Face detection
Captures every face passing in front of the camera once and stores it as an event with a face image.
Face detection finds every face passing through the analytics area, tracks it and reports it once. Even if the same person stands in front of the camera for minutes, only one event is created; each event includes a cropped image of the face. This way, who passed through the entrances is archived with timestamped face images, even without setting up a gallery. Instead of watching hours of video in recordings, looking at faces in the event list is enough. Face detection is also the basis for selecting faces from the camera when adding a new person to the person gallery. All face processing is done on your own server.
How it works
Draw the area where faces will be seen; in front of doors and turnstiles is ideal.
Set the confidence threshold; raise it so only clear faces are recorded.
ProVife finds faces in the area and assigns each a tracking ID.
Each face is reported once; the event includes a cropped image of the face.
Events are listed by date and camera on the AI Playback and AI search screens.
Settings
| Area | The zone where faces are searched for. |
|---|---|
| Confidence threshold | The minimum confidence required to accept a face. |
| Analytics hours | The days and hours face detection runs. |
| Notification hours | The hours notifications are sent. |
| Alert method | Recording only, main screen notification or pop-up window. |
What it produces
- Face detection event (analytics.face.detected): face tracking ID, confidence value and face image.
- Event image with the face marked by a red box.
- Live number of faces in the area on the smart panel.
- An illustrated list of face events on the AI search screen.
What you can do with the event
- A WhatsApp notification with an image to security for a face seen at the door after hours.
- Showing the event in a pop-up window on the operator screen.
- Exporting events with face images to a PDF report.
- Forwarding the event via MQTT to visitor management software.
Where it is used
Office
Face images of everyone passing the building entrance are archived with timestamps.
Retail
Customer faces at the store door are found quickly for post-incident review.
Logistics
People passing through the warehouse staff door after hours are captured on video.
Residential complexes
People passing through the building entrance are logged for management.
Camera placement
- Position the camera to see faces close to eye level, at most slightly from above.
- Choose narrow passages where people walk toward the camera.
- Favor a narrow-angle lens so faces appear large enough in the image.
- Block daylight coming from behind the door, or use wide dynamic range.