<s>
Facial	B-Operating_System
motion	I-Operating_System
capture	I-Operating_System
is	O
the	O
process	O
of	O
electronically	O
converting	O
the	O
movements	O
of	O
a	O
person	O
's	O
face	O
into	O
a	O
digital	O
database	O
using	O
cameras	O
or	O
laser	B-Algorithm
scanners	I-Algorithm
.	O
</s>
<s>
A	O
facial	B-Operating_System
motion	I-Operating_System
capture	I-Operating_System
database	O
describes	O
the	O
coordinates	O
or	O
relative	O
positions	O
of	O
reference	O
points	O
on	O
the	O
actor	O
's	O
face	O
.	O
</s>
<s>
Facial	B-Operating_System
motion	I-Operating_System
capture	I-Operating_System
is	O
related	O
to	O
body	O
motion	B-Application
capture	I-Application
,	O
but	O
is	O
more	O
challenging	O
due	O
to	O
the	O
higher	O
resolution	O
requirements	O
to	O
detect	O
and	O
track	O
subtle	O
expressions	O
possible	O
from	O
small	O
movements	O
of	O
the	O
eyes	O
and	O
lips	O
.	O
</s>
<s>
Facial	O
expression	O
capture	O
is	O
similar	O
to	O
facial	B-Operating_System
motion	I-Operating_System
capture	I-Operating_System
.	O
</s>
<s>
This	O
technology	O
is	O
discussed	O
and	O
demonstrated	O
at	O
CMU	O
,	O
IBM	O
,	O
University	O
of	O
Manchester	O
(	O
where	O
much	O
of	O
this	O
started	O
with	O
,	O
Gareth	O
Edwards	O
and	O
Chris	O
Taylor	O
)	O
and	O
other	O
locations	O
,	O
using	O
active	B-General_Concept
appearance	I-General_Concept
models	I-General_Concept
,	O
principal	B-Application
component	I-Application
analysis	I-Application
,	O
eigen	B-General_Concept
tracking	I-General_Concept
,	O
and	O
other	O
techniques	O
to	O
track	O
the	O
desired	O
facial	O
features	O
from	O
frame	O
to	O
frame	O
.	O
</s>
<s>
Typical	O
limitations	O
of	O
vision	O
based	O
approaches	O
are	O
resolution	O
and	O
frame	O
rate	O
,	O
both	O
of	O
which	O
are	O
decreasing	O
as	O
issues	O
as	O
high	O
speed	O
,	O
high	O
resolution	O
CMOS	B-Architecture
cameras	I-Architecture
become	O
available	O
from	O
multiple	O
sources	O
.	O
</s>
<s>
The	O
technology	O
for	O
markerless	O
face	B-Operating_System
tracking	I-Operating_System
is	O
related	O
to	O
that	O
in	O
a	O
Facial	O
recognition	O
system	O
,	O
</s>
<s>
of	O
video	O
,	O
resulting	O
in	O
face	B-Operating_System
tracking	I-Operating_System
.	O
</s>
<s>
2D	O
face	B-Operating_System
tracking	I-Operating_System
with	O
no	O
person-specific	O
training	O
;	O
their	O
system	O
was	O
also	O
amongst	O
the	O
best-performing	O
facial	O
recognition	O
systems	O
in	O
the	O
U.S.	O
Government	O
's	O
2002	O
Facial	O
Recognition	O
Vendor	O
Test	O
(	O
FRVT	O
)	O
.	O
</s>
<s>
whether	O
they	O
need	O
an	O
additional	O
source	O
of	O
information	O
such	O
as	O
projected	O
patterns	O
or	O
invisible	O
paint	O
such	O
as	O
used	O
in	O
the	O
Mova	B-Device
system	O
.	O
</s>
<s>
Using	O
digital	B-Device
cameras	I-Device
,	O
the	O
input	O
user	O
's	O
expressions	O
are	O
processed	O
to	O
provide	O
the	O
head	O
pose	B-Architecture
,	O
which	O
allows	O
the	O
software	O
to	O
then	O
find	O
the	O
eyes	O
,	O
nose	O
and	O
mouth	O
.	O
</s>
<s>
It	O
is	O
expected	O
that	O
this	O
will	O
become	O
a	O
major	O
input	B-Device
device	I-Device
for	O
computer	O
games	O
once	O
the	O
software	O
is	O
available	O
in	O
an	O
affordable	O
format	O
,	O
but	O
the	O
hardware	O
and	O
software	O
do	O
not	O
yet	O
exist	O
,	O
despite	O
the	O
research	O
for	O
the	O
last	O
15	O
years	O
producing	O
results	O
that	O
are	O
almost	O
usable	O
.	O
</s>
