Comparative Study of VGG16 and MobileNetV2 for Masked Face Recognition

Indonesia is one of the countries affected by coronavirus pandemic, which has taken too many lives. The pandemic forces us to continue wear masks daily, especially when working break chain spread coronavirus. Before face recognition for attendance used entire as input data, so results were accurate. However, during this all employees use masks, including attendance, can reduce level accuracy using masks. In research, we a deep learning technique recognize masked faces. We propose transfer pre-trained models perform feature extraction and classification image data. techniques due small amount data used. analyzed two models, namely VGG16 MobileNetV2. parameters batch size number epochs evaluate each model. best model obtained with value 32 50 in showed that MobileNetV2 was more accurate than VGG16, an 95.42%. study provide overview recognition.