<s>
Object	B-General_Concept
detection	I-General_Concept
is	O
a	O
computer	O
technology	O
related	O
to	O
computer	B-Application
vision	I-Application
and	O
image	B-Algorithm
processing	I-Algorithm
that	O
deals	O
with	O
detecting	O
instances	O
of	O
semantic	O
objects	O
of	O
a	O
certain	O
class	O
(	O
such	O
as	O
humans	O
,	O
buildings	O
,	O
or	O
cars	O
)	O
in	O
digital	O
images	O
and	O
videos	O
.	O
</s>
<s>
Well-researched	O
domains	O
of	O
object	B-General_Concept
detection	I-General_Concept
include	O
face	B-General_Concept
detection	I-General_Concept
and	O
pedestrian	B-General_Concept
detection	I-General_Concept
.	O
</s>
<s>
Object	B-General_Concept
detection	I-General_Concept
has	O
applications	O
in	O
many	O
areas	O
of	O
computer	B-Application
vision	I-Application
,	O
including	O
image	B-General_Concept
retrieval	I-General_Concept
and	O
video	O
surveillance	O
.	O
</s>
<s>
It	O
is	O
widely	O
used	O
in	O
computer	B-Application
vision	I-Application
tasks	O
such	O
as	O
image	B-Application
annotation	I-Application
,	O
vehicle	O
counting	O
,	O
activity	B-Application
recognition	I-Application
,	O
face	B-General_Concept
detection	I-General_Concept
,	O
face	O
recognition	O
,	O
video	B-Algorithm
object	I-Algorithm
co-segmentation	I-Algorithm
.	O
</s>
<s>
It	O
is	O
also	O
used	O
in	O
tracking	B-Operating_System
objects	I-Operating_System
,	O
for	O
example	O
tracking	O
a	O
ball	O
during	O
a	O
football	O
match	O
,	O
tracking	O
movement	O
of	O
a	O
cricket	O
bat	O
,	O
or	O
tracking	O
a	O
person	O
in	O
a	O
video	O
.	O
</s>
<s>
Often	O
,	O
the	O
test	O
images	O
are	O
sampled	O
from	O
a	O
different	O
data	O
distribution	O
,	O
making	O
the	O
object	B-General_Concept
detection	I-General_Concept
task	O
significantly	O
more	O
difficult	O
.	O
</s>
<s>
Among	O
other	O
uses	O
,	O
cross-domain	O
object	B-General_Concept
detection	I-General_Concept
is	O
applied	O
in	O
autonomous	O
driving	O
,	O
where	O
models	O
can	O
be	O
trained	O
on	O
a	O
vast	O
amount	O
of	O
video	O
game	O
scenes	O
,	O
since	O
the	O
labels	O
can	O
be	O
generated	O
without	O
manual	O
labor	O
.	O
</s>
<s>
Every	O
object	O
class	O
has	O
its	O
own	O
special	O
features	B-Algorithm
that	O
help	O
in	O
classifying	O
the	O
class	O
–	O
for	O
example	O
all	O
circles	O
are	O
round	O
.	O
</s>
<s>
Object	O
class	O
detection	O
uses	O
these	O
special	O
features	B-Algorithm
.	O
</s>
<s>
A	O
similar	O
approach	O
is	O
used	O
for	O
face	O
identification	O
where	O
eyes	O
,	O
nose	O
,	O
and	O
lips	O
can	O
be	O
found	O
and	O
features	B-Algorithm
like	O
skin	O
color	O
and	O
distance	O
between	O
eyes	O
can	O
be	O
found	O
.	O
</s>
<s>
Methods	O
for	O
object	B-General_Concept
detection	I-General_Concept
generally	O
fall	O
into	O
either	O
neural	O
network-based	O
or	O
non-neural	O
approaches	O
.	O
</s>
<s>
For	O
non-neural	O
approaches	O
,	O
it	O
becomes	O
necessary	O
to	O
first	O
define	O
features	B-Algorithm
using	O
one	O
of	O
the	O
methods	O
below	O
,	O
then	O
using	O
a	O
technique	O
such	O
as	O
support	B-Algorithm
vector	I-Algorithm
machine	I-Algorithm
(	O
SVM	B-Algorithm
)	O
to	O
do	O
the	O
classification	O
.	O
</s>
<s>
On	O
the	O
other	O
hand	O
,	O
neural	O
techniques	O
are	O
able	O
to	O
do	O
end-to-end	O
object	B-General_Concept
detection	I-General_Concept
without	O
specifically	O
defining	O
features	B-Algorithm
,	O
and	O
are	O
typically	O
based	O
on	O
convolutional	B-Architecture
neural	I-Architecture
networks	I-Architecture
(	O
CNN	B-Architecture
)	O
.	O
</s>
<s>
Region	B-Algorithm
Proposals	I-Algorithm
(	O
R-CNN	B-Algorithm
,	O
Fast	O
R-CNN	B-Algorithm
,	O
Faster	O
R-CNN	B-Algorithm
,	O
cascade	O
R-CNN	B-Algorithm
.	O
)	O
</s>
