interval to go back to a central position will increase to a couple of minutes, and the
central position will be restored only if the last detected face has changed position.
• If in the room there are sources of noises, sometimes the camera is digitally pan/tilted to
the wrong direction or the algorithm needs more time to detect the speaker. It is also
possible that the audio algorithm is able to detect the speaker, but the face is not
recognized (too far or not in front position). In those cases:
o Check that the camera is not manually panned left or right respect to the
central position.
o Check that there are no sources of noise in the room.
o Check that the room is properly lighted.
o Assure that the distance of the talkers from the camera is not more that 3.0-
3.5 meters, and that the face of the talker is in front of the camera.
o Keep talking for a few seconds with your face towards the camera to have the
camera detect you.
• See Optimal Huddle Room Setup for supported arrangement of the room. Up to 4
different faces can be recognized for Speaker Tracking.
• The audio detection algorithm for Speaker Tracking is now available also when a BT or
USB microphone is used.
• Speaker Tracking is available for third-party apps and in USB Bar mode
Figure 4: Speaker Tracking Disabled vs Enabled
Auto-Framing (i.e. Group Framing)
When the Camera Tracking is active, and the current mode is Auto-Framing, the selected camera
is automatically pan/tilt/zoomed to include all the people present in the room (based on face
detection) inside the best possible frame.
If the people in the room change their relative positions, or enter/exit the room, the frame is
automatically adjusted to include all the present people (identified by their faces), with the best
pan/tilt/zoom positioning for a Group Framing.
Important
• Camera Auto Framing is applicable only to the embedded HD camera Camera Auto
Framing is available for third-party apps and in USB Bar mode
• Only faces in front of the camera are detected.
o Check that the room is properly lighted.
o Check that the room does not contain mirrors or other highly reflecting
surfaces, or pictures on the walls/furniture depicting faces.
o Assure that the distance of the people from the camera is not more that 3.0-
3.5 meters, and that the faces are towards the camera.
o Faces covered by masks could be not detected in this release.
• See Optimal Huddle Room Setup for supported arrangement of the room. Up to 4
different faces can be recognized for Auto Framing.
Figure 5: Auto Framing Disabled vs Enabled, with Four Faces Detected in the Room
Figure 6: Auto Framing Enabled: Faces Populating the Room Enlarge the Frame
Important – BETA BUILD
• Do not activate auto-framing when using an additional monitor. The video could
freeze or the device could hang.
To Next Page
Loading...
