SAFR SDK do the following:

• Takes a video feed as input and performs both detection (locally) and recognition (against the cloud or locally deployed server).
• Object Tracker - Feeds into SDK raw frames. Responds via a listener to any changes to the analysis as to what is in the frame. Will generate events such as new face detection and face recognition.
• The images are submitted to a SAFR Server for recognition. You can configure the system to not do any recognitions, thus causing face detection information to be reported back to you.
• Event Reporter - Takes reported information and pushes it into the event reporter which will generate events back to the SAFR Server for the purpose of event analytics and generating events.