Masked Face Recognition using YOLOv5 and Generative Adversarial Networks Algorithm for Attendance Machine

The COVID-19 pandemic has driven the development of touchless and automatic face recognition technologies. Detection and recognition of masked faces is a relevant solution in the effort to control the spread of the virus. To prevent the transmission of the COVID-19 virus from becoming more widespread, there is a need for a human contactless system that can analyze and identify individuals wearing masks. This research focuses on face verification that determines whether the faces belong to the same person and classifies the faces into specific identities for the attendance process. This research is built and designed on the case of imperfectly exposed faces by testing the capabilities of You Only Look Once version 5 (YOLOv5) and Generative Adversarial Networks (GAN) algorithms by looking at the results of their standard metrics including loss, accuracy, precision, recall and f1-score. The accuracy achieved in the masked face recognition system using GAN is 99.8%. On the other hand, the accuracy using the validation set by GAN is 99.7%. While the training loss that has been achieved in this study is 0.64 and the validation loss achieved is 6.45. With these results the system successfully detects according to the time of detection on the web attendance system. The YOLOv5 algorithm is implemented in the detection process to get the bounding box on the registered face. Meanwhile, the GAN algorithm is used for masked face recognition that has been integrated into the designed web attendance system. The various trainings that have been conducted resulted in an accuracy rate of 99.8% for real-time implementation. The proposed system successfully detects the name and its accuracy rate despite the use of masks.