<s>
Proposed	O
as	O
an	O
extension	O
of	O
image	B-Algorithm
epitomes	I-Algorithm
in	O
the	O
field	O
of	O
video	O
content	O
analysis	O
,	O
video	B-Algorithm
imprint	I-Algorithm
is	O
obtained	O
by	O
recasting	O
video	O
contents	O
into	O
a	O
fixed-sized	O
tensor	O
representation	O
regardless	O
of	O
video	O
resolution	O
or	O
duration	O
.	O
</s>
<s>
It	O
is	O
claimed	O
that	O
both	O
spatio-temporal	O
interdependences	O
are	O
accounted	O
for	O
and	O
redundancies	O
are	O
mitigated	O
during	O
the	O
computation	O
of	O
video	B-Algorithm
imprints	I-Algorithm
.	O
</s>
<s>
The	O
option	O
of	O
computing	O
video	B-Algorithm
imprints	I-Algorithm
exploiting	O
the	O
epitome	B-Algorithm
model	O
has	O
the	O
advantage	O
of	O
more	O
flexible	O
input	O
feature	O
formats	O
and	O
more	O
efficient	O
training	O
stage	O
for	O
video	O
content	O
analysis	O
.	O
</s>
