Cameras send cropped images of each person's face to SAFR edge processors (i.e. VIRGO, SAFR clients, and SDKs), which then send the images to SAFR Servers for recognition requests.