Face Recognition Configuration in FortiCentral


FortiRecorder and FortiCentral supports easy to use face detection for your ever-evolving security needs. This recipe guides you through the process of getting the best facial recognition results and configuring facial recognition in FortiCentral.

 Note: FortiRecorder automatically initializes and configures facial recognition when detected. No manual configuration is required.

 Ensuring Best Results

There are a few steps you should take to ensure the best possible results for a face match 

  1. Make sure the individual at eye-level with the camera and is facing forward and looking into the camera.
  2. Keep the area well lit to avoid harsh shadows.
  3. Do not pose the subject of the picture against a bright background, since this will darken their faces and make facial recognition difficult.
  4. Keep the subject relatively still to reduce motion blur.



 FortiCentral Configuration

  1. Select the settings cog.
  2. Select Settings. 
  3. Select Analytics.
  4. Select the recorder receiving the results of the face analytics from the Face Recognition Database dropdown menu. 
  5. Enable “Face Recognition in a snapshot” to allow FortiRecorder to receive pictures of individuals for later identification.
  6. Enable “Display EBC/GPC Demos” to electronically welcome guests based on recognition.
  7. Enable “Show Analytics Alarms” to start a playback loop of the moment a person triggered an alarm.
  8. Select OK.



 Facial Analytics Configuration

Face Analytics is the first implementation of a video analytics engine for FortiCentral. Setting it up is simple

  1. Select a camera stream on a pane.
  2. Select the pane’s settings gear menu.
  3. Select Face Recognition from the Algorithm dropdown menu to activate analytics on the camera’s stream.
  4. Enable Visualize to enable the overlay of facial landmarks on a live video. This could help determine if the processing is functional. The displayed overlay has four different colors:
    White: Face detected but the quality is too low to generate a representation.
    Yellow: First frame taken as a snapshot and to generate facial representation.
    Magenta: The individual being tracked after initial detection.
    Cyan/Green: The individual track continued from face matching.
  5. Enable Show Snapshots to enable the display of the best snapshots associated with each track in a gallery.
  6. Enable Look For Names to activate a search for detected faces in the face database. If an enrolled person resembles someone in the database, their name is added to the bottom of the facial landmark.
  7. Select OK.