Deep Learning-Based Models For Mask Detection In Times of Covid-19

COVID-19 has been one of the most severe problems in the world, causing a strong crisis in the social, economic and health fields, millions of people were infected, to avoid several preventive measures were taken, such as wearing masks. There is related research on the pandemic and application of Deep Learning methods. The objective of this study is to evaluate the per-formance of models based on Deep Learning such as YOLOv5, EfficientNet and MobileNetV2, for the detection of mask use, uasndo two different da-tasets for training, the first dataset with 1161 for YOLOv5 and the second with 10000 images for the other two models, in the test 60 images were con-sidered for all models, As a methodology, three processes were carried out: First process was the data collection, it was possible to collect the FMD data-base and the other of a project in Kaggle, second process was the data treat-ment, allowed to label and classify the images with more veneer and without mask, third process was manipulation of the model, allowed the training and development of the models. in the results, the YOLOv5 model had the best performance with 98.3% in accuracy, 96.7% in accuracy, 100% in recall and 98.3% in F1-Score. It is concluded that YOLO v5 is one of the algorithms with the highest performance for the detection of mask use, in addition an investigation that uses YOLOv5 was not found so far, likewise it can be im-plemented in future social research, for contagion environments.