Face Detection In Development Expected in Q4
What it does:
Determines whether a face is in a video feed when a person is detected in that same video.
Allows you to browse events by face. Allows you to run and validate chained classification and recognition models. Does not collect biometric information or create databases of faces.
Because Face Detection only allows you to know if a face is visible in the feed, but not whose face it is, it is not generally considered biometrics, so if your country or jurisdiction has banned the collection of biometric information, Face Detection may be able to still be used (consult your local laws).
Requires people detection and its camera placement requirements. Face Detection only runs once a person has been detected.
Additional Camera Placement Requirement:
Faces must be at least 20x20 pixels in size, requiring the person to be substantially closer than the person detector. At least 60% of the face must be visible. So, it works when the person is facing the camera or turned away as much as 40%.
If the camera is mounted high and looking downward, it will be looking down at people and it will not be effective. Face detection / recognition will be impossible if the horizontal angle of the camera is so high, as you will record video of the top of heads rather than of faces.
Interaction with Facial Recognition
In order to do facial recognition, survail must first locate a face. Face detections are used to determine the area of an image that has a candidate for facial recognition or visual face search.
Face Detection, People Grouping, and Search inside the Survail user interface:
People grouping works on either (or both) facial recognition and people re id data points. The Faces found are used as in the UI as the icons for the groups.
In the below example, the names of each face have been assigned by the user. If no names have been assigned, a number will be assigned to each person.
This model was made by NVIDIA.