In the wake of the COVID-19, face masks have been considered an essential weapon in the fight against the pandemic. As governments around the world move to ease their lockdowns to allow their citizens out to mingle in the wider world again, many are enforcing mandatory mask-wearing measures in public. PHANTOM is an AI-powered system for detecting individuals faces and recognizing when he or she is not wearing face masks. The model was developed using DeepLearning technology using a labeled database of more than 200,000 faces. The model is accurate enough to handle rich variations of head position and orientation, skin color and resolution, as well as masks of several types and shapes. PHANTOM can also detect when someone is attempting to hack the system by covering the facial area to conceal the absence of a mask. To make it operational, PHANTOM was deployed using NVIDIA AI-edge devices, can plug into any CCTV camera and has the capability to process an incoming streaming video at a 21 frame per second with up to 1080ps high definition resolution with up to 500 faces in a single frame. The device can connect to WIFI in order to transmit signals for the appropriate action needed.
“There’s a growing evidence about the nature of COVID-19 that suggests masks may be necessary at all times when in social settings” Dr. Wang, director of Center for Policy, Outcomes and Prevention at Stanford University