Audio-Visual Broadcast Transcription System in the Era of Covid-19

This paper deals with the methods and algorithms for face (mask) detection and recognition in the system for automatic audio-visual TV broadcast transcription. In the era of Covid-19, traditional methods for face detection and recognition were failing as large numbers of people wearing face masks began to appear in television video recordings. Therefore, we focused our work on finding a sufficiently robust detector and classifier for face (masked/unmasked) detection and recognition. As part of this work, a face mask detector and a face recognizer were trained.