear, disgust, and neutral emotions. The initial training and validation system is based on the CK+ Dataset that contains images distributed by emotions. This tool was developed for the WEB in Python Flask, which in addition to automatic recognition in real-time, generates statistics of the emotions of the people evaluated with 75% accuracy. To validate the tool, videoconferencing programs were used, the emotions of a group of students were evaluated, and open videos were available online on YouTube. With this study, it was possible to re-know the emotions of the people who attended the class, which allows the teacher to take measures if the students do not carry out the planned activities. © 2022, Springer Nature Switzerland AG.