--On Friday, July 24, 2015 7:30 PM +0000 Christian Huitema
<huitema(_at_)microsoft(_dot_)com> wrote:
John, we are in 2015. Computer vision can definitely help
tracking the speaker. You can find face tracking algorithms in
many packages, including open source variants. Computer
systems can easily track the speaker's face, crop the image
around that, and send the speaker-focused image over the
video. This is not rocket science. The pink box is obsolete.
Christian.
In my limited and very backward experience, face tracking
systems are really good at keeping track of the movements of one
speaker, even a moving one. When there are multiple people in
the front of the room, with multiple microphones, etc., plus a
microphone line or three in the middle of the room, maybe not so
good. That is, however, neither here nor there. My main point
was that we need to get more serious about making sure remote
participants can participate and tell what is going on and that
means resources. whether those resources are more people
monitoring or working the systems or more intelligent cameras
and software is a decision I'm happy to leave to those who are
more expert than I am (which doesn't take much). However, I
remain convinced that more resources and commitment are needed,
whether that is mostly weighted toward individual commitments
(standing in boxes, giving names clearly), additional staffing
to aim and select cameras, or higher- technology equipment.
best,
john