Automatic camerawork system for lecture archives
Haruna Tsutsumi, Naoshi Kaneko, Katsuto Nakajima · 2024
We propose an automatic camerawork system to reduce the cost of editing lecture videos for online education. The system detects the lecturer's movement from a fixed camera image by recognizing the direction of their face and positions of their hands, and changes the scale and perspective angle accordingly. The effectiveness of the system was assessed through subjective questionnaires that asked respondents to rate its performance in terms of different technical parameters. The results highlighted that the proposed system could effectively create useful videos for teachers. Overall, the proposed system can greatly reduce the time spent by teachers in preparing for their lessons and assist them in more efficiently utilizing their time in the classroom.