I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
1
1
,
No
.
2
,
A
p
r
il
2
0
2
1
,
p
p
.
1
6
5
6
~
1
6
6
5
I
SS
N:
2
0
8
8
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/
ijece
.
v
1
1
i
2
.
pp
1
6
5
6
-
1
6
6
5
1656
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
S
ea
rching
surv
eil
la
nce video
conte
nts using
conv
o
lu
tiona
l
neura
l net
wo
rk
Dua
a
M
o
ha
m
m
a
d,
I
na
d Alj
a
rr
a
h,
M
o
a
t
h J
a
rr
a
h
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter E
n
g
in
e
e
rin
g
,
Jo
r
d
a
n
Un
i
v
e
rsity
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
l
o
g
y
(JU
S
T),
Jo
rd
a
n
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
2
4
,
2
0
2
0
R
ev
is
ed
Au
g
4
,
2
0
2
0
Acc
ep
ted
No
v
4
,
2
0
2
0
M
a
n
u
a
l
v
i
d
e
o
i
n
sp
e
c
ti
o
n
,
se
a
rc
h
in
g
,
a
n
d
a
n
a
ly
z
in
g
is
e
x
h
a
u
stin
g
a
n
d
in
e
fficie
n
t.
Th
is
p
a
p
e
r
p
re
se
n
ts
a
n
i
n
telli
g
e
n
t
sy
ste
m
t
o
se
a
rc
h
su
rv
e
il
lan
c
e
v
id
e
o
c
o
n
ten
ts
u
si
n
g
d
e
e
p
lea
rn
i
n
g
.
Th
e
p
r
o
p
o
se
d
s
y
ste
m
re
d
u
c
e
d
t
h
e
a
m
o
u
n
t
o
f
wo
rk
th
a
t
is
n
e
e
d
e
d
to
p
e
rfo
r
m
v
id
e
o
se
a
rc
h
in
g
a
n
d
imp
r
o
v
e
d
th
e
sp
e
e
d
a
n
d
a
c
c
u
ra
c
y
.
A
p
re
-
train
e
d
VG
G
-
1
6
CNN
s
m
o
d
e
l
is
u
se
d
f
o
r
d
a
tas
e
t
train
in
g
.
In
a
d
d
i
ti
o
n
,
k
e
y
fra
m
e
s
o
f
v
i
d
e
o
s
we
re
e
x
trac
ted
i
n
o
rd
e
r
to
sa
v
e
sp
a
c
e
,
re
d
u
c
e
th
e
a
m
o
u
n
t
o
f
w
o
rk
,
a
n
d
re
d
u
c
e
th
e
e
x
e
c
u
ti
o
n
ti
m
e
.
Th
e
e
x
trac
ted
k
e
y
fr
a
m
e
s
we
re
p
ro
c
e
ss
e
d
u
sin
g
th
e
s
o
b
e
l
o
p
e
ra
to
r
e
d
g
e
d
e
tec
to
r
a
n
d
t
h
e
m
a
x
-
p
o
o
li
n
g
in
o
r
d
e
r
to
e
li
m
in
a
te
re
d
u
n
d
a
n
c
y
.
T
h
i
s
in
c
re
a
se
s
c
o
m
p
a
c
ti
o
n
a
n
d
a
v
o
id
s
sim
il
a
rit
ie
s
b
e
twe
e
n
e
x
trac
ted
fra
m
e
s.
A
tex
t
fil
e
,
th
a
t
c
o
n
tain
s
k
e
y
fra
m
e
in
d
e
x
,
ti
m
e
o
f
o
c
c
u
rre
n
c
e
,
a
n
d
t
h
e
c
las
sifica
ti
o
n
o
f
th
e
VGG
-
1
6
m
o
d
e
l
is p
ro
d
u
c
e
d
.
T
h
e
tex
t
fil
e
e
n
a
b
les
h
u
m
a
n
s
t
o
e
a
sily
se
a
rc
h
fo
r
o
b
jec
ts
o
f
in
tere
st.
VIRA
T
a
n
d
IVY
LAB
d
a
tas
e
ts
we
re
u
se
d
in
t
h
e
e
x
p
e
rime
n
ts.
In
a
d
d
it
i
o
n
,
1
2
8
d
iff
e
re
n
t
c
las
se
s
we
re
id
e
n
ti
fied
in
th
e
d
a
tas
e
ts.
Th
e
c
las
se
s
re
p
re
se
n
t
imp
o
r
tan
t
o
b
jec
ts
f
o
r
su
r
v
e
il
lan
c
e
sy
ste
m
s.
Ho
we
v
e
r,
u
se
rs
c
a
n
id
e
n
ti
fy
o
t
h
e
r
c
las
se
s
a
n
d
u
ti
l
ize
th
e
p
ro
p
o
se
d
m
e
th
o
d
o
l
o
g
y
.
Ex
p
e
rime
n
ts
a
n
d
e
v
a
lu
a
ti
o
n
s
h
o
we
d
th
a
t
th
e
p
ro
p
o
se
d
s
y
ste
m
o
u
tp
e
rfo
rm
e
d
e
x
isti
n
g
m
e
th
o
d
s
in
a
n
o
r
d
e
r
o
f
m
a
g
n
it
u
d
e
.
Th
e
s
y
ste
m
a
c
h
iev
e
d
th
e
b
e
st
re
su
lt
s in
s
p
e
e
d
wh
il
e
p
ro
v
i
d
in
g
a
h
ig
h
a
c
c
u
ra
c
y
i
n
c
las
sifica
ti
o
n
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
Key
f
r
am
es
So
b
el
d
etec
to
r
VGG
-
16
Vid
eo
co
n
ten
t a
n
aly
s
is
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Mo
ath
J
ar
r
ah
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
J
o
r
d
an
Un
iv
e
r
s
ity
o
f
Scien
ce
a
n
d
T
ec
h
n
o
lo
g
y
C
o
m
p
u
ter
an
d
I
n
f
o
r
m
atio
n
C
o
lleg
e,
P.O.
B
o
x
3
0
3
0
,
I
r
b
id
,
2
2
1
1
0
,
J
o
r
d
an
E
m
ail:
m
jar
r
ah
@
ju
s
t.e
d
u
.
jo
1.
I
NT
RO
D
UCT
I
O
N
Vid
eo
co
n
te
n
t
an
al
y
s
is
(
VC
A
)
is
th
e
m
et
h
o
d
o
f
an
aly
zi
n
g
v
id
eo
s
tr
ea
m
s
to
d
etec
t
an
d
d
eter
m
in
e
tem
p
o
r
al
an
d
s
p
atial
ev
en
ts
to
f
in
d
wh
at
a
v
id
eo
r
ep
r
esen
ts
an
d
th
e
ty
p
e
o
f
in
f
o
r
m
atio
n
it h
as [
1
]
.
I
t is u
s
ed
b
y
d
if
f
er
en
t
ap
p
licatio
n
s
th
at
r
eq
u
ir
e
h
ig
h
s
ec
u
r
ity
to
d
etec
t
in
tr
u
d
er
s
an
d
ab
n
o
r
m
al
ev
e
n
ts
esp
ec
ially
f
r
au
d
ac
tio
n
s
[
2
]
.
Air
p
o
r
ts
,
h
o
tels
,
b
an
k
s
,
an
d
o
t
h
er
p
u
b
lic
p
lace
s
n
ee
d
VC
A
to
en
s
u
r
e
a
s
ec
u
r
e
en
v
ir
o
n
m
en
t
f
o
r
clien
ts
an
d
s
taf
f
.
Ma
r
k
et
o
w
n
er
s
ca
n
im
p
r
o
v
e
th
eir
p
r
o
d
u
ct
iv
ity
b
y
u
n
d
er
s
tan
d
in
g
th
eir
c
u
s
to
m
er
r
ea
ctio
n
s
,
n
ee
d
s
,
an
d
d
esire
s
.
Mo
r
e
o
v
er
,
VC
A
i
s
o
f
h
ig
h
im
p
o
r
tan
ce
i
n
s
u
b
way
s
tatio
n
s
to
d
etec
t
d
an
g
er
o
u
s
s
itu
atio
n
s
an
d
s
ec
u
r
e
th
e
a
r
ea
s
[
3
]
.
T
r
a
n
s
p
o
r
tatio
n
s
y
s
tem
s
r
ely
o
n
V
C
A
to
en
s
u
r
e
p
ass
en
g
e
r
s
s
ec
u
r
ity
,
v
e
h
icle
co
n
tr
o
l,
a
n
d
b
e
t
t
e
r
t
r
a
c
k
i
n
g
m
e
t
h
o
d
s
[
4
]
.
F
u
r
t
h
e
r
m
o
r
e
,
v
i
d
e
o
a
n
a
l
y
s
i
s
i
s
u
s
e
d
t
o
d
e
t
e
c
t
u
n
d
e
r
w
a
t
e
r
o
b
j
e
c
t
s
[
5
]
.
H
e
n
c
e
,
v
id
eo
co
n
ten
t
s
ea
r
ch
in
g
s
y
s
tem
s
m
u
s
t
b
e
ab
le
to
s
ea
r
ch
th
e
v
id
eo
i
n
a
f
ast
way
.
Ho
wev
er
,
th
e
am
o
u
n
t
o
f
d
ata
th
at
is
r
ep
o
r
ted
b
y
s
u
r
v
eillan
ce
ca
m
er
as
is
h
u
g
e,
wh
ich
p
u
ts
a
ch
allen
g
e
o
n
t
h
e
s
ea
r
ch
p
r
o
ce
s
s
.
I
t
is
im
p
o
r
tan
t
t
o
o
r
g
a
n
i
z
e
a
n
d
s
e
a
r
c
h
t
h
e
c
o
n
t
e
n
t
s
o
f
v
i
d
e
o
s
i
n
a
w
a
y
w
h
e
r
e
u
s
e
r
s
c
a
n
f
i
n
d
o
b
j
e
c
t
s
o
f
i
n
t
e
r
e
s
t
w
i
t
h
i
n
a
s
h
o
r
t
t
i
m
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
ng
I
SS
N:
2
0
8
8
-
8
7
0
8
S
ea
r
ch
in
g
s
u
r
ve
illa
n
ce
vid
e
o
co
n
ten
ts
u
s
in
g
co
n
vo
lu
tio
n
a
l n
eu
r
a
l n
etw
o
r
k
(
Du
a
a
Mo
h
a
m
ma
d
)
1657
Dee
p
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NNs)
ar
e
u
s
ed
in
d
if
f
er
en
t
d
o
m
ain
s
s
u
ch
as
im
ag
e
class
if
icatio
n
[
6
,
7
]
,
o
b
ject
d
etec
tio
n
[
8
,
9
]
,
a
n
d
n
atu
r
al
l
an
g
u
ag
e
p
r
o
ce
s
s
in
g
[
1
0
,
1
1
]
.
T
h
e
n
etwo
r
k
s
ar
e
co
m
p
u
tatio
n
ally
in
ten
s
iv
e
tas
k
s
.
Ho
wev
e
r
,
th
ey
p
r
o
v
id
e
h
i
g
h
ac
cu
r
ac
y
i
n
o
b
ject
d
etec
ti
o
n
.
C
NNs
co
n
tain
th
r
ee
ty
p
es
o
f
lay
e
r
s
wh
ich
a
r
e:
co
n
v
o
lu
tio
n
al
lay
er
s
,
p
o
o
l
in
g
lay
er
s
,
a
n
d
f
u
lly
co
n
n
ec
ted
lay
er
s
.
I
n
s
o
m
e
ap
p
licatio
n
s
,
an
ac
tiv
atio
n
f
u
n
ctio
n
(
i.e
.
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U)
)
lay
e
r
f
o
llo
w
s
th
e
co
n
v
o
lu
tio
n
al
lay
er
s
.
T
h
er
e
h
a
v
e
b
ee
n
s
u
cc
es
s
f
u
l
ap
p
licatio
n
s
o
f
C
NNs
f
o
r
im
ag
e
class
if
icatio
n
s
u
ch
as
Alex
Net
m
o
d
el
wh
ich
a
c
h
i
e
v
e
d
1
5
.
3
%
t
o
p
-
5
t
e
s
t
e
r
r
o
r
r
a
t
e
i
n
I
L
S
V
R
C
-
2
0
1
2
[
6
]
.
I
n
a
d
d
i
t
i
o
n
,
Z
F
N
e
t
[
1
2
]
a
n
d
G
o
o
g
L
e
N
e
t
m
o
d
e
l
[
1
3
]
ac
h
iev
ed
ex
ce
llen
t
p
er
f
o
r
m
an
ce
.
VGG
-
1
6
m
o
d
el
an
d
R
esNet
m
o
d
el
wer
e
d
es
ig
n
ed
b
y
He
et
a
l
.
f
o
r
o
b
ject
d
etec
tio
n
[
1
4
,
1
5
]
.
C
NNs
ar
e
also
u
s
ed
in
f
ac
e
r
ec
o
g
n
itio
n
s
u
ch
as
th
e
wo
r
k
b
y
L
i
et
a
l
.
[
1
6
]
.
L
i
et
a
l
.
p
r
o
p
o
s
ed
a
m
u
lti
-
r
eso
lu
tio
n
C
NN
ca
s
ca
d
e
f
o
r
f
ast
f
ac
e
d
ete
ctio
n
.
Fu
r
th
e
r
m
o
r
e,
Su
n
et
a
l
.
p
r
o
p
o
s
ed
two
d
ee
p
n
eu
r
al
n
etwo
r
k
ar
ch
itectu
r
es,
Dee
p
I
D3
,
an
d
ac
h
iev
ed
9
9
.
5
3
%
ac
cu
r
ac
y
in
L
FW
f
ac
e
v
er
if
icatio
n
an
d
9
6
.
0
%
L
FW
r
an
k
-
1
f
ac
e
id
en
tific
atio
n
[
1
7
]
.
I
n
a
d
d
itio
n
,
C
NNs
h
av
e
b
ee
n
u
s
ed
f
o
r
m
ed
ical
im
a
g
e
an
aly
s
is
[
1
8
,
1
9
]
.
I
n
1
9
9
4
,
t
h
e
C
NNs we
r
e
u
s
ed
to
d
etec
t th
e
m
icr
o
-
ca
lcif
icatio
n
s
in
d
ig
ital m
am
m
o
g
r
ap
h
y
[
2
0
]
.
M
u
c
h
o
f
t
h
e
p
a
s
t
a
n
d
o
n
-
g
o
i
n
g
r
e
s
e
a
r
c
h
a
i
m
s
t
o
a
n
a
l
y
z
e
v
i
d
e
o
c
o
n
t
e
n
t
s
u
s
i
n
g
d
i
f
f
e
r
e
n
t
m
e
t
h
o
d
s
.
L
a
o
et
a
l
.
p
r
o
p
o
s
ed
a
s
y
s
tem
f
o
r
s
em
an
tical
an
aly
s
is
o
f
h
u
m
an
b
eh
av
i
o
r
s
in
a
m
o
n
o
cu
lar
s
u
r
v
eillan
ce
v
id
eo
c
a
p
t
u
r
e
d
b
y
a
c
o
n
s
u
m
e
r
c
a
m
e
r
a
[
2
1
]
.
T
h
e
a
u
t
h
o
r
s
i
n
c
o
r
p
o
r
a
t
e
d
a
t
r
a
j
e
c
t
o
r
y
e
s
t
i
m
a
t
i
o
n
m
e
t
h
o
d
b
e
s
i
d
e
s
h
u
m
a
n
-
b
o
d
y
m
o
d
e
l
i
n
g
t
o
c
o
m
p
r
e
h
e
n
d
t
h
e
s
e
m
a
n
t
i
c
a
n
a
l
y
s
i
s
o
f
h
u
m
a
n
a
c
t
i
v
i
t
i
e
s
a
n
d
e
v
e
n
t
s
i
n
v
i
d
e
o
s
eq
u
en
ce
s
.
Z
h
ao
an
d
C
ai
em
p
lo
y
ed
a
s
h
o
r
t
-
tim
e
m
em
o
r
y
m
o
d
el
t
o
s
eg
m
en
t
a
g
i
v
en
v
i
d
eo
an
d
to
s
p
ec
if
y
th
e
s
ce
n
e
im
p
o
r
ta
n
ce
f
o
r
k
ey
f
r
a
m
e
s
e
x
t
r
a
c
t
i
o
n
[
2
2
]
.
B
e
r
t
i
n
i
e
t
a
l
.
p
r
e
s
e
n
t
e
d
a
f
r
a
m
e
w
o
r
k
f
o
r
e
v
e
n
t
a
n
d
o
b
j
e
c
t
e
x
t
r
a
c
t
i
o
n
o
f
s
o
c
c
e
r
v
i
d
e
o
s
[
2
3
]
.
T
h
e
au
th
o
r
s
ap
p
lied
s
em
an
tic
tr
an
s
co
d
in
g
to
t
h
e
f
r
am
es
th
at
c
o
n
tain
ev
en
ts
an
d
h
u
m
an
f
ac
es.
Fo
u
r
class
es
o
f
ev
en
ts
wer
e
d
etec
ted
in
th
eir
f
r
am
ewo
r
k
.
Ko
le
k
ar
in
tr
o
d
u
ce
d
a
p
r
o
b
ab
ilis
tic
ap
p
r
o
ac
h
f
o
r
v
id
eo
an
aly
s
is
an
d
in
d
ex
in
g
,
b
ased
o
n
b
a
y
esian
b
elief
n
etwo
r
k
(
B
B
N)
[
2
4
]
.
T
h
ey
u
s
ed
a
h
ier
ar
ch
al
class
if
icatio
n
f
r
a
m
ewo
r
k
to
ex
tr
ac
t
f
ea
tu
r
es
f
r
o
m
v
id
eo
s
an
d
th
en
th
e
B
B
N
ass
ig
n
s
th
e
s
em
an
tic
lab
el
f
o
r
ea
ch
ev
en
t
in
v
id
eo
clip
s
.
Fu
r
th
er
m
o
r
e
,
C
h
en
an
d
Z
h
an
g
p
r
o
p
o
s
ed
a
v
id
e
o
co
n
ten
t
a
n
aly
s
is
s
y
s
tem
u
s
i
n
g
a
u
t
o
r
e
g
r
e
s
s
i
v
e
(
A
R
)
m
o
d
e
l
i
n
g
t
o
m
o
d
e
l
t
h
e
f
e
a
t
u
r
e
s
e
q
u
e
n
c
e
o
f
f
r
a
m
e
s
o
v
e
r
t
i
m
e
[
2
5
]
.
S
u
n
e
t
a
l
.
i
n
t
r
o
d
u
c
e
d
a
v
i
d
e
o
a
n
a
l
y
s
i
s
m
e
t
h
o
d
t
h
a
t
d
e
p
e
n
d
s
o
n
c
o
l
o
r
d
i
s
t
r
i
b
u
t
i
o
n
s
b
e
t
w
e
e
n
f
r
a
m
e
s
[
2
6
]
.
T
h
e
d
is
tr
ib
u
tio
n
s
a
r
e
u
s
ed
to
s
ea
r
ch
th
e
v
id
e
o
f
r
a
m
es.
Fu
r
th
er
m
o
r
e
,
Sh
ar
if
et
a
l
.
p
r
o
p
o
s
ed
a
d
etec
tio
n
s
y
s
tem
u
s
in
g
en
t
r
o
p
y
m
ea
s
u
r
e
to
p
ar
titi
o
n
a
v
id
eo
in
to
s
m
a
ll
s
p
atial
-
tem
p
o
r
al
p
atch
es
[
2
7
]
.
Ho
wev
er
,
th
eir
s
y
s
tem
m
ea
s
u
r
es
th
e
b
ac
k
g
r
o
u
n
d
f
ea
tu
r
es
o
n
ly
an
d
d
o
es
n
o
t
ass
ess
th
e
b
eh
av
io
r
s
o
f
in
d
iv
i
d
u
als
an
d
m
o
v
in
g
o
b
j
ec
ts
.
C
er
n
ek
o
v
et
a
l
.
ex
t
r
ac
ted
k
ey
f
r
am
es
u
s
in
g
m
u
tu
al
in
f
o
r
m
atio
n
a
n
d
j
o
in
t
en
tr
o
p
y
f
o
r
ea
s
e
o
f
s
ea
r
ch
o
f
v
id
eo
co
n
ten
ts
[
2
8
]
.
Z
en
g
et
a
l
.
ap
p
lied
a
b
lo
c
k
-
b
ased
m
ar
k
o
v
r
an
d
o
m
f
ield
(
MRF
)
m
o
d
el
to
s
eg
m
en
t
th
e
m
o
v
in
g
o
b
jects
o
b
tain
ed
f
r
o
m
v
id
eo
f
r
am
es
to
an
aly
ze
v
id
e
o
co
n
ten
ts
;
an
d
u
s
ed
b
ac
k
tr
ac
k
in
g
to
s
elec
t
t
h
e
k
e
y
f
r
am
es
[
2
9
]
.
Z
h
o
u
et
a
l
.
p
r
o
p
o
s
ed
a
n
o
n
-
u
n
i
f
o
r
m
s
a
m
p
l
i
n
g
m
e
t
h
o
d
a
s
w
e
l
l
a
s
a
s
i
m
p
l
e
u
n
i
f
o
r
m
s
a
m
p
l
e
r
(
U
n
i
)
f
o
r
s
u
m
m
a
r
i
z
i
n
g
l
o
n
g
v
i
d
e
o
c
o
n
t
e
n
t
[
3
0
]
.
Af
ter
war
d
,
th
e
p
r
o
p
o
s
ed
s
am
p
lin
g
m
eth
o
d
e
x
tr
ac
ts
im
p
o
r
tan
t
f
ea
tu
r
es
an
d
p
r
o
d
u
ce
a
s
h
o
r
t
v
id
eo
wh
er
e
u
s
er
s
c
an
s
ea
r
ch
it
f
aster
.
T
h
eir
s
y
s
tem
tak
e
s
a
s
ec
o
n
d
to
r
etim
e
e
ac
h
v
id
e
o
a
n
d
ten
s
ec
o
n
d
s
to
r
en
d
er
ea
ch
f
r
am
e
.
On
th
e
o
th
er
h
a
n
d
,
B
ai
et
a
l
.
in
tr
o
d
u
ce
d
a
v
id
e
o
s
em
an
tic
co
n
ten
t a
n
aly
s
is
f
r
am
ewo
r
k
th
at
d
ep
en
d
s
o
n
d
o
m
ai
n
o
n
to
lo
g
y
[
3
1
]
.
T
h
e
au
th
o
r
s
u
s
ed
lo
w
-
lev
e
l
alg
o
r
ith
m
s
to
e
x
t
r
ac
t
b
o
th
h
ig
h
le
v
e
l
an
d
lo
w
-
l
ev
el
f
ea
tu
r
es
i
n
th
e
v
id
eo
s
.
T
h
e
v
id
e
o
ev
en
t
d
ete
ctio
n
is
p
er
f
o
r
m
e
d
m
an
u
ally
.
Fo
g
g
ia
et
a
l
.
in
tr
o
d
u
ce
d
a
f
i
r
e
d
etec
tio
n
s
y
s
tem
an
aly
s
is
f
o
r
s
u
r
v
eillan
ce
v
id
eo
s
[
3
2
]
.
T
h
eir
p
r
o
p
o
s
ed
s
y
s
t
em
r
elies
o
n
co
lo
r
,
s
h
ap
e
v
a
r
iatio
n
,
an
d
m
o
tio
n
an
aly
s
is
to
d
etec
t
a
f
ir
e.
T
h
ey
u
s
ed
YUV
co
lo
r
s
p
ac
e
a
n
d
th
e
s
ca
le
-
in
v
ar
ian
t
f
ea
t
u
r
e
tr
an
s
f
o
r
m
(
SIFT
)
d
escr
ip
to
r
s
f
o
r
b
lo
b
s
m
o
v
em
en
ts
’
d
etec
tio
n
.
T
h
en
th
e
m
u
l
ti
-
ex
p
er
t
s
y
s
tem
(
ME
S)
p
r
o
d
u
ce
s
th
e
p
r
ed
ictio
n
.
T
h
e
s
y
s
tem
ac
h
iev
ed
a
c
o
n
s
id
er
ab
le
r
ate
o
f
f
alse p
o
s
itiv
es.
T
h
is
p
ap
er
p
r
o
p
o
s
es
a
s
y
s
tem
f
o
r
s
ea
r
ch
in
g
s
u
r
v
eillan
ce
v
id
eo
co
n
ten
ts
(
SS
VC
)
.
SS
V
C
s
y
s
tem
u
s
es
C
NN
f
o
r
o
b
ject
r
ec
o
g
n
itio
n
an
d
class
if
icatio
n
s
.
Sp
ec
if
ically
,
it u
s
es th
e
V
GGNe
t m
o
d
el
wh
ich
was d
ev
elo
p
ed
b
y
Ox
f
o
r
d
'
s
v
is
u
al
g
eo
m
et
r
y
g
r
o
u
p
(
VGG)
[
1
4
]
.
T
h
e
m
o
d
e
l
s
co
r
ed
th
e
f
ir
s
t
p
lace
i
n
im
a
g
e
lo
ca
lizatio
n
a
n
d
th
e
s
ec
o
n
d
p
lace
in
im
ag
e
cl
ass
if
icatio
n
[
3
3
]
.
T
h
er
e
ar
e
d
if
f
er
en
t
co
n
f
ig
u
r
atio
n
s
o
f
th
e
VGGN
et
s
u
ch
as
V
G
G
-
1
6
a
n
d
V
G
G
-
1
9
.
V
G
G
-
1
6
h
a
s
1
3
c
o
n
v
o
l
u
t
i
o
n
a
l
l
a
y
e
r
s
a
n
d
3
f
u
l
l
y
c
o
n
n
e
c
t
e
d
l
a
y
e
r
s
.
S
S
V
C
s
y
s
t
e
m
u
s
es
V
G
G
-
1
6
.
I
t
g
e
n
e
r
a
t
e
s
a
t
e
x
t
f
i
l
e
t
h
a
t
c
o
n
t
a
i
n
s
d
i
f
f
e
r
e
n
t
c
l
a
s
s
e
s
o
f
t
h
e
d
e
t
e
c
t
e
d
o
b
j
e
c
t
s
a
n
d
t
h
e
t
i
m
e
o
f
ap
p
ea
r
an
ce
o
f
ea
ch
o
b
ject
in
th
e
v
id
eo
,
as
well
as
th
e
f
r
am
e
in
d
ex
.
T
o
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
an
c
e,
SS
VC
s
y
s
t
e
m
p
r
o
c
e
s
s
e
s
o
n
l
y
a
s
p
e
c
i
f
i
c
n
u
m
b
e
r
o
f
f
r
a
m
e
s
t
h
a
t
h
o
l
d
m
o
s
t
o
f
t
h
e
i
n
f
o
r
m
a
t
i
o
n
(
k
e
y
f
r
a
m
e
s
)
.
F
u
r
t
h
e
r
m
o
r
e
,
a
m
a
t
c
h
i
n
g
p
r
o
c
e
s
s
i
s
p
e
r
f
o
r
m
e
d
t
o
e
l
i
m
i
n
a
t
e
r
e
d
u
n
d
a
n
t
k
e
y
f
r
a
m
e
s
.
A
s
a
r
e
s
u
l
t
,
V
G
G
-
1
6
e
l
i
m
i
n
a
t
e
s
m
a
n
y
o
f
t
h
e
e
x
t
r
a
c
t
e
d
f
r
a
m
e
s
,
w
h
i
c
h
i
m
p
r
o
v
e
s
t
h
e
s
p
e
e
d
o
f
t
h
e
s
y
s
t
e
m
.
V
I
R
A
T
a
n
d
I
V
Y
L
A
B
d
a
t
a
s
e
t
s
a
r
e
u
s
e
d
i
n
t
h
e
e
x
p
e
r
i
m
e
n
t
s
.
R
e
s
u
l
t
s
s
h
o
w
t
h
a
t
S
S
V
C
s
y
s
t
e
m
o
u
t
p
e
r
f
o
r
m
s
p
r
e
v
i
o
u
s
l
y
p
r
o
p
o
s
e
d
m
eth
o
d
s
in
ter
m
s
o
f
s
p
ee
d
.
T
h
e
r
es
t
o
f
th
is
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
ws.
Sectio
n
2
p
r
esen
ts
th
e
m
et
h
o
d
o
l
o
g
y
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
t
e
m
a
n
d
t
h
e
d
i
f
f
e
r
e
n
t
t
e
c
h
n
i
q
u
e
s
t
h
a
t
a
r
e
e
m
p
l
o
y
e
d
t
o
e
n
h
a
n
c
e
v
i
d
e
o
a
n
a
l
y
s
i
s
.
A
d
d
i
t
i
o
n
a
l
l
y
,
t
h
e
ex
p
er
im
en
ts
ar
e
p
r
esen
te
d
to
ev
alu
ate
ea
c
h
tech
n
iq
u
e.
Sectio
n
3
co
n
clu
d
e
s
th
e
p
ap
er
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
2
,
Ap
r
il 2
0
2
1
:
1
6
5
6
-
1665
1658
2.
M
E
T
H
O
DO
L
O
G
Y
AND
E
XP
E
RI
M
E
NT
AL
R
E
SU
L
T
S
MA
T
L
AB
i
s
u
s
ed
f
o
r
th
e
im
p
lem
en
ta
tio
n
an
d
tr
ain
i
n
g
o
f
t
h
e
p
r
e
-
tr
ain
e
d
C
NN
m
o
d
el.
I
n
ad
d
itio
n
,
FF
m
p
eg
is
u
s
ed
f
o
r
k
ey
f
r
am
es
ex
tr
ac
tio
n
lik
e
th
e
wo
r
k
in
[
3
4
]
.
FF
m
p
eg
s
u
p
p
o
r
ts
d
if
f
e
r
e
n
t
v
id
eo
an
d
a
u
d
io
f
o
r
m
ats.
Fo
r
C
NN
r
e
-
tr
ain
in
g
,
NVI
DI
A
T
esla K
8
0
is
u
s
ed
wh
ich
h
as 2
4
9
6
C
UDA
co
r
es.
2
.
1
.
M
et
ho
do
lo
g
y
I
n
th
is
wo
r
k
,
th
e
VGG
-
1
6
is
r
e
-
tr
ain
ed
u
s
in
g
a
d
ataset
th
at
co
n
tain
s
4
8
0
0
0
im
ag
es.
T
h
e
VGG
-
1
6
m
o
d
el
was
n
o
t
tr
ain
ed
f
r
o
m
s
cr
atch
f
o
r
tim
e
s
av
in
g
p
u
r
p
o
s
es.
I
n
ad
d
itio
n
,
tr
ain
in
g
th
e
V
GG
-
1
6
m
o
d
el
o
n
a
r
elativ
ely
s
m
all
d
ataset
ca
u
s
es
n
etwo
r
k
o
v
er
f
i
ttin
g
.
Hen
ce
,
a
tr
an
s
f
er
lear
n
in
g
is
u
s
ed
f
o
r
a
VGG
-
1
6
m
o
d
el
th
at
was
tr
ain
ed
o
n
a
v
er
y
lar
g
e
d
ataset
(
I
m
ag
eNe
t,
wh
ich
co
n
tain
s
1
.
2
m
illi
o
n
im
ag
es
with
1
0
0
0
ca
teg
o
r
ies).
T
h
e
last
f
u
lly
-
co
n
n
ec
te
d
lay
er
o
f
th
e
tr
ain
ed
m
o
d
el
was
r
em
o
v
ed
an
d
tailo
r
ed
f
o
r
th
e
1
2
8
ca
teg
o
r
ies
o
f
o
u
r
s
y
s
t
em
.
Als
o
,
ap
p
r
o
p
r
iate
tr
ai
n
in
g
p
ar
am
eter
s
v
alu
es
wer
e
s
elec
ted
.
T
h
e
v
alu
es
wer
e
ch
o
s
en
m
an
u
ally
with
tr
ial
an
d
er
r
o
r
wh
ile
m
o
n
ito
r
in
g
th
e
p
er
f
o
r
m
a
n
ce
an
d
th
e
ac
c
u
r
ac
y
o
f
th
e
VGG
-
1
6
m
o
d
el.
T
h
e
tr
ain
in
g
d
ataset
is
ch
o
s
e
n
ac
co
r
d
in
g
t
o
th
e
p
r
o
b
ab
ilit
y
o
f
ap
p
e
a
r
an
ce
s
in
s
u
r
v
eillan
ce
v
id
e
o
s
.
T
ab
le
1
lis
ts
th
e
ca
teg
o
r
ies th
at
wer
e
s
elec
ted
in
o
u
r
VC
A
s
y
s
tem
.
T
h
ese
o
b
jects a
r
e
o
f
in
te
r
est in
s
u
r
v
eillan
ce
v
id
eo
s
.
Nev
er
th
eless
,
u
s
er
s
ca
n
ad
d
o
r
r
e
m
o
v
e
o
b
jects
as
th
ey
p
r
ef
e
r
an
d
s
till
f
o
llo
w
th
e
m
eth
o
d
o
l
o
g
y
o
f
th
e
pr
o
p
o
s
ed
s
y
s
tem
.
T
h
e
s
y
s
tem
co
n
s
id
er
s
1
2
8
d
if
f
e
r
en
t
o
b
je
cts.
So
m
e
o
f
th
e
s
elec
ted
o
b
j
ec
ts
ar
e
co
n
s
id
er
ed
d
a
n
g
e
r
o
u
s
a
n
d
m
a
y
h
a
r
m
h
u
m
a
n
s
s
u
c
h
a
s
f
i
r
e
,
s
m
o
k
e
,
g
u
n
s
,
r
e
v
o
l
v
e
r
s
,
n
a
i
l
s
,
a
n
d
d
r
i
l
l
t
o
o
l
s
.
I
n
a
d
d
i
t
i
o
n
,
a
n
i
m
a
ls
l
i
k
e
s
c
o
r
p
i
o
n
s
,
s
n
a
k
e
s
,
a
n
d
s
p
i
d
e
r
s
a
r
e
c
o
n
s
i
d
e
r
e
d
.
T
h
e
d
a
t
a
s
e
t
a
l
s
o
h
a
s
n
o
r
m
a
l
o
b
j
e
c
t
s
t
h
a
t
a
r
e
m
a
i
n
l
y
u
s
e
d
b
y
h
u
m
a
n
s
l
i
k
e
c
a
r
s
,
m
o
t
o
r
b
i
k
e
,
b
a
g
s
,
c
a
m
e
r
a
s
,
m
o
b
i
l
e
p
h
o
n
e
s
,
c
o
m
p
u
t
e
r
s
,
A
T
M
s
,
i
r
o
n
,
a
n
d
v
a
c
u
u
m
c
l
e
a
n
e
r
.
S
o
m
e
e
x
a
m
p
l
e
s
o
f
o
b
j
e
c
t
s
t
h
a
t
w
e
r
e
e
x
c
l
u
d
e
d
a
r
e
s
e
a
o
b
j
e
c
t
s
,
c
l
o
t
h
e
s
,
a
n
d
m
u
s
i
c
a
l
i
n
s
t
r
u
m
e
n
t
s
.
A
f
t
e
r
t
h
e
d
a
t
a
s
e
t
h
a
s
b
e
e
n
p
r
e
p
a
r
e
d
,
e
a
c
h
i
m
a
g
e
i
s
l
a
b
e
l
e
d
a
n
d
u
s
e
d
i
n
t
h
e
t
r
a
i
n
i
n
g
o
f
t
h
e
p
r
e
-
tr
ain
ed
VGG
-
1
6
m
o
d
el.
T
ab
le
1
.
T
h
e
s
elec
ted
o
b
ject
c
ateg
o
r
ies in
th
e
d
ataset
T
h
e
c
a
t
e
g
o
r
i
e
s
t
h
a
t
w
e
r
e
se
l
e
c
t
e
d
i
n
o
u
r
V
C
A
s
y
s
t
e
m
P
a
c
k
e
t
A
i
r
p
l
a
n
e
To
a
s
t
e
r
A
l
l
i
g
a
t
o
r
A
mb
u
l
a
n
c
e
A
p
i
a
r
y
,
b
e
e
h
o
u
se
A
p
p
l
e
A
sh
c
a
n
,
g
a
r
b
a
g
e
c
a
n
P
r
i
n
t
e
r
B
a
l
l
V
e
n
d
i
n
g
mac
h
i
n
e
B
a
n
a
n
a
B
a
n
n
i
st
e
r
,
b
a
n
i
s
t
e
r
B
a
r
r
e
l
,
c
a
s
k
B
a
ss
i
n
e
t
B
i
c
y
c
l
e
S
c
h
o
o
l
b
u
s
C
a
b
,
t
a
x
i
,
t
a
x
i
c
a
b
P
o
p
b
o
t
t
l
e
C
a
l
c
u
l
a
t
o
r
C
a
mer
a
C
a
n
o
p
e
n
e
r
C
a
n
d
l
e
,
t
a
p
e
r
,
Car
S
n
a
k
e
C
a
sse
t
t
e
p
l
a
y
e
r
M
o
v
i
n
g
v
a
n
Cat
C
e
l
l
u
l
a
r
p
h
o
n
e
C
h
a
i
n
C
h
a
i
r
C
h
i
l
d
,
k
i
d
Ti
g
e
r
C
o
m
p
u
t
e
r
mo
u
se
G
a
smas
k
C
o
n
f
e
c
t
i
o
n
e
r
y
,
c
a
n
d
y
s
t
o
r
e
C
o
r
k
screw
,
b
o
t
t
l
e
screw
C
r
a
n
e
C
r
a
sh
h
e
l
m
e
t
C
r
u
t
c
h
Tr
a
m,
t
r
o
l
l
e
y
c
a
r
D
r
i
l
l
P
e
o
p
l
e
Ea
g
l
e
El
e
c
t
r
i
c
f
a
n
Es
p
r
e
ss
o
mak
e
r
F
e
n
c
e
,
f
e
n
c
i
n
g
F
i
r
e
S
mo
k
e
F
l
o
w
e
r
p
o
t
R
a
c
i
n
g
c
a
r
F
o
r
k
l
i
f
t
F
o
u
n
t
a
i
n
F
r
y
i
n
g
p
a
n
G
a
r
b
a
g
e
t
r
u
c
k
G
a
r
d
e
n
c
a
r
t
O
r
a
n
g
e
H
a
t
c
h
e
t
S
c
r
e
w
B
a
g
G
r
o
c
e
r
y
st
o
r
e
H
a
i
r
d
r
y
e
r
H
a
mm
e
r
H
a
mste
r
F
i
r
e
t
r
u
c
k
B
o
o
k
S
p
i
d
e
r
B
o
o
k
c
a
s
e
H
e
n
H
o
r
se
i
P
o
d
I
r
o
n
P
e
n
C
a
r
t
o
n
Tr
a
c
t
o
r
C
a
s
h
m
a
c
h
i
n
e
,
A
TM
K
n
i
f
e
La
d
l
e
La
mp
La
p
t
o
p
R
a
b
b
i
t
C
o
m
b
i
n
a
t
i
o
n
l
o
c
k
W
a
t
e
r
b
o
t
t
l
e
C
o
m
p
u
t
e
r
Li
g
h
t
e
r
Li
m
o
u
s
i
n
e
M
a
c
a
w
M
a
i
l
b
o
x
,
l
e
t
t
e
r
b
o
x
S
c
o
r
p
i
o
n
D
i
a
l
t
e
l
e
p
h
o
n
e
S
a
w
D
o
g
M
i
c
r
o
w
a
v
e
o
v
e
n
M
i
l
k
c
a
n
M
o
n
i
t
o
r
M
o
t
o
r
b
i
k
e
S
o
f
a
b
e
d
F
i
sh
M
i
c
r
o
p
h
o
n
e
,
mi
k
e
F
l
o
w
e
r
M
u
g
M
u
s
h
r
o
om
N
a
i
l
N
e
c
k
l
a
c
e
S
t
r
a
w
b
e
r
r
y
Le
mo
n
A
ssau
l
t
r
i
f
l
e
,
C
a
r
t
Jee
p
,
l
a
n
d
r
o
v
e
r
P
e
r
f
u
me,
e
sse
n
c
e
P
i
p
e
,
t
u
b
e
P
l
a
t
e
Ta
b
l
e
V
a
c
u
u
m
c
l
e
a
n
e
r
A
u
t
o
ma
t
i
c
w
a
sh
e
r
B
i
r
d
C
l
o
c
k
R
e
f
r
i
g
e
r
a
t
o
r
,
i
c
e
b
o
x
R
e
m
o
t
e
c
o
n
t
r
o
l
R
e
v
o
l
v
e
r
Tr
a
f
f
i
c
l
i
g
h
t
Tr
a
i
n
,
r
a
i
l
r
o
a
d
t
r
a
i
n
S
t
r
e
e
t
si
g
n
Tu
r
t
l
e
S
y
r
i
n
g
e
S
c
r
e
w
d
r
i
v
e
r
S
h
e
e
p
S
h
o
v
e
l
Ap
p
r
o
p
r
iate
tr
ain
in
g
p
ar
am
et
er
s
ar
e
ass
ig
n
ed
an
d
u
s
ed
d
u
r
in
g
th
e
tr
ain
in
g
o
f
th
e
VGG
-
1
6
m
o
d
el.
T
ab
le
2
s
h
o
ws
th
e
ass
ig
n
ed
v
alu
es
f
o
r
th
e
m
o
s
t
im
p
o
r
ta
n
t
tr
ain
in
g
p
a
r
am
eter
s
o
f
th
e
VGG
-
1
6
m
o
d
el.
T
h
e
a
s
s
i
g
n
e
d
v
a
l
u
e
s
w
e
r
e
t
e
s
t
e
d
a
n
d
v
e
r
i
f
i
e
d
t
o
p
r
o
v
i
d
e
t
h
e
b
e
s
t
r
e
s
u
l
t
s
t
h
r
o
u
g
h
t
r
i
a
l
a
n
d
e
r
r
o
r
.
T
h
e
s
e
p
ar
am
eter
s
ar
e
im
p
o
r
tan
t to
f
it th
e
tr
ain
in
g
im
ag
es in
th
e
m
em
o
r
y
a
n
d
to
a
v
o
id
o
v
er
f
itti
n
g
o
f
t
h
e
C
NN
n
etwo
r
k
.
Th
e
s
cr
ip
t
is
ex
ec
u
ted
3
0
0
ti
m
es
T
ab
le
2
.
At
ea
c
h
iter
atio
n
,
1
6
0
im
ag
es
a
r
e
ch
o
s
en
at
r
an
d
o
m
f
r
o
m
t
h
e
t
r
a
i
n
i
n
g
s
e
t
.
T
h
e
n
e
t
w
o
r
k
l
a
y
e
r
s
a
r
e
t
r
a
i
n
e
d
o
v
e
r
t
h
e
s
e
i
m
a
g
e
s
a
n
d
t
h
e
p
r
e
d
i
c
t
i
o
n
i
s
c
o
m
p
a
r
e
d
a
g
a
i
n
s
t
t
h
e
g
r
o
u
n
d
t
r
u
t
h
.
F
i
g
u
r
e
1
s
h
o
w
s
t
h
e
t
r
a
i
n
i
n
g
r
e
s
u
l
t
s
.
T
h
e
n
e
t
w
o
r
k
r
e
a
c
h
e
s
a
n
a
c
c
u
r
a
c
y
o
f
8
8
.
9
1
%
.
F
i
g
u
r
e
2
a
n
d
F
i
g
u
r
e
3
s
h
o
w
s
d
if
f
er
en
t te
s
tin
g
e
x
am
p
les o
f
t
h
e
tr
ain
ed
n
etwo
r
k
w
h
er
e
th
e
n
etwo
r
k
was a
b
le
to
c
o
r
r
ec
tly
id
en
tify
o
b
jects.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
ng
I
SS
N:
2
0
8
8
-
8
7
0
8
S
ea
r
ch
in
g
s
u
r
ve
illa
n
ce
vid
e
o
co
n
ten
ts
u
s
in
g
co
n
vo
lu
tio
n
a
l n
eu
r
a
l n
etw
o
r
k
(
Du
a
a
Mo
h
a
m
ma
d
)
1659
T
ab
le
2
.
T
r
ai
n
in
g
p
ar
am
eter
s
P
a
r
a
me
t
e
r
s
S
p
e
c
i
f
i
c
a
t
i
o
n
N
u
mb
e
r
o
f
c
l
a
sse
s
1
2
8
d
i
f
f
e
r
e
n
t
c
a
t
e
g
o
r
i
e
s
Tr
a
i
n
i
n
g
i
m
a
g
e
s
4
8
,
0
0
0
V
a
l
i
d
a
t
i
o
n
i
ma
g
e
s
1
4
,
0
0
0
M
i
n
i
b
a
t
c
h
s
i
z
e
1
6
0
N
u
mb
e
r
o
f
i
t
e
r
a
t
i
o
n
s
3
0
0
I
n
i
t
i
a
l
l
e
a
r
n
i
n
g
r
a
t
e
0
.
0
0
0
1
N
u
mb
e
r
o
f
e
p
o
c
h
s
30
Fig
u
r
e
1
.
T
h
e
ac
cu
r
ac
y
o
f
VGG
-
1
6
ag
ain
s
t th
e
n
u
m
b
er
o
f
iter
atio
n
s
Fig
u
r
e
2
.
A
ca
n
o
p
e
n
er
is
d
ete
cted
in
th
e
f
r
am
e
F
ig
u
r
e
3
.
E
x
am
p
les o
f
o
b
j
ec
ts
th
at
wer
e
co
r
r
ec
tly
d
etec
ted
b
y
t
h
e
VGG
-
16
2
.
2
.
Sp
ee
di
ng
up
t
he
VCA
s
y
s
t
e
m
T
o
en
h
a
n
ce
th
e
s
p
ee
d
o
f
th
e
VC
A
s
y
s
tem
,
th
r
ee
tech
n
iq
u
e
s
wer
e
u
s
ed
to
im
p
r
o
v
e
th
e
p
er
f
o
r
m
a
n
ce
.
T
h
e
tech
n
iq
u
es a
r
e
ex
tr
ac
tio
n
k
ey
f
r
a
m
es,
s
o
b
e
l e
d
g
e
d
etec
t
o
r
,
an
d
Ma
x
-
p
o
o
lin
g
as f
o
llo
w
s
.
2
.
2
.
1
.
K
ey
f
ra
mes
I
n
a
v
id
eo
,
th
e
en
tire
in
f
o
r
m
atio
n
ex
is
ts
in
th
e
(
I
)
f
r
am
e,
wh
ich
is
ca
lled
al
s
o
th
e
k
ey
f
r
am
e.
T
h
e
s
et
o
f
k
e
y
f
r
a
m
es
is
s
im
ilar
to
h
a
v
in
g
a
s
h
o
r
t
r
e
p
r
esen
tatio
n
o
f
a
v
id
e
o
.
T
o
in
cr
ea
s
e
th
e
s
p
e
ed
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
,
o
n
ly
k
e
y
f
r
am
es
wer
e
u
s
ed
.
FF
m
p
eg
p
r
o
g
r
am
is
u
s
ed
to
e
x
tr
ac
t
th
e
f
r
am
es.
T
h
e
ex
tr
a
ctio
n
is
p
e
r
f
o
r
m
e
d
f
o
r
d
i
f
f
e
r
e
n
t
f
r
a
m
e
r
a
t
e
s
s
u
c
h
a
s
2
5
,
3
0
,
a
n
d
6
0
f
r
a
m
e
s
p
e
r
s
e
c
o
n
d
.
M
o
r
e
o
v
e
r
,
D
i
f
f
e
r
e
n
t
s
u
r
v
eillan
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
2
,
Ap
r
il 2
0
2
1
:
1
6
5
6
-
1665
1660
v
id
eo
s
f
o
r
in
d
o
o
r
an
d
o
u
td
o
o
r
ac
tiv
ities
ca
n
h
a
v
e
r
e
d
u
n
d
an
c
y
in
th
e
k
ey
f
r
am
es
b
ec
a
u
s
e
o
f
s
m
all
tr
an
s
itio
n
s
.
Similar
ities
in
d
if
f
er
en
t
s
eq
u
en
ce
s
o
f
th
e
k
ey
f
r
am
es
wer
e
f
o
u
n
d
.
Hen
ce
,
th
e
s
o
b
el
d
etec
to
r
is
u
s
ed
f
o
r
en
h
an
ce
m
e
n
t to
g
ain
m
o
r
e
d
is
cr
im
in
atio
n
o
f
th
e
f
r
am
es.
2
.
2
.
2
.
So
bel e
dg
e
det
ec
t
o
r
T
h
e
s
ec
o
n
d
s
te
p
in
s
p
ee
d
in
g
u
p
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
is
u
s
in
g
th
e
s
o
b
el
o
p
er
ato
r
ed
g
e
d
etec
to
r
to
d
eter
m
in
e
m
o
s
t
u
s
ef
u
l
r
ep
r
esen
tatio
n
o
f
th
e
k
e
y
f
r
am
es.
So
b
el
o
p
er
ato
r
d
etec
ts
th
e
ed
g
es
i
n
th
e
ex
tr
ac
ted
k
e
y
f
r
am
es,
wh
ich
is
u
s
ed
later
to
f
in
d
m
atch
i
n
g
s
b
etwe
en
c
o
n
s
ec
u
tiv
e
f
r
am
es.
I
f
th
e
s
eq
u
e
n
ce
s
o
f
f
r
am
e
s
h
a
v
e
m
atch
in
g
s
ac
co
r
d
in
g
t
o
a
th
r
e
s
h
o
ld
,
th
en
th
e
s
y
s
tem
elim
in
ates
th
e
s
ec
o
n
d
f
r
am
e
b
ec
a
u
s
e
th
er
e
is
n
o
n
ee
d
to
class
if
y
it
b
y
VGG
-
1
6
n
etwo
r
k
.
So
b
el
o
p
er
ato
r
h
as
a
h
ig
h
a
cc
u
r
ac
y
in
d
etec
tin
g
e
d
g
es
o
f
im
ag
es.
I
t
u
s
es
tw
o
3×
3
k
er
n
els.
T
h
e
f
ir
s
t o
n
e
is
f
o
r
th
e
h
o
r
izo
n
tal
an
d
th
e
s
ec
o
n
d
o
n
e
is
f
o
r
th
e
v
er
t
ical
d
if
f
er
en
ce
s
[
3
5
]
.
Fig
u
r
e
4
s
h
o
ws
s
o
b
el
ed
g
e
d
etec
to
r
f
ilter
s
.
Fig
u
r
e
4
.
So
b
el
ed
g
e
d
etec
to
r
f
ilter
s
T
h
e
an
d
f
ilter
s
ar
e
c
o
n
v
o
lv
e
d
with
th
e
im
ag
e
a
n
d
r
esu
lt
i
n
a
g
r
ad
ien
t
o
f
a
m
ag
n
itu
d
e
th
at
is
co
m
p
u
ted
u
s
in
g
(
1
)
.
T
h
e
g
r
ad
ien
t
d
ir
ec
tio
n
is
co
m
p
u
ted
u
s
in
g
(
2
)
[
3
5
]
.
T
h
e
s
o
b
el
ed
g
e
d
etec
to
r
is
ap
p
lied
to
th
e
ex
tr
ac
ted
k
ey
f
r
am
es
a
n
d
t
h
e
ed
g
es
a
r
e
th
e
n
co
m
p
u
ted
.
T
h
e
o
u
t
p
u
t
im
ag
e
h
as
th
e
s
am
e
s
ize
as
t
h
e
o
r
ig
in
al
o
n
e
(
i.e
.
2
2
4
×
2
2
4
)
.
=
√
2
+
2
(
1
)
Ɵ
=
(
2
)
2
.
2
.
3
.
M
a
x
-
p
o
o
lin
g
Af
ter
ap
p
ly
in
g
th
e
s
o
b
el
ed
g
e
d
etec
to
r
,
a
3
×3
m
ax
-
p
o
o
lin
g
s
lid
in
g
win
d
o
w
is
ap
p
lied
t
o
f
in
d
th
e
m
ax
im
u
m
n
u
m
b
er
o
f
t
h
e
ar
e
a.
T
h
e
e
x
tr
ac
ted
f
ea
tu
r
es
a
r
e
co
m
p
a
r
ed
to
m
ak
e
th
e
m
atc
h
in
g
d
ec
is
io
n
m
o
r
e
ac
cu
r
ate.
Similar
f
r
am
es
ar
e
ig
n
o
r
e
d
.
T
h
is
m
eth
o
d
s
p
ee
d
s
u
p
th
e
s
y
s
tem
s
in
ce
th
e
an
aly
s
is
an
d
class
if
icatio
n
ar
e
p
er
f
o
r
m
ed
o
n
a
r
ed
u
ce
d
n
u
m
b
er
o
f
k
ey
f
r
am
es.
Ma
x
-
p
o
o
lin
g
s
lid
in
g
w
in
d
o
w
o
f
s
ize
3
×3
an
d
a
f
r
am
e
s
ize
o
f
2
2
4
×2
2
4
was
u
s
ed
.
T
h
e
m
at
ch
in
g
d
ec
is
io
n
is
b
ased
o
n
t
h
e
n
u
m
b
er
o
f
m
atch
in
g
p
ix
els
r
elativ
e
to
th
e
to
tal
n
u
m
b
er
o
f
p
ix
els as sh
o
wn
in
(
3
)
.
ℎ
=
(
ℎ
/
)
100%
(
3
)
A
th
r
esh
o
ld
o
f
0
.
7
is
u
s
ed
f
o
r
th
e
m
atch
in
g
p
e
r
ce
n
tag
e.
I
f
th
e
to
tal
m
atch
p
er
ce
n
tag
e
is
less
th
an
0
.
7
,
th
en
th
e
f
r
am
e
is
d
if
f
er
en
t
f
r
o
m
th
e
n
ex
t
f
r
am
e
an
d
n
ee
d
s
to
b
e
an
aly
ze
d
.
T
a
b
le
3
s
h
o
ws
d
if
f
er
en
t
th
r
esh
o
l
d
v
alu
es
th
at
wer
e
test
ed
.
A
v
al
u
e
o
f
0
.
7
was
f
o
u
n
d
to
b
e
r
ela
tiv
ely
b
etter
in
p
r
o
d
u
cin
g
ac
c
u
r
ate
r
esu
lts
f
o
r
t
h
e
k
e
y
f
r
a
m
e
s
.
T
o
v
a
l
i
d
a
t
e
t
h
e
s
e
l
e
c
t
e
d
t
h
r
e
s
h
o
l
d
v
a
l
u
e
,
O
n
e
L
e
a
v
e
S
h
o
p
2
c
o
r
v
i
d
e
o
f
r
o
m
C
A
V
I
A
R
d
a
t
a
s
e
t
[
3
6
]
is
u
s
ed
.
T
h
e
v
i
d
eo
co
n
tain
s
6
3
k
ey
f
r
am
es.
H
o
wev
er
,
th
e
r
e
a
r
e
s
im
ilar
ities
b
etwe
en
d
if
f
er
en
t
f
r
am
es.
T
a
b
l
e
4
s
h
o
w
s
s
a
m
p
l
e
i
m
a
g
e
s
u
s
i
n
g
t
h
e
t
h
r
e
s
h
o
l
d
v
a
l
u
e
o
f
0
.
7
.
W
h
e
n
u
s
i
n
g
t
h
e
v
a
l
u
e
s
0
.
9
o
r
1
.
0
f
o
r
t
h
e
th
r
esh
o
ld
,
th
e
s
im
ilar
it
y
b
ec
o
m
es less
th
an
th
e
th
r
esh
o
ld
an
d
all
o
f
th
e
f
r
a
m
es will b
e
s
elec
ted
an
d
class
i
f
ied
.
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
elim
in
ati
n
g
r
e
d
u
n
d
an
t
f
r
am
es
is
ap
p
lie
d
to
a
n
o
u
td
o
o
r
s
u
r
v
eillan
ce
v
id
eo
with
a
len
g
th
o
f
6
8
s
ec
o
n
d
s
.
On
e
ex
t
r
ac
ted
f
r
am
e
tak
es
0
.
1
3
5
3
5
9
s
ec
o
n
d
s
to
b
e
class
if
ied
b
y
th
e
VGG
-
1
6
n
etwo
r
k
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
tak
es
1
6
.
8
0
4
3
0
4
s
ec
o
n
d
s
to
an
aly
ze
th
e
v
id
eo
with
o
u
t
th
e
a
p
p
lic
atio
n
o
f
th
e
s
o
b
e
l
d
etec
to
r
an
d
m
a
x
-
p
o
o
lin
g
m
eth
o
d
s
.
On
th
e
o
th
er
h
an
d
,
i
t
tak
es
1
0
.
5
0
4
6
8
8
s
ec
o
n
d
s
wh
en
th
e
en
h
a
n
ce
d
m
eth
o
d
s
ar
e
u
s
ed
.
T
h
is
s
h
o
w
s
h
o
w
th
e
s
o
b
el
d
etec
to
r
a
n
d
m
ax
-
p
o
o
lin
g
en
h
a
n
ce
th
e
s
y
s
tem
s
p
ee
d
with
o
u
t
alter
in
g
ac
cu
r
ac
y
.
T
h
e
d
if
f
er
e
n
t
s
tep
s
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
ar
e
s
u
m
m
ar
ized
as sh
o
wn
i
n
Fig
u
r
e
5.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
ng
I
SS
N:
2
0
8
8
-
8
7
0
8
S
ea
r
ch
in
g
s
u
r
ve
illa
n
ce
vid
e
o
co
n
ten
ts
u
s
in
g
co
n
vo
lu
tio
n
a
l n
eu
r
a
l n
etw
o
r
k
(
Du
a
a
Mo
h
a
m
ma
d
)
1661
T
ab
le
3
.
T
h
r
esh
o
ld
test
in
g
Th
r
e
s
h
o
l
d
#
o
f
K
e
y
F
r
a
m
e
s
#
o
f
E
l
i
m
i
n
a
t
e
d
K
e
y
F
r
a
m
e
s
#
o
f
A
n
a
l
y
z
e
d
F
r
a
mes
0
.
1
63
63
0
0
.
2
63
63
0
0
.
3
63
60
3
0
.
4
63
55
8
0
.
5
63
53
10
0
.
6
63
52
11
0
.
7
63
50
13
0
.
8
63
8
55
0
.
9
63
0
63
1
63
0
63
T
ab
le
4
.
T
esti
n
g
th
e
th
r
esh
o
ld
o
.
7
u
s
in
g
On
eL
ea
v
eSh
o
p
2
co
r
v
id
eo
0
.
7
Fig
u
r
e
5
.
T
h
e
f
r
a
m
ewo
r
k
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
2
.
3
T
esting
a
nd
ev
a
lua
t
io
n
T
h
e
v
id
e
o
c
o
n
ten
t
a
n
aly
s
is
s
y
s
tem
is
test
ed
o
n
a
v
ar
iety
o
f
s
u
r
v
eillan
ce
v
id
eo
s
in
cl
u
d
in
g
d
if
f
e
r
en
t
o
b
jects,
m
o
v
em
en
ts
,
an
d
ev
e
n
ts
.
T
h
e
s
elec
ted
v
id
eo
s
ar
e
tak
en
f
r
o
m
th
e
im
ag
e
v
i
d
eo
s
y
s
tem
lab
(
I
VY)
d
ataset
[
3
7
]
an
d
v
i
d
eo
im
ag
e
r
etr
iev
al
an
aly
s
is
to
o
l
(
VI
R
AT
)
v
id
e
o
d
ataset
[
3
8
]
.
T
h
e
e
x
p
e
r
im
en
tal
r
esu
lts
ar
e
s
h
o
wn
in
T
ab
le
5
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
is
test
ed
u
s
in
g
n
in
e
v
id
eo
s
.
T
h
ese
v
id
e
o
s
ar
e
d
if
f
er
en
t
f
r
o
m
ea
ch
o
th
er
i
n
tim
e
,
f
r
a
m
e
r
ate
,
an
d
th
e
n
u
m
b
er
o
f
k
e
y
f
r
am
es.
T
h
e
ex
ec
u
tio
n
tim
e
in
clu
d
es
th
e
l
o
ad
in
g
o
f
t
h
e
tr
ain
ed
n
etwo
r
k
,
ex
tr
ac
tin
g
f
r
am
es,
ap
p
ly
in
g
s
o
b
el
d
etec
to
r
,
ap
p
ly
in
g
m
ax
-
p
o
o
lin
g
,
th
e
m
atch
in
g
s
tag
e,
an
d
s
av
in
g
o
f
th
e
r
esu
lts
in
to
a
te
x
t
f
ile.
T
h
e
ex
p
e
r
im
en
ts
we
r
e
p
er
f
o
r
m
ed
u
s
in
g
I
n
tel
i7
PC
an
d
T
esla
K8
0
GPU
ca
r
d
.
T
h
e
VGG
-
1
6
n
etwo
r
k
r
eq
u
ir
es
1
1
m
s
o
n
av
er
a
g
e
t
o
class
if
y
o
n
e
f
r
am
e.
Vid
eo
s
n
u
m
b
er
s
7
,
8
,
a
n
d
9
h
av
e
o
r
ig
in
ally
5
1
9
,
6
1
2
,
an
d
3
1
k
ey
f
r
am
es,
r
esp
ec
tiv
ely
.
Ho
wev
er
,
u
s
in
g
o
u
r
m
ec
h
an
is
m
s
o
f
m
atch
in
g
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
2
,
Ap
r
il 2
0
2
1
:
1
6
5
6
-
1665
1662
i
g
n
o
r
in
g
s
im
ilar
f
r
a
m
es,
th
e
p
r
o
p
o
s
ed
s
y
s
tem
class
if
ied
2
9
,
1
1
,
an
d
1
2
f
r
am
es,
r
esp
ec
t
iv
ely
.
I
n
s
p
ec
tin
g
th
e
v
id
eo
s
r
ev
e
als
th
at
th
ey
wer
e
r
ec
o
r
d
e
d
b
y
f
i
x
ed
ca
m
er
as
an
d
th
e
s
ce
n
e
o
f
th
e
s
ite
h
as
litt
le
in
f
o
r
m
atio
n
with
s
m
all
tr
an
s
itio
n
s
.
Hen
ce
,
th
e
p
r
o
p
o
s
ed
s
y
s
tem
is
u
s
ef
u
l f
o
r
s
ea
r
ch
in
g
v
id
eo
c
o
n
ten
ts
.
T
ab
le
5
.
E
x
p
er
im
en
tal
r
esu
lts
f
o
r
n
in
e
v
id
e
o
s
I
n
a
d
d
itio
n
,
u
s
in
g
T
esla
K8
0
s
p
ee
d
s
u
p
th
e
s
y
s
tem
1
2
X
th
a
n
u
s
in
g
C
PU
o
n
ly
.
T
h
e
o
u
tp
u
t
te
x
t
f
ile
f
o
r
v
i
d
e
o
2
i
s
l
i
s
t
e
d
b
e
l
o
w
.
V
i
d
e
o
2
c
o
n
t
a
i
n
s
5
6
k
e
y
f
r
a
m
e
s
(
I
)
t
h
a
t
r
e
p
r
e
s
e
n
t
t
w
o
c
h
i
l
d
r
e
n
w
a
l
k
i
n
g
i
n
th
e
h
allway
.
T
h
e
f
r
am
es th
at
wer
e
an
aly
ze
d
an
d
class
if
ied
wer
e
1
2
a
s
a
r
esu
lt
o
f
ap
p
ly
in
g
th
e
s
o
b
e
l f
ilter
an
d
m
ax
-
p
o
o
lin
g
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
t
em
was
ab
le
to
an
aly
ze
th
e
f
r
am
es
in
0
.
7
1
s
e
co
n
d
s
u
s
in
g
th
e
GPU
an
d
7
.
8
0
s
ec
o
n
d
s
u
s
in
g
th
e
C
PU.
A
s
s
ee
n
f
r
o
m
th
e
tex
t
f
ile
o
f
v
id
eo
2
,
th
er
e
was
n
o
u
s
e
f
u
l
in
f
o
r
m
atio
n
b
ef
o
r
e
tim
e
1
7
.
2
0
s
ec
o
n
d
s
.
T
h
e
n
,
th
e
ch
ild
r
en
ap
p
ea
r
in
th
e
v
id
e
o
.
T
h
is
is
an
ef
f
icien
t w
ay
to
f
in
d
o
b
jects o
f
in
ter
est v
id
eo
s
in
s
tead
o
f
s
ea
r
ch
in
g
m
an
u
ally
an
d
s
p
en
d
in
g
a
lo
t
o
f
tim
e.
A
u
s
er
ca
n
s
ea
r
ch
with
in
th
e
tex
t
f
ile
to
f
in
d
a
tim
e
o
f
in
ter
est
an
d
th
en
f
ast
f
o
r
wa
r
d
th
e
v
id
eo
(
e.
g
.
to
tim
e
1
7
.
2
s
ec
s
)
.
T
h
e
tex
t
f
ile
f
o
r
v
id
e
o
9
(
AT
M
m
ac
h
i
n
e)
is
s
h
o
wn
b
elo
w
.
Vid
eo
9
co
n
tain
s
3
1
o
r
ig
in
al
k
ey
f
r
am
es
(
I
)
th
at
r
ep
r
esen
t
th
e
ar
ea
o
f
th
e
AT
M
m
ac
h
in
e.
T
h
e
class
if
ied
f
r
am
es
wer
e
o
n
ly
1
2
f
r
am
es.
Ag
ain
,
o
u
r
s
y
s
tem
was
ab
le
to
an
aly
ze
th
e
v
id
eo
v
er
y
f
a
s
t
(
0
.
7
2
s
e
c
o
n
d
s
o
n
G
P
U
a
n
d
9
.
2
5
s
e
c
o
n
d
s
o
n
C
P
U
)
w
h
e
n
c
o
m
p
a
r
e
d
t
o
t
h
e
l
e
n
g
t
h
o
f
t
h
e
v
i
d
e
o
(
5
9
s
e
c
o
n
d
s
)
.
C
o
n
ten
ts
o
f
th
e
T
ex
t File
f
o
r
Vid
eo
2
D:
\
H
A
LLWA
Y
_
A
-
0
1
.
j
p
e
g
0
0
.
0
0
0
0
0
0
A
s
h
c
a
n
,
g
a
r
b
a
g
e
c
a
n
,
s
c
o
r
e
0
.
7
9
D:
\
H
A
LLWA
Y
_
A
-
4
4
.
j
p
e
g
5
1
6
1
7
.
2
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
sc
o
r
e
0
.
7
9
D:
\
H
A
LLWA
Y
_
A
-
4
5
.
j
p
e
g
5
2
8
1
7
.
6
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
sc
o
r
e
0
.
7
1
D:
\
H
A
LLWA
Y
_
A
-
4
6
.
j
p
e
g
5
4
0
1
8
.
0
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
sc
o
r
e
0
.
4
2
D:
\
H
A
LLWA
Y
_
A
-
4
7
.
j
p
e
g
5
5
2
1
8
.
4
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
s
c
o
r
e
0
.
6
3
D:
\
H
A
LLWA
Y
_
A
-
4
8
.
j
p
e
g
5
6
4
1
8
.
8
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
s
c
o
r
e
0
.
6
6
D:
\
H
A
LLWA
Y
_
A
-
4
9
.
j
p
e
g
5
7
6
1
9
.
2
0
0
0
0
0
C
h
i
l
d
,
k
i
d
sc
o
r
e
0
.
4
4
D:
\
H
A
LLWA
Y
_
A
-
5
0
.
j
p
e
g
5
8
8
1
9
.
6
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
sc
o
r
e
0
.
6
5
D:
\
H
A
LLWA
Y
_
A
-
5
1
.
j
p
e
g
6
0
0
2
0
.
0
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
sc
o
r
e
0
.
8
0
D:
\
H
A
LLWA
Y
_
A
-
5
2
.
j
p
e
g
6
1
2
2
0
.
4
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
s
c
o
r
e
0
.
5
9
D:
\
H
A
LLWA
Y
_
A
-
5
3
.
j
p
e
g
6
2
4
2
0
.
8
0
0
0
0
0
C
h
i
l
d
,
k
i
d
,
s
c
o
r
e
0
.
8
6
D:
\
H
A
LLWA
Y
_
A
-
5
4
.
j
p
e
g
6
3
6
2
1
.
2
0
0
0
0
0
C
a
s
h
m
a
c
h
i
n
e
,
A
T
M
,
sc
o
r
e
0
.
9
6
C
o
n
ten
ts
o
f
th
e
T
ex
t File
f
o
r
Vid
eo
9
D:
\
A
T
M
1
0
5
.
j
p
e
g
2
4
0
8
.
0
0
8
0
0
0
F
l
o
w
e
r
p
o
t
,
sc
o
r
e
0
.
7
9
D:
\
A
T
M
1
0
6
.
j
p
e
g
3
0
0
1
0
.
0
1
0
0
0
0
F
l
o
w
e
r
p
o
t
,
sc
o
r
e
0
.
8
1
D:
\
A
T
M
1
1
0
.
j
p
e
g
5
2
4
1
7
.
4
8
4
1
3
3
P
e
o
p
l
e
,
sc
o
r
e
0
.
8
0
D:
\
A
T
M
1
1
2
.
j
p
e
g
6
4
4
2
1
.
4
8
8
1
3
3
C
h
i
l
d
,
k
i
d
,
s
c
o
r
e
0
.
8
4
D:
\
A
T
M
1
1
3
.
j
p
e
g
7
0
4
2
3
.
4
9
0
1
3
3
C
h
i
l
d
,
k
i
d
,
sc
o
r
e
0
.
8
1
D:
\
A
T
M
1
1
7
.
j
p
e
g
9
2
8
3
0
.
9
6
4
2
6
7
C
h
i
l
d
,
k
i
d
,
s
c
o
r
e
0
.
6
8
D:
\
A
T
M
1
1
8
.
j
p
e
g
9
8
8
3
2
.
9
6
6
2
6
7
B
a
n
n
i
s
t
e
r
,
b
a
n
i
st
e
r
sc
o
r
e
0
.
5
4
D:
\
A
T
M
1
2
1
.
j
p
e
g
1
1
6
8
3
8
.
9
7
2
2
6
7
C
h
i
l
d
,
k
i
d
,
sc
o
r
e
0
.
6
8
D:
\
A
T
M
1
2
2
.
j
p
e
g
1
2
2
8
4
0
.
9
7
4
2
6
7
P
e
o
p
l
e
,
s
c
o
r
e
0
.
5
4
D:
\
A
T
M
1
2
5
.
j
p
e
g
1
3
9
2
4
6
.
4
4
6
4
0
0
P
e
o
p
l
e
,
sc
o
r
e
0
.
6
6
D:
\
A
T
M
1
2
9
.
j
p
e
g
1
6
3
2
5
4
.
4
5
4
4
0
0
B
a
n
n
i
st
e
r
,
b
a
n
i
s
t
e
r
sc
o
r
e
0
.
8
D:
\
A
T
M
1
3
0
.
j
p
e
g
1
6
9
2
5
6
.
4
5
6
4
0
0
B
a
n
n
i
st
e
r
,
b
a
n
i
s
t
e
r
,
sc
o
r
e
1
.
0
0
V
i
d
e
o
#
Le
n
g
t
h
(
S
e
c
)
To
t
a
l
N
o
.
o
f
f
r
a
mes
N
o
.
o
f
K
e
y
f
r
a
mes
C
l
a
s
si
f
i
e
d
f
r
a
mes
F
r
a
me
r
a
t
e
(
F
r
a
me/
S
e
c
)
C
P
U
E
x
e
c
u
t
i
o
n
Ti
me
(
S
e
c
)
G
P
U
Ex
e
c
u
t
i
o
n
Ti
me
(
S
e
c
)
V
i
d
e
o
1
(A
TM
)
29
7
2
5
61
15
25
8
0
.
6
6
6
V
i
d
e
o
2
(
H
a
l
l
w
a
y
)
22
6
6
0
56
12
30
7
.
8
0
0
.
7
1
V
i
d
e
o
3
(
S
t
a
i
r
)
24
7
2
0
59
6
30
7
.
7
0
0
.
6
2
5
V
i
d
e
o
4
(
j
u
n
c
t
i
o
n
)
2
6
5
3
6
6
3
2
5
5
5
2
7
1
2
9
25
1
1
7
7
9
9
.
6
8
V
i
d
e
o
5
(
p
a
r
k
i
n
g
l
o
t
)
6
8
9
1
9
9
8
1
83
58
29
9
2
.
4
1
5
1
1
.
0
8
9
V
i
d
e
o
6
(
o
u
t
d
o
o
r
)
3
0
2
8
7
5
8
37
18
29
4
3
.
0
7
1
3
.
8
7
2
V
i
d
e
o
7
(
p
a
r
k
i
n
g
l
o
t
)
5
0
5
1
5
1
5
0
5
1
9
29
30
5
5
.
70
3
4
.
7
5
9
V
i
d
e
o
8
(
p
a
r
k
i
n
g
l
o
t
)
2
4
4
7
0
7
6
6
1
2
11
29
1
0
0
8
.
6
5
V
i
d
e
o
9
(
A
TM
)
59
1
7
1
1
31
12
29
9
.
2
5
0
.
7
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
ng
I
SS
N:
2
0
8
8
-
8
7
0
8
S
ea
r
ch
in
g
s
u
r
ve
illa
n
ce
vid
e
o
co
n
ten
ts
u
s
in
g
co
n
vo
lu
tio
n
a
l n
eu
r
a
l n
etw
o
r
k
(
Du
a
a
Mo
h
a
m
ma
d
)
1663
2
.
4
.
Co
m
pa
riso
n wit
h t
he
s
t
a
t
e
o
f
t
he
a
rt
T
ab
le
6
s
h
o
ws
a
co
m
p
ar
is
o
n
b
etwe
en
o
u
r
s
y
s
tem
an
d
th
r
ee
s
y
s
tem
s
th
at
wer
e
p
r
o
p
o
s
ed
in
th
e
liter
atu
r
e.
L
ee
et
a
l
.
p
r
o
p
o
s
ed
a
v
id
eo
co
n
ten
t
m
eth
o
d
u
s
in
g
a
r
eg
r
ess
io
n
m
o
d
el
t
h
at
d
etec
t
s
im
p
o
r
tan
t
o
b
jects
an
d
o
r
g
a
n
izes
th
em
as
a
s
eq
u
en
ce
o
f
im
ag
es
[
3
9
]
.
T
h
e
s
y
s
tem
in
[
3
9
]
n
ee
d
s
1
s
ec
o
n
d
f
o
r
ea
ch
f
r
am
e
to
b
e
an
aly
ze
d
with
a
n
ac
c
u
r
ac
y
o
f
6
8
.
7
5
%.
Me
g
h
d
ad
i
a
n
d
I
r
an
i
in
tr
o
d
u
ce
d
a
n
an
al
y
s
is
s
y
s
te
m
th
at
p
r
o
ce
s
s
es
3
f
r
am
es
p
er
s
ec
o
n
d
with
an
a
cc
u
r
ac
y
o
f
a
b
o
u
t
6
9
%
[
4
0
]
.
T
h
e
s
y
s
tem
in
[
4
0
]
wo
r
k
s
b
y
ex
tr
ac
tin
g
m
u
ltip
le
fr
am
es
b
ased
o
n
m
o
tio
n
a
n
d
t
h
en
v
is
u
alizin
g
tr
ajec
to
r
ies
o
f
th
e
o
b
jects
in
s
u
r
v
eillan
ce
s
tr
ea
m
s
.
Z
h
o
u
et
a
l
.
p
r
o
p
o
s
ed
an
ap
p
r
o
ac
h
f
o
r
v
i
d
e
o
an
aly
s
is
u
s
in
g
a
s
p
ac
e
-
tim
e
s
alien
cy
m
eth
o
d
an
d
a
f
aster
r
e
-
tim
in
g
m
eth
o
d
to
tack
le
lo
n
g
v
id
e
o
in
d
e
x
in
g
a
n
d
s
u
m
m
ar
izat
io
n
[
3
0
]
.
T
h
e
s
y
s
tem
in
[
3
0
]
r
e
q
u
ir
es
1
1
s
ec
o
n
d
s
to
p
r
o
ce
s
s
o
n
e
f
r
am
e.
T
ab
le
6
s
h
o
ws th
at
th
e
p
r
o
p
o
s
ed
s
y
s
tem
o
u
tp
er
f
o
r
m
s
th
e
r
elate
d
wo
r
k
with
an
o
r
d
er
o
f
m
ag
n
itu
d
e.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
p
r
o
ce
s
s
es
6
f
r
am
es
p
er
s
ec
o
n
d
.
T
h
e
n
u
m
b
e
r
o
f
f
r
am
es
o
f
t
h
e
m
atch
in
g
p
r
o
ce
s
s
ca
n
r
ea
ch
u
p
to
4
0
f
r
am
es p
er
s
ec
o
n
d
.
T
ab
le
6
.
C
o
m
p
u
tatio
n
al
tim
e
c
o
m
p
ar
is
o
n
R
e
l
a
t
e
d
w
o
r
k
C
o
m
p
u
t
a
t
i
o
n
a
l
t
i
m
e
Le
e
e
t
a
l
.
[
3
9
]
1
F
r
a
m
e
p
e
r
s
e
c
o
n
d
M
e
g
h
d
a
d
i
a
n
d
I
r
a
n
i
[
4
0
]
3
f
r
a
m
e
s
p
e
r
s
e
c
o
n
d
Zh
o
u
e
t
a
l
.
[
3
0
]
1
f
r
a
m
e
p
e
r
1
1
se
c
o
n
d
Th
e
p
r
o
p
o
s
e
d
S
S
V
C
s
y
st
e
m
6
f
r
a
m
e
s
p
e
r
s
e
c
o
n
d
3.
CO
NCLU
SI
O
N
A
n
ew
s
y
s
tem
f
o
r
v
id
eo
co
n
ten
t
an
aly
s
is
h
as
b
ee
n
p
r
o
p
o
s
ed
.
T
h
e
s
y
s
tem
u
s
es
VGG
-
1
6
d
ee
p
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
etwo
r
k
f
o
r
o
b
ject
d
etec
tio
n
a
n
d
id
e
n
tific
atio
n
.
T
h
e
n
u
m
b
er
o
f
d
if
f
er
en
t
class
es
o
f
o
b
jects
th
at
wer
e
co
n
s
id
er
ed
f
o
r
s
ec
u
r
ity
an
d
s
u
r
v
eillan
ce
s
y
s
tem
s
i
s
1
2
8
.
Ho
wev
er
,
u
s
e
r
s
ca
n
d
ef
in
e
m
o
r
e
class
es
o
f
th
eir
in
te
r
est.
I
n
ad
d
itio
n
,
th
r
ee
tech
n
iq
u
es
wer
e
u
s
ed
to
r
ed
u
ce
th
e
am
o
u
n
t
o
f
in
f
o
r
m
atio
n
th
at
n
ee
d
s
to
b
e
p
r
o
ce
s
s
ed
b
y
th
e
VGG
-
1
6
n
etwo
r
k
.
T
h
e
th
r
ee
tech
n
iq
u
es
ar
e
ex
tr
ac
tio
n
o
f
k
ey
f
r
am
es
u
s
in
g
FF
m
p
eg
,
s
o
b
e
l
ed
g
e
d
etec
to
r
,
an
d
Ma
x
-
p
o
o
lin
g
.
T
h
e
s
o
b
el
d
etec
to
r
an
d
m
ax
-
p
o
o
lin
g
ar
e
u
s
ed
in
o
r
d
er
to
f
u
r
th
er
r
e
d
u
ce
th
e
n
u
m
b
er
o
f
f
r
am
es th
at
ar
e
n
ee
d
ed
f
o
r
p
r
o
c
ess
in
g
.
T
h
e
o
u
tp
u
t o
f
th
e
s
y
s
tem
is
a
tex
t f
ile
th
at
is
r
ea
d
ab
le
b
y
h
u
m
a
n
s
s
o
th
ey
ca
n
s
ea
r
ch
in
it
f
o
r
th
eir
o
b
j
ec
ts
o
f
in
ter
est.
T
h
e
tex
t
f
ile
co
n
tain
s
th
e
o
b
je
cts
an
d
m
o
v
em
en
ts
th
at
o
cc
u
r
i
n
a
v
id
eo
.
I
n
s
tead
o
f
m
a
n
u
al
v
id
eo
in
s
p
ec
tio
n
f
o
r
lo
n
g
p
e
r
io
d
s
,
th
e
p
r
esen
ted
s
y
s
tem
p
r
o
v
id
ed
u
s
er
s
with
an
ea
s
y
an
d
s
im
p
le
way
f
o
r
v
id
eo
co
n
ten
t
s
ea
r
ch
in
g
.
T
h
e
tim
e
to
p
r
o
d
u
ce
th
e
tex
t
f
ile
is
n
eg
lig
ib
le
co
m
p
ar
ed
with
th
e
s
iz
e
an
d
tim
e
o
f
th
e
a
n
aly
ze
d
v
id
eo
.
T
h
e
r
esu
lts
h
av
e
s
h
o
wed
th
at
th
e
p
r
o
p
o
s
ed
s
y
s
tem
o
u
tp
er
f
o
r
m
e
d
ex
is
tin
g
m
eth
o
d
s
in
an
o
r
d
e
r
o
f
m
ag
n
itu
d
e.
RE
F
E
R
E
NC
E
S
[1
]
Aljarra
h
I
.
a
n
d
M
o
h
a
m
m
a
d
D
.
,
"
Vi
d
e
o
c
o
n
ten
t
a
n
a
ly
sis
u
si
n
g
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
s,"
2
0
1
8
9
t
h
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
I
n
fo
r
ma
ti
o
n
a
n
d
Co
mm
u
n
ica
ti
o
n
S
y
ste
ms
(ICICS
)
,
Irb
id
,
2
0
1
8
,
pp
.
1
2
2
-
126
.
[2
]
S
.
F
a
k
h
a
r
A.
G
.
,
e
t
a
l.
,
"
De
v
e
lo
p
m
e
n
t
o
f
a
p
o
rtab
le
c
o
m
m
u
n
it
y
v
i
d
e
o
su
r
v
e
il
lan
c
e
s
y
ste
m
,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
ter
En
g
in
e
e
rin
g
(IJ
ECE
)
,
v
o
l.
9
,
n
o
.
3
,
p
p
.
1
8
1
4
-
1
8
2
1
,
2
0
1
9
.
[3
]
Kra
u
sz
B
.
a
n
d
He
rp
e
rs
R
.
,
"
M
e
tro
S
u
r
v
:
d
e
tec
ti
n
g
e
v
e
n
ts
i
n
su
b
wa
y
sta
ti
o
n
s,
"
M
u
lt
ime
d
i
a
T
o
o
ls
a
n
d
Ap
p
li
c
a
ti
o
n
s
,
v
o
l.
5
0
,
n
o
.
1
,
p
p
.
1
2
3
-
1
4
7
,
2
0
1
0
.
[4
]
P
ra
k
o
so
,
P
.
B.
,
a
n
d
S
a
ri,
Y.
,
"
Ve
h
icle
d
e
tec
ti
o
n
u
sin
g
b
a
c
k
g
ro
u
n
d
su
b
trac
ti
o
n
a
n
d
c
lu
ste
ri
n
g
a
lg
o
ri
th
m
s,
"
T
EL
KOM
NIKA
(
T
e
lec
o
mm
u
n
ic
a
t
io
n
,
Co
mp
u
ti
n
g
,
El
e
c
tro
n
ics
a
n
d
Co
n
tro
l),
v
o
l
.
1
7
,
n
o
.
3
,
p
p
.
1
3
9
3
-
1
3
9
8
,
2
0
1
9
.
[5
]
P
ra
b
o
w
o
,
M
.
R.
,
Hu
d
a
y
a
n
i
,
N.,
P
u
rwiy
a
n
t
i,
S
.
,
S
u
l
isti
y
a
n
ti
,
S
.
R
.
,
S
e
ty
a
wa
n
,
F
.
A.
,
"
A
m
o
v
i
n
g
o
b
je
c
ts
d
e
tec
ti
o
n
in
u
n
d
e
rwa
ter
v
i
d
e
o
u
si
n
g
s
u
b
trac
ti
o
n
o
f
th
e
b
a
c
k
g
ro
u
n
d
m
o
d
e
l,
"
i
n
2
0
1
7
4
th
I
n
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
El
e
c
trica
l
En
g
i
n
e
e
rin
g
,
Co
m
p
u
ter
S
c
ie
n
c
e
a
n
d
I
n
fo
rm
a
ti
c
s (E
EC
S
I)
,
Yo
g
y
a
k
a
rta
,
2
0
1
7
,
p
p
.
1
-
4.
[6
]
Kriz
h
e
v
sk
y
A
.
,
e
t
a
l.
,
"
Im
a
g
e
N
e
t
c
las
sifica
ti
o
n
with
d
e
e
p
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
r
k
s,
"
Ad
v
a
n
c
e
s
in
n
e
u
ra
l
in
fo
rm
a
ti
o
n
p
ro
c
e
ss
in
g
sy
ste
ms
,
p
p
.
1
0
9
7
-
1
1
0
5
,
2
0
1
2
.
[7
]
As
il
,
H.,
a
n
d
Ba
g
h
e
rz
a
d
e
h
,
J.,
"
P
r
o
p
o
si
n
g
a
n
e
w
m
e
th
o
d
o
f
ima
g
e
c
las
sifica
ti
o
n
b
a
se
d
o
n
t
h
e
Ad
a
Bo
o
st
d
e
e
p
b
e
li
e
f
n
e
two
rk
h
y
b
ri
d
m
e
th
o
d
,
"
T
EL
K
OM
NIKA
(T
e
lec
o
mm
u
n
ica
ti
o
n
,
Co
mp
u
t
in
g
,
El
e
c
tro
n
ics
a
n
d
C
o
n
tro
l)
,
v
o
l.
1
7
,
n
o
.
5
,
p
p
.
2
6
5
0
-
2
6
5
8
,
2
0
1
9
.
[8
]
Erh
a
n
D
.
,
e
t
a
l.
,
"
S
c
a
lab
le
o
b
jec
t
d
e
tec
ti
o
n
u
si
n
g
d
e
e
p
n
e
u
ra
l
n
e
two
rk
s,"
2
0
1
4
IEE
E
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
Vi
sio
n
a
n
d
P
a
t
ter
n
Rec
o
g
n
it
i
o
n
,
Co
lu
m
b
u
s,
OH
,
2
0
1
4
,
p
p
.
2
1
4
7
-
2
1
5
4
.
[9
]
Yu
a
n
,
Y.,
M
o
u
,
L
.
a
n
d
L
u
,
X.,
"
S
c
e
n
e
re
c
o
g
n
it
i
o
n
b
y
m
a
n
if
o
l
d
re
g
u
lariz
e
d
d
e
e
p
lea
rn
i
n
g
a
rc
h
it
e
c
tu
re
,
"
IE
EE
tra
n
sa
c
ti
o
n
s
o
n
n
e
u
r
a
l
n
e
two
rk
s a
n
d
lea
rn
in
g
sy
ste
ms
,
v
o
l
.
2
6
,
n
o
.
1
0
,
p
p.
2
2
2
2
-
2
2
3
3
,
2
0
1
5
.
[1
0
]
Co
ll
o
b
e
rt
R
a
n
d
Wes
to
n
J
.
,
"
A
u
n
ifi
e
d
a
rc
h
it
e
c
tu
re
fo
r
n
a
tu
ra
l
l
a
n
g
u
a
g
e
p
ro
c
e
ss
in
g
:
De
e
p
n
e
u
ra
l
n
e
two
rk
s
wi
t
h
m
u
lt
it
a
sk
lea
rn
i
n
g
,
"
T
h
e
2
5
t
h
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
M
a
c
h
in
e
lea
rn
in
g
,
2
0
0
8
,
p
p
.
1
6
0
-
1
6
7
.
[1
1
]
S
a
n
to
s,
C.
N.
D.,
Xia
n
g
,
B.
,
a
n
d
Zh
o
u
,
B
.
,
"
Clas
sify
in
g
re
lati
o
n
s
b
y
ra
n
k
i
n
g
with
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
s,
"
Pro
c
e
e
d
in
g
s
o
f
th
e
5
3
rd
An
n
u
a
l
M
e
e
ti
n
g
o
f
t
h
e
Asso
c
ia
ti
o
n
f
o
r
C
o
mp
u
ta
ti
o
n
a
l
L
i
n
g
u
isti
c
s a
n
d
t
h
e
7
th
I
n
ter
n
a
t
io
n
a
l
J
o
in
t
C
o
n
fer
e
n
c
e
o
n
N
a
tu
ra
l
L
a
n
g
u
a
g
e
Pro
c
e
ss
in
g
,
v
o
l
.
1
,
2
0
15
,
p
p
.
6
2
6
-
6
3
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
2
,
Ap
r
il 2
0
2
1
:
1
6
5
6
-
1665
1664
[1
2
]
Zeiler
M
.
D
.
a
n
d
F
e
rg
u
s
R
.
,
"
V
isu
a
li
z
in
g
a
n
d
u
n
d
e
rsta
n
d
i
n
g
c
o
n
v
o
lu
ti
o
n
a
l
n
e
two
r
k
s,"
Eu
ro
p
e
a
n
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
ter
v
isio
n
,
v
o
l
.
8
6
8
9
,
2
0
1
4
,
pp
.
8
1
8
-
8
3
3
.
[1
3
]
S
z
e
g
e
d
y
C
.
,
e
t
a
l.
,
"
G
o
i
n
g
d
e
e
p
e
r
with
c
o
n
v
o
lu
t
io
n
s,"
2
0
1
5
I
EE
E
Co
n
fer
e
n
c
e
o
n
C
o
mp
u
t
e
r
Vi
si
o
n
a
n
d
P
a
tt
e
rn
Rec
o
g
n
it
io
n
(CVP
R)
,
Bo
st
o
n
,
M
A
,
2
0
1
5
,
p
p
.
1
-
9.
[1
4
]
S
imo
n
y
a
n
K
.
a
n
d
Zi
ss
e
rm
a
n
A
.
,
"
Ve
ry
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
two
rk
s
f
o
r
lar
ge
-
sc
a
le
ima
g
e
re
c
o
g
n
i
ti
o
n
,
"
a
rXiv:1
4
0
9
.
1
5
5
6
,
2
0
1
4
.
[1
5
]
He
K
.
,
e
t
a
l.
,
"
De
e
p
re
sid
u
a
l
lea
rn
in
g
f
o
r
ima
g
e
re
c
o
g
n
i
ti
o
n
,
"
2
0
1
6
I
EE
E
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
t
e
r
Vi
sio
n
a
n
d
Pa
tt
e
rn
Rec
o
g
n
it
io
n
(CVP
R)
,
Las
Ve
g
a
s,
NV
,
2
0
1
6
,
p
p
.
7
7
0
-
7
7
8
.
[1
6
]
Li
H
.
,
e
t
a
l.
,
"
A
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
c
a
sc
a
d
e
fo
r
f
a
c
e
d
e
tec
ti
o
n
,
"
2
0
1
5
I
EE
E
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
ter
Vi
sio
n
a
n
d
P
a
tt
e
rn
Rec
o
g
n
it
i
o
n
(
CVP
R)
,
B
o
s
to
n
,
M
A
,
2
0
1
5
,
p
p
.
5
3
2
5
-
5
3
3
4
.
[1
7
]
S
u
n
Y
.
,
e
t
a
l.
,
"
De
e
p
i
d
3
:
F
a
c
e
re
c
o
g
n
it
io
n
wit
h
v
e
ry
d
e
e
p
n
e
u
ra
l
n
e
two
rk
s,"
a
rXiv:
1
5
0
2
.
0
0
8
7
3
,
2
0
1
5
.
[1
8
]
Ba
sh
e
e
r,
N.
M
.
,
a
n
d
M
o
h
a
m
m
e
d
,
M
.
H.
,
"
Clas
sifica
ti
o
n
o
f
b
re
a
st
m
a
ss
e
s
in
d
ig
it
a
l
m
a
m
m
o
g
ra
m
s
u
sin
g
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
s,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Ad
v
a
n
c
e
d
Res
e
a
rc
h
i
n
Co
mp
u
ter
S
c
ien
c
e
a
n
d
S
o
ft
wa
re
En
g
i
n
e
e
rin
g
,
v
o
l.
3
,
n
o
.
1
0
,
p
p
.
5
7
-
6
3
,
2
0
1
3
.
[1
9
]
S
h
a
d
e
e
d
,
G
.
A.,
Taw
f
e
e
q
,
M
.
A.,
M
a
h
m
o
u
d
,
S
.
M
.
,
"
De
e
p
l
e
a
rn
in
g
m
o
d
e
l
f
o
r
t
h
o
ra
x
d
is
e
a
se
s
d
e
tec
ti
o
n
,
"
T
EL
KOM
NIKA
(
T
e
lec
o
mm
u
n
ic
a
t
io
n
,
Co
mp
u
ti
n
g
,
El
e
c
tro
n
ics
a
n
d
Co
n
tro
l)
,
v
o
l
.
1
8
,
n
o
.
1
,
p
p
.
4
4
1
-
4
4
9
,
2
0
2
0
.
[2
0
]
Zh
a
n
g
W,
e
t
a
l.
,
"
C
o
m
p
u
teriz
e
d
d
e
tec
ti
o
n
o
f
c
l
u
ste
re
d
m
icro
c
a
lci
fica
ti
o
n
s
i
n
d
ig
it
a
l
m
a
m
m
o
g
ra
m
s
u
sin
g
a
s
h
ift
‐
in
v
a
rian
t
a
rti
ficia
l
n
e
u
ra
l
n
e
two
r
k
,
"
M
e
d
ica
l
Ph
y
sic
s
,
v
o
l.
2
1
,
n
o
.
4
,
p
p
.
5
1
7
-
5
2
4
,
1
9
9
4
.
[2
1
]
Lao
W
.
,
e
t
a
l.
,
"
Au
t
o
m
a
ti
c
v
i
d
e
o
-
b
a
se
d
h
u
m
a
n
m
o
ti
o
n
a
n
a
ly
z
e
r
fo
r
c
o
n
su
m
e
r
su
rv
e
il
la
n
c
e
sy
ste
m
,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Co
n
su
me
r E
lec
tro
n
ics
,
v
o
l
.
5
5
,
n
o
.
2
,
p
p
.
5
9
1
-
5
9
8
,
2
0
0
9
.
[2
2
]
Zh
a
o
Z
.
C
.
a
n
d
Ca
i
A
.
N
.
,
"
Ex
tra
c
ti
o
n
o
f
se
m
a
n
ti
c
k
e
y
fra
m
e
s
b
a
se
d
o
n
v
isu
a
l
a
tt
e
n
t
io
n
a
n
d
a
ffe
c
ti
v
e
m
o
d
e
ls,"
2
0
0
7
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
t
a
ti
o
n
a
l
In
telli
g
e
n
c
e
a
n
d
S
e
c
u
ri
ty
(CIS
2
0
0
7
)
,
Ha
rb
in
,
p
p
.
3
7
1
-
3
7
5
,
2
0
0
7
.
[2
3
]
Be
rti
n
i
M
.
,
e
t
a
l.
,
"
An
in
te
g
ra
t
e
d
fra
m
e
wo
rk
f
o
r
se
m
a
n
ti
c
a
n
n
o
tatio
n
a
n
d
a
d
a
p
tati
o
n
,
"
M
u
lt
im
e
d
ia
T
o
o
ls
a
n
d
Ap
p
li
c
a
ti
o
n
s
,
v
o
l
.
2
6
,
n
o
.
3
,
p
p
.
3
4
5
-
3
6
3
,
2
0
0
5
.
[2
4
]
Ko
lek
a
r
M
.
H
.,
"
Ba
y
e
sia
n
b
e
li
e
f
n
e
two
r
k
b
a
se
d
b
ro
a
d
c
a
st
s
p
o
rts
v
id
e
o
i
n
d
e
x
in
g
,
"
M
u
lt
ime
d
ia
T
o
o
ls
a
n
d
Ap
p
li
c
a
ti
o
n
s
,
v
o
l
.
5
4
,
n
o
.
1
,
p
p
.
27
-
5
4
,
2
0
1
1
.
[2
5
]
Ch
e
n
W
.
E
.,
a
n
d
Zh
a
n
g
Y
.
J
.
,
"
Vid
e
o
se
g
m
e
n
tat
io
n
a
n
d
k
e
y
fra
m
e
e
x
trac
ti
o
n
wit
h
p
a
ra
m
e
tri
c
m
o
d
e
l,
"
2
0
0
8
3
r
d
In
ter
n
a
t
io
n
a
l
S
y
mp
o
si
u
m o
n
C
o
m
mu
n
ica
t
io
n
s,
Co
n
tro
l
a
n
d
S
i
g
n
a
l
P
ro
c
e
ss
in
g
,
S
t
Ju
l
ian
s
,
2
0
0
8
,
p
p
.
1
0
2
0
-
1
0
2
3
.
[2
6
]
S
u
n
Z
.
,
e
t
a
l.
,
"
Vid
e
o
k
e
y
fra
m
e
e
x
trac
ti
o
n
b
a
se
d
o
n
sp
a
ti
a
l
-
tem
p
o
ra
l
c
o
l
o
r
d
istri
b
u
ti
o
n
,
"
2
0
0
8
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
In
tell
ig
e
n
t
In
f
o
rm
a
ti
o
n
Hi
d
in
g
a
n
d
M
u
lt
i
me
d
i
a
S
ig
n
a
l
Pro
c
e
ss
in
g
,
Ha
rb
i
n
,
2
0
0
8
,
p
p
.
1
9
6
-
1
9
9
.
[2
7
]
S
h
a
rif,
M
.
H.,
a
n
d
Dje
ra
b
a
,
C.
,
"
An
e
n
tro
p
y
a
p
p
r
o
a
c
h
fo
r
a
b
n
o
rm
a
l
a
c
ti
v
it
ies
d
e
tec
ti
o
n
in
v
id
e
o
str
e
a
m
s,"
Pa
tt
e
rn
re
c
o
g
n
it
i
o
n
,
v
o
l.
4
5
,
n
o
.
7
,
p
p
.
2
5
4
3
-
2
5
6
1
,
2
0
1
2
.
[2
8
]
Ce
rn
e
k
o
v
a
Z
.
,
e
t
a
l.
,
"
I
n
fo
rm
a
ti
o
n
t
h
e
o
ry
-
b
a
se
d
s
h
o
t
c
u
t/
fa
d
e
d
e
tec
ti
o
n
a
n
d
v
i
d
e
o
su
m
m
a
r
iza
ti
o
n
,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Circ
u
it
s
a
n
d
S
y
st
e
ms
fo
r V
id
e
o
T
e
c
h
n
o
lo
g
y
,
v
o
l.
1
6
,
n
o
.
1
,
p
p
.
82
-
9
1
,
2
0
0
6
.
[2
9
]
Zen
g
W
.
,
e
t
a
l.
,
"
Ro
b
u
st
m
o
v
in
g
o
b
jec
t
se
g
m
e
n
tati
o
n
o
n
H.
2
6
4
/AVC
c
o
m
p
re
ss
e
d
v
id
e
o
u
si
n
g
th
e
b
l
o
c
k
-
b
a
se
d
M
RF
m
o
d
e
l
,
"
Re
a
l
-
T
ime
Ima
g
i
n
g
,
v
o
l.
1
1
,
n
o
.
4
,
p
p
.
2
9
0
-
29
9
,
2
0
0
5
.
[3
0
]
Zh
o
u
F
.
,
e
t
a
l.
,
"
Ti
m
e
-
m
a
p
p
i
n
g
u
sin
g
s
p
a
c
e
-
ti
m
e
sa
li
e
n
c
y
,
"
2
0
1
4
IEE
E
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
Vi
sio
n
a
n
d
Pa
tt
e
rn
Rec
o
g
n
it
io
n
,
C
o
lu
m
b
u
s,
OH
,
2
0
1
4
,
pp
.
3
3
5
8
-
3
3
6
5
.
[3
1
]
Ba
i
L
.
,
e
t
a
l.
,
"
Vid
e
o
se
m
a
n
ti
c
c
o
n
ten
t
a
n
a
l
y
sis
b
a
se
d
o
n
o
n
to
l
o
g
y
,
"
I
n
ter
n
a
t
io
n
a
l
M
a
c
h
i
n
e
Vi
s
io
n
a
n
d
Ima
g
e
Pro
c
e
ss
in
g
Co
n
fer
e
n
c
e
(IM
VIP
2
0
0
7
)
,
Kil
d
a
re
,
2
0
0
7
,
p
p
.
1
1
7
-
1
2
4
.
[3
2
]
F
o
g
g
ia
P
.
,
e
t
a
l.
,
"
Re
a
l
-
ti
m
e
fire
d
e
tec
ti
o
n
fo
r
v
i
d
e
o
-
s
u
rv
e
il
lan
c
e
a
p
p
li
c
a
ti
o
n
s
u
sin
g
a
c
o
m
b
i
n
a
ti
o
n
o
f
e
x
p
e
rts
b
a
se
d
o
n
c
o
lo
r
,
s
h
a
p
e
,
a
n
d
m
o
ti
o
n
,
"
IE
EE
T
r
a
n
sa
c
ti
o
n
s
o
n
Circ
u
it
s
a
n
d
S
y
ste
ms
fo
r
Vi
d
e
o
T
e
c
h
n
o
lo
g
y
,
v
o
l.
2
5
,
n
o
.
9
,
pp.
15
45
-
15
5
6
,
2
0
1
5
.
[3
3
]
Be
g
e
ja
L
.
,
e
t
a
l.
,
"
Vi
d
c
a
t:
a
n
i
m
a
g
e
a
n
d
v
id
e
o
a
n
a
ly
sis
se
rv
ic
e
fo
r
p
e
rso
n
a
l
m
e
d
ia
m
a
n
a
g
e
m
e
n
t,
"
M
u
l
ti
me
d
i
a
Co
n
ten
t
a
n
d
M
o
b
il
e
De
v
ice
s
, v
o
l.
8
6
6
7
,
2
0
1
3
.
[3
4
]
“
IM
AG
ENET
,
”
2
0
2
0
.
[On
li
n
e
].
Av
a
il
a
b
le:
h
tt
p
:/
/www
.
ima
g
e
-
n
e
t.
o
rg
/ch
a
ll
e
n
g
e
s/LS
VRC/
2
0
1
4
/res
u
lt
s.
[
3
5
]
K
i
t
t
l
e
r
J
.
,
"
O
n
t
h
e
a
c
c
u
r
a
c
y
o
f
t
h
e
S
o
b
e
l
e
d
g
e
d
e
t
e
c
t
o
r
,
"
I
m
a
g
e
a
n
d
V
i
s
i
o
n
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
,
n
o
.
1
,
p
p
.
37
-
4
2
,
1
9
8
3
.
[3
6
]
“
CAV
IAR p
ro
jec
t,
”
2
0
2
0
.
[On
l
in
e
].
Av
a
il
a
b
le:
h
tt
p
:/
/g
r
o
u
p
s.i
n
f.
e
d
.
a
c
.
u
k
/v
isi
o
n
/CAVIA
R/CAVIA
RDA
T
A1
.
[3
7
]
“
IM
AG
E
a
n
d
Vid
e
o
S
YS
TE
M
S
LAB,
”
2
0
2
0
.
[On
li
n
e
].
Av
a
il
a
b
le:
h
tt
p
:/
/i
v
y
la
b
.
k
a
ist.
a
c
.
k
r/d
e
fa
u
l
t
/.
[3
8
]
“
VIRA
T
v
id
e
o
d
a
tas
e
t,
”
2
0
2
0
.
[O
n
li
n
e
].
Av
a
il
a
b
le:
h
tt
p
:/
/www
.
v
iratd
a
ta.o
r
g
/.
[3
9
]
Lee
Y
.
J
.
,
e
t
a
l.
,
"
Disc
o
v
e
ri
n
g
i
m
p
o
rtan
t
p
e
o
p
le
a
n
d
o
b
jec
ts
fo
r
e
g
o
c
e
n
tri
c
v
i
d
e
o
s
u
m
m
a
riza
ti
o
n
,
"
2
0
1
2
IEE
E
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
ter
Vi
si
o
n
a
n
d
Pa
tt
e
rn
Rec
o
g
n
it
io
n
,
P
r
o
v
id
e
n
c
e
,
RI
,
2
0
1
2
,
p
p
1
3
4
6
-
1
3
5
3
.
[4
0
]
M
e
g
h
d
a
d
i
A
.
H
.
a
n
d
I
r
a
n
i
P
.
,
"
I
n
t
e
r
a
c
t
i
v
e
e
x
p
l
o
r
a
t
i
o
n
o
f
s
u
r
v
e
i
l
l
a
n
c
e
v
i
d
e
o
t
h
r
o
u
g
h
a
c
t
i
o
n
s
h
o
t
s
u
m
m
a
r
i
z
a
t
i
o
n
a
n
d
t
r
a
j
e
c
t
o
r
y
v
i
s
u
a
l
i
z
a
t
i
o
n
,
"
I
E
E
E
T
r
a
n
s
a
c
t
i
o
n
s
o
n
V
i
s
u
a
l
i
z
a
t
i
o
n
a
n
d
C
o
m
p
u
t
e
r
G
r
a
p
h
i
c
s
,
v
o
l
.
1
9
,
n
o
.
1
2
,
p
p
.
2119
-
21
2
8
,
2
0
1
3
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Dua
a
Mo
h
a
m
m
a
d
re
c
e
iv
e
d
h
e
r
B.
S
.
d
e
g
re
e
fro
m
th
e
De
p
a
rtm
e
n
t
o
f
Co
m
p
u
ter
a
n
d
Ne
two
r
k
En
g
i
n
e
e
rin
g
a
t
Al
-
Ba
lq
a
Ap
p
li
e
d
Un
iv
e
rsit
y
(BAU
)
in
2
0
1
4
.
S
h
e
is
c
u
rre
n
tl
y
a
M
a
ste
r
stu
d
e
n
t
a
t
th
e
De
p
a
rtme
n
t
o
f
C
o
m
p
u
ter
En
g
in
e
e
rin
g
a
t
Jo
r
d
a
n
Un
i
v
e
rsity
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
lo
g
y
.
S
h
e
is
a
lso
wo
rk
in
g
a
s
a
se
c
u
rit
y
n
e
tw
o
rk
e
n
g
i
n
e
e
r
a
t
th
e
Lan
d
Tran
s
p
o
rt
Re
g
u
lat
o
ry
Co
m
m
issio
n
in
Am
m
a
n
,
Jo
rd
a
n
.
He
r
re
se
a
rc
h
in
t
e
re
sts in
c
lu
d
e
a
rti
ficia
l
i
n
telli
g
e
n
c
e
a
n
d
m
a
c
h
in
e
lea
rn
i
n
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
ng
I
SS
N:
2
0
8
8
-
8
7
0
8
S
ea
r
ch
in
g
s
u
r
ve
illa
n
ce
vid
e
o
co
n
ten
ts
u
s
in
g
co
n
vo
lu
tio
n
a
l n
eu
r
a
l n
etw
o
r
k
(
Du
a
a
Mo
h
a
m
ma
d
)
1665
Ina
d
Alj
a
r
r
a
h
is
a
n
As
so
c
iate
P
ro
fe
ss
o
r
a
t
th
e
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter
En
g
in
e
e
ri
n
g
a
t
Jo
r
d
a
n
Un
iv
e
rsity
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
lo
g
y
.
He
re
c
e
iv
e
d
h
is
B.
S
.
i
n
El
e
c
tri
c
a
l
En
g
i
n
e
e
rin
g
fro
m
Jo
r
d
a
n
Un
iv
e
rsity
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
l
o
g
y
i
n
1
9
9
9
.
He
re
c
e
iv
e
d
h
is
M
a
ste
rs
a
n
d
P
h
.
D.
d
e
g
re
e
s
in
El
e
c
tri
c
a
l
a
n
d
Co
m
p
u
ter
En
g
in
e
e
rin
g
fro
m
Oh
i
o
U
n
iv
e
rsit
y
,
Ath
e
n
s,
O
h
io
,
USA
in
t
h
e
y
e
a
r
s
2
0
0
2
a
n
d
2
0
0
6
r
e
s
p
e
c
t
i
v
e
l
y
.
H
i
s
r
e
s
e
a
r
c
h
i
n
t
e
r
e
s
t
s
a
r
e
c
o
m
p
u
t
e
r
v
i
s
i
o
n
,
i
m
a
g
e
p
r
o
c
e
s
s
i
n
g
a
n
d
a
n
a
l
y
s
i
s
,
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
t
s
y
s
t
e
m
s
.
C
u
r
r
e
n
t
l
y
h
e
i
s
a
v
i
s
i
t
i
n
g
p
r
o
f
e
s
s
o
r
a
t
t
h
e
El
e
c
tri
c
a
l
a
n
d
Co
m
p
u
ter
e
n
g
in
e
e
rin
g
d
e
p
a
rtme
n
t
a
t
N
o
rth
Ca
ro
li
n
a
S
tate
U
n
iv
e
rsit
y
Ra
leig
h
,
NC.
Mo
a
t
h
J
a
r
r
a
h
is
a
n
As
so
c
iate
P
ro
fe
ss
o
r
a
t
th
e
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter
En
g
in
e
e
rin
g
a
t
Jo
r
d
a
n
Un
iv
e
rsity
o
f
S
c
ien
c
e
a
n
d
Tec
h
n
o
lo
g
y
(JU
S
T)
.
He
h
a
s
re
c
e
iv
e
d
h
is M
a
ste
r
a
n
d
P
h
.
D.
d
e
g
re
e
s
fro
m
th
e
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
a
n
d
Co
m
p
u
ter
En
g
i
n
e
e
rin
g
a
t
t
h
e
Un
iv
e
rsity
o
f
Ariz
o
n
a
,
T
u
c
so
n
,
USA,
in
2
0
0
5
a
n
d
2
0
0
8
,
re
sp
e
c
ti
v
e
ly
.
He
re
c
e
iv
e
d
th
e
B.
S
d
e
g
re
e
fro
m
th
e
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
a
n
d
Co
m
p
u
ter
E
n
g
i
n
e
e
rin
g
a
t
JU
S
T
in
2
0
0
2
.
Ja
rra
h
wo
rk
e
d
a
re
se
a
rc
h
fe
ll
o
w
a
t
Ro
ll
s
-
Ro
y
c
e
c
o
r
p
o
ra
te
lab
a
t
Na
n
y
a
n
g
Tec
h
n
o
lo
g
ica
l
U
n
iv
e
rsit
y
(NTU
),
S
i
n
g
a
p
o
re
;
a
n
d
a
s
a
sc
h
o
lar
re
se
a
rc
h
e
r
a
t
th
e
De
p
a
rtme
n
t
o
f
Cre
a
ti
v
e
IT
En
g
i
n
e
e
rin
g
a
n
d
a
t
P
OST
ECH
Un
i
v
e
rsity
,
S
o
u
t
h
Ko
re
a
in
th
e
p
e
rio
d
o
f
2
0
1
4
t
o
2
0
1
6
.
He
h
a
s
c
o
-
a
u
t
h
o
re
d
a
n
d
p
u
b
li
sh
e
d
m
a
n
y
p
a
p
e
rs
in
t
o
p
-
t
ier
jo
u
r
n
a
ls
a
n
d
c
o
n
fe
re
n
c
e
s.
His
re
se
a
rc
h
in
tere
st
s
in
c
lu
d
e
m
o
d
e
li
n
g
a
n
d
sim
u
lat
io
n
,
d
istri
b
u
te
d
sy
ste
m
s,
c
lo
u
d
c
o
m
p
u
ti
n
g
,
sm
a
rt
g
rid
,
m
a
c
h
in
e
lea
rn
i
n
g
,
a
n
d
o
p
ti
m
iza
ti
o
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.