Abstract: In order to effectively prevent the spread of COVID19 virus, almost everyone wears a mask during corona virus epidemic. This almost makes conventional face recognition technology ineffective in many cases, like community access control, face access control, facial attendance, facial security checks at train stations, etc. Therefore, it's very urgent to enhance the popularity performance of the prevailing face recognition technology on the masked faces. Most current advanced face recognition approaches are designed supported deep learning, which depend upon an outsized number of face samples. Deep Learning has proven its effectiveness in recognition and classification through image processing.
However, at the present, there are not any publicly available masked face recognition datasets. To this end, this work proposes three kinds of masked face datasets, including Masked Face Detection Dataset (MFDD), Real- world Masked Face Recognition Dataset (RMFRD) and Simulated Masked Face Recognition Dataset (SMFRD). Among them, to the only of our knowledge, RMFRD is currently the world’s largest real-world masked face dataset. These datasets are freely available to industry and academia, supported which various applications on masked faces are often developed. The multigranularity masked face recognition model we developed achieves 92% accuracy, exceeding the results reported by the industry.
Keywords: Deep Learning, Face Recognition Dataset,COVID19.
|
DOI:
10.17148/IJARCCE.2021.10771
