We are a team of students of the Higher School of Economics, specializing in machine learning and application.
Increasing productivity of the online and offline learning is an essential task in the nowadays world. There is a tendency that online education is taking more and more over offline and the need in professors labor is declining while the efficiency in delivery of content is growing. However, multiple research shows that the online education still has problems such as high attrition rates, lack of interaction and student isolation. Given that, monitoring of the course quality, meaningful interaction with students and streamlining the content is highly important. We propose a novel video-based approach to analyse public presentations. Implemented by means of state-of-the-art computer vision algorithms and neural networks, Boremeter is an auditory tracking application which determines interest and involvement of an audience during an event, extracts gender and age characteristics of listeners. Boremeter has multiple real-life applications. It can help to detect the most resonant and most boring parts of a presentation, automatically extract the most involved and interested people from an audience, give a summary on socio-demographic characteristics of an auditory. This application can be used in developing online education projects as well as in assessment of public lectures in universities and IT-companies.