A Novel Masked Facial Recognition System Using CNN

The present work consists of the elaboration of models for facial recognition of individuals with a mask and to present suggestions of different methodologies that can be applied to adapt existing models. There is a need to evaluate the accuracy of these architectures in real scenarios, especially when it comes to applications that help prevent COVID-19. In this work convolutional neural network is applied to facial recognition. In addition, the ability of transfer learning to assist in the creation of new models will be evaluated. So, the steps include pre-processing of the chosen database, Model development, Training of models with hyperparameter variation, and Validation of models. In a problem of deep convolutional networks with multiclasses, the obtained results were able to classify the vast majority of images and be consistent with the current state of the art.