Masked Face Recognition Using MobileNet V2 with Transfer Learning

Corona virus (COVID-19) is once in a life time calamity that has resulted thousands of deaths and security concerns. People are using face masks on regular basis to protect themselves help reduce corona transmission. During the on-going coronavirus outbreak, one major priorities for researchers discover effective solution. As important parts obscured, identification verification becomes exceedingly difficult. The suggested method transfer learning MobileNet V2 based technology uses deep feature such as extraction model, identify problem masked identification. In first stage, we applying mask detector mask. Then, proposed approach datasets from Canadian Institute Advanced Research10 (CIFAR10), Modified National Standards Technology Database (MNIST), Real World Masked Face Recognition (RMFRD), Stimulated (SMFRD). model achieving recognition accuracy 99.82% with dataset. This article employs four pre-programmed models VGG16, VGG19, ResNet50 ResNet101. To extract features faces VGG16 99.30% accuracy, VGG19 99.54% 78.70% ResNet101 98.64% own comparative analysis shows, our performs better result all previous existing models. fundamental contribution this study monitor without decreases pace detect persons wearing masks.