Real-Time Facemask Detection and Analytics

The COVID-19 pandemic has emphasized the need for preventive measures in any pandemic situation. As we are now well aware, wearing a mask while stepping out in public places is imperative. In this context, our current work aims to build an end-to-end application-based system that scans people’s faces to determine whether they are wearing a face mask or not, or if they are wearing it improperly. It is a real-time system that provides feedback instantly. At the same time, different types of data obtained from the triple classification system are stored in a simple database to further visualize the same. The system we have created utilizes Single Shot Detector (SSD), a deep learning architecture, to extract a face from a given video frame, and MobileNetV2, another deep learning architecture, to classify it into one of the three categories mentioned above. Those without or improperly wearing a mask, are warned by the system. This work also makes use of QR (Quick Response) codes to identify the people in the video feed. Dataframes from Pandas in Python are used to organize the acquired data into a structured format which is then exported to Tableau. From this, we have visualized and observed the data and from the inferences made, meaningful insights have been arrived at, which can then be used to make informed decisions.