Applying Super Resolution and Optical Flow to Implement a Roll Call System for a Small Classroom by Webcam
Ming‐Yang Su, Wang-Di Xu · 2023
Traditional facial recognition technology has been widely adopted in the technology industry and has almost achieved maximum recognition accuracy. However, it is still challenging to implement a cost-effective attendance system using low-cost webcams in a small classroom measuring approximately 8.5m*6m that can accommodate 20-25 people. The primary reason for this is that webcams are designed for personal use and can only recognize faces up to about 2 meters, making it impossible to recognize the faces of everyone in the classroom. In this study, we utilized YOLOX for small face detection, improved the features of distant small faces through Super-Resolution (SR), extracted facial features, and finally used Dlib for facial recognition. We also used OpenCV's optical flow to track the person's body frame's movement. Therefore, even if the person looks down or turns around, our system can still determine whether the person is still in the classroom. Experiments have demonstrated that this system can extend the webcam range to approximately 4-5 meters. Additionally, under clear facial conditions, this system can recognize whether the person has closed their eyes, serving as a reference for classroom participation.