Face recognition is an application-rich, widely-used technology. However, the rise of face masks during the COVID-19 epidemic poses a hurdle to precise facial recognition. Creating a facial recognition system that can recognize people even while they are wearing masks is the aim of this research. They will be able to open the locker once the face has been confirmed. If not, entry would be refused. First, faces in the stream are identified using face identification techniques such as MTCNN. Next, each recognized face's orientation and scale are normalized through the use of an alignment procedure. Feature extraction using a deep learning model, like Facenet, is the next stage. This model converts every aligned face into an embedding, which is a high-dimensional vector representation of the face's key traits and features. Finally, the embedded vectors are subjected to a softmax classifier to authenticate or categorize faces. Furthermore, activities such as face verification and grouping can be enabled by using similarity measures such as cosine similarity to assess the similarity of embedded vectors.