I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
22
,
No
.
3
,
J
u
n
e
202
1
,
p
p
.
1548
~
1
5
5
5
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ee
cs.v
2
2
.i
3
.
pp
1
5
4
8
-
1
5
5
5
1548
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Pedestr
ia
n ag
e
est
i
m
a
tion
ba
sed o
n
deep learning
Na
w
a
l Y
o
un
is
Abdu
lla
h
,
M
o
ha
mm
ed
T
a
la
l
G
ha
za
l,
Na
j
wa
n Wa
is
i
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter E
n
g
in
e
e
rin
g
T
e
c
h
n
o
lo
g
y
,
No
rth
e
rn
T
e
c
h
n
ica
l
Un
iv
e
rsit
y
,
M
o
su
l,
Ira
q
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
2
4
,
2
0
2
1
R
ev
i
s
ed
Ma
r
2
3
,
2
0
2
1
A
cc
ep
ted
Ma
r
2
9
,
2
0
2
1
T
h
e
larg
e
-
s
c
a
le
d
istri
b
u
ti
o
n
o
f
c
a
m
e
ra
n
e
tw
o
rk
s
in
th
e
traf
f
ic
a
re
a
re
su
lt
e
d
i
n
th
e
in
c
re
a
sin
g
p
o
p
u
larity
o
f
v
i
d
e
o
su
rv
e
il
lan
c
e
sy
ste
m
s.
A
s
p
e
d
e
strian
d
e
tec
ti
o
n
a
n
d
trac
k
in
g
a
re
th
e
c
rit
ica
l
m
o
n
it
o
rin
g
targ
e
ts
in
traf
f
i
c
su
rv
e
il
lan
c
e
,
m
a
n
y
stu
d
ies
f
o
c
u
s
o
n
p
e
d
e
strian
d
e
tec
ti
o
n
a
lg
o
ri
th
m
s
a
c
ro
ss
c
a
m
e
ra
s.
T
h
is p
a
p
e
r
a
d
d
re
ss
e
d
th
e
e
ff
e
c
t
o
f
u
sin
g
th
e
a
g
e
e
sti
m
a
ti
o
n
b
a
se
d
o
n
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
n
e
u
ra
l
n
e
tw
o
rk
(CNN
)
a
s
a
c
o
n
v
e
n
ien
c
e
f
o
r
p
e
d
e
strian
m
o
n
it
o
rin
g
w
h
o
is
c
ro
ss
in
g
a
t
i
n
t
e
rse
c
ti
o
n
s.
Tw
o
p
o
p
u
lar
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
(DCN
Ns
)
p
re
-
train
e
d
m
o
d
e
ls
h
a
v
e
b
e
e
n
u
se
d
in
th
is
w
o
rk
,
w
h
ich
h
a
v
e
re
c
e
n
tl
y
a
c
h
ie
v
e
d
th
e
b
e
st
p
e
rf
o
rm
a
n
c
e
in
f
a
c
i
a
l
f
e
a
tu
re
s
e
x
trac
ti
o
n
tas
k
s:
V
G
G
-
F
a
c
e
a
n
d
Re
sN
e
t
-
5
0
.
W
e
c
o
m
b
in
e
d
th
e
se
t
w
o
m
o
d
e
ls
to
in
c
re
a
se
th
e
e
f
f
icie
n
c
y
o
f
th
e
p
r
o
p
o
se
d
sy
ste
m
.
W
e
ra
n
o
u
r
e
x
p
e
rim
e
n
ts
to
e
v
a
lu
a
te
th
e
s
y
ste
m
b
a
se
d
o
n
t
h
e
V
G
G
F
a
c
e
2
d
a
tas
e
t
c
o
n
sisti
n
g
o
f
3
.
3
1
m
il
li
o
n
f
a
c
e
i
m
a
g
e
s.
F
ro
m
th
e
e
x
p
e
rim
e
n
tal
re
su
lt
s,
w
e
o
b
se
rv
e
d
a
g
a
p
in
t
h
e
d
e
tec
ti
o
n
p
e
rf
o
rm
a
n
c
e
s
b
e
t
w
e
e
n
th
o
se
a
g
e
g
ro
u
p
s:
c
h
i
ld
re
n
f
ro
m
(0
0
-
10
)
y
e
a
rs
a
n
d
e
ld
e
rly
w
it
h
5
5
y
e
a
r
s
a
n
d
m
o
re
.
M
o
re
o
v
e
r,
it
n
o
te
d
th
a
t
t
h
e
p
ro
p
o
se
d
p
e
d
e
strian
a
g
e
e
sti
m
a
ti
o
n
m
o
d
e
l
p
e
rf
o
rm
a
n
c
e
is
h
ig
h
,
a
lso
a
g
o
o
d
re
su
lt
c
a
n
b
e
o
b
tai
n
e
d
b
y
u
sin
g
th
e
m
o
d
e
l
f
o
r
n
e
w
p
u
rp
o
se
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
e
u
r
al
n
e
t
w
o
r
k
s
Dee
p
lear
n
in
g
P
ed
estrian
Re
sN
e
t
-
50
VGG
-
Face
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
h
a
m
m
ed
T
alal
Gh
az
al
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
T
ec
h
n
o
lo
g
y
No
r
th
er
n
T
ec
h
n
ical
U
n
i
v
er
s
it
y
Mo
s
u
l,
I
r
aq
E
m
ail:
m
o
h
a
m
m
ed
.
g
h
az
al
@
n
t
u
.
ed
u
.
iq
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
W
o
r
ld
Hea
lth
Or
g
a
n
izati
o
n
(
W
HO)
r
ep
o
r
ted
th
e
s
tati
s
t
ics
f
o
r
th
e
n
u
m
b
er
o
f
r
o
ad
tr
af
f
ic
d
ea
th
s
in
2
0
1
8
,
th
e
r
ate
o
f
d
ea
th
r
ea
ch
ed
1
.
3
5
m
i
llio
n
i
n
2
0
1
6
w
it
h
an
a
v
er
ag
e
r
ate
o
f
2
7
.
5
d
ea
th
s
p
er
1
0
0
,
0
0
0
p
o
p
u
latio
n
as
s
h
o
w
n
in
Fi
g
u
r
e
1
.
T
h
e
r
ep
o
r
t
r
ec
o
m
m
e
n
d
ed
th
e
n
ee
d
to
im
p
r
o
v
e
th
e
r
o
ad
s
in
f
r
astr
u
ct
u
r
e,
w
h
ic
h
r
ep
r
esen
ts
th
e
m
ain
c
au
s
e
o
f
th
e
f
a
tal
in
j
u
r
y
i
n
r
o
ad
tr
af
f
ic
co
llis
io
n
s
[
1
]
.
P
e
d
estrian
m
o
n
ito
r
in
g
s
y
s
te
m
s
p
r
o
v
id
e
q
u
a
n
titati
v
e
in
s
i
g
h
ts
in
to
t
h
e
p
ed
estrian
tr
af
f
ic
s
tate,
al
th
o
u
g
h
t
h
e
y
ar
e
ea
s
y
to
i
n
s
ta
ll
an
d
g
o
o
d
u
n
d
er
n
o
r
m
al
co
n
d
itio
n
s
,
b
u
t
th
e
y
ar
e
n
o
t
1
0
0
%
ac
cu
r
ate
[
2
]
.
T
h
is
w
o
r
k
f
o
cu
s
es
o
n
ag
e
esti
m
atio
n
f
o
r
s
p
ec
if
ic
ag
e
g
r
o
u
p
s
,
s
i
n
ce
eld
er
l
y
p
eo
p
le
f
all
d
u
r
in
g
ac
r
o
s
s
th
e
s
tr
ee
t
is
a
m
o
n
g
th
e
m
o
s
t
d
a
m
ag
in
g
ev
e
n
t
an
d
th
e
m
o
s
t
co
m
m
o
n
o
cc
u
r
r
en
ce
,
b
esid
e
th
e
ch
ild
r
en
b
et
w
ee
n
t
h
o
s
e
t
w
o
p
h
a
s
es,
ag
e
s
4
th
r
o
u
g
h
1
0
.
T
h
e
m
aj
o
r
ity
o
f
p
ed
estrian
i
n
j
u
r
ies
h
ap
p
en
in
m
id
-
b
lo
ck
cr
o
s
s
w
al
k
s
,
wh
er
e
y
o
u
n
g
ch
ild
r
en
tr
y
to
cr
o
s
s
a
s
tr
ee
t
w
it
h
o
u
t
ch
ec
k
i
n
g
f
o
r
tr
af
f
ic,
o
r
at
in
ter
s
ec
tio
n
s
[
3
]
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
i
s
a
p
ar
t
o
f
th
e
co
m
p
u
ter
v
is
io
n
f
ield
,
w
h
ich
p
la
y
s
a
s
ig
n
i
f
ica
n
t
r
o
le
in
f
ac
e
r
ec
o
g
n
itio
n
ap
p
licatio
n
s
[
4
]
.
B
r
ief
l
y
,
th
e
co
m
p
u
ter
v
i
s
io
n
al
g
o
r
ith
m
s
ta
s
k
is
t
o
e
x
tr
ac
t
th
e
f
ac
ial
f
ea
t
u
r
es
f
r
o
m
i
m
a
g
es
a
n
d
co
m
p
ar
e
t
h
e
m
w
i
th
d
ataset
s
o
f
f
ac
e
p
r
o
f
iles
[
5
]
.
R
ec
en
tl
y
d
ee
p
lear
n
i
n
g
p
la
y
s
a
m
aj
o
r
r
o
le
as
a
co
m
p
u
ter
v
i
s
io
n
to
o
l
[
6
]
,
[
7
]
.
I
n
d
ee
p
lear
n
in
g
tec
h
n
o
lo
g
y
,
a
co
n
v
o
lu
t
io
n
al
n
eu
r
al
n
et
w
o
r
k
(
C
NN)
is
a
d
ee
p
n
eu
r
al
n
et
w
o
r
k
ar
ch
itect
u
r
e
cl
ass
t
h
at
i
s
m
o
s
tl
y
u
s
ed
to
an
a
l
y
ze
t
h
e
v
is
u
al
i
m
a
g
er
y
[
8
]
,
[
9
]
.
A
d
ee
p
C
NN
's
m
at
h
e
m
a
tical
co
n
s
tr
u
ctio
n
co
n
s
i
s
ts
o
f
s
ev
er
al
t
y
p
es
o
f
la
y
e
r
s
:
co
n
v
o
lu
tio
n
al
la
y
er
s
,
p
o
o
li
n
g
,
a
n
d
a
co
n
n
ec
ted
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
ed
estr
ia
n
a
g
e
esti
ma
tio
n
b
a
s
ed
o
n
d
ee
p
lea
r
n
in
g
(
N
a
w
a
l Y
o
u
n
is
A
b
d
u
lla
h
)
1549
la
y
er
.
T
h
e
co
n
v
o
lu
tio
n
a
n
d
p
o
o
lin
g
la
y
er
s
'
f
u
n
ctio
n
ar
e
to
p
er
f
o
r
m
f
ea
t
u
r
e
ex
tr
ac
ti
o
n
,
w
h
ile
t
h
e
f
u
ll
y
co
n
n
ec
ted
la
y
er
f
ed
s
w
it
h
t
h
e
r
esu
lt o
f
t
h
ese
p
r
o
ce
s
s
e
s
to
g
et
th
e
f
i
n
al
cla
s
s
i
f
icatio
n
d
ec
is
io
n
[
1
0
]
,
[
1
1
]
.
Fig
u
r
e
1
.
Dea
th
r
ate
f
r
o
m
r
o
ad
tr
af
f
ic
f
o
r
th
e
p
er
io
d
(
2
0
0
0
–
2
0
1
6
)
p
er
1
0
0
,
0
0
0
p
o
p
u
lati
on
VGG
-
Face
an
d
R
esNet
-
5
0
ar
e
to
p
-
r
ated
d
ee
p
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
w
o
r
k
(
DC
N
N)
m
o
d
el
s
;
w
e
p
r
o
p
o
s
e
to
u
s
e
th
e
s
e
t
w
o
m
o
d
els
to
i
m
p
r
o
v
e
th
e
esti
m
atio
n
o
f
p
ed
estrian
a
g
e.
U
s
i
n
g
th
ese
ar
ch
itec
tu
r
es
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
f
r
o
m
i
m
ag
e
s
f
ac
i
litates
tr
ad
itio
n
al
m
et
h
o
d
s
'
d
i
f
f
icu
lt
y
b
ec
a
u
s
e
i
t
is
n
ee
d
ed
to
s
elec
t
w
h
ic
h
f
ea
t
u
r
es
ar
e
s
i
g
n
i
f
ica
n
t
i
n
e
ac
h
g
iv
e
n
i
m
a
g
e
[
1
2
]
,
[
1
3
]
.
W
e
d
o
n
'
t
r
eq
u
ir
e
to
u
s
e
f
ea
tu
r
e
d
escr
ip
to
r
s
:
s
ca
le
-
i
n
v
ar
ian
t
f
ea
t
u
r
e
tr
an
s
f
o
r
m
(
SIFT
)
,
s
p
ee
d
ed
u
p
r
o
b
u
s
t
f
ea
t
u
r
es
(
SU
R
F
)
,
B
R
I
E
F
,
h
is
to
g
r
a
m
-
o
r
ien
te
d
g
r
ad
ien
t
(
HOG)
an
d
S
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
e
(
SVM)
f
o
r
f
u
r
t
h
er
r
ec
o
g
n
it
io
n
o
r
class
i
f
icatio
n
tas
k
s
[
1
4
]
.
Ma
ch
i
n
e
lear
n
in
g
i
s
also
n
o
t
f
r
ee
o
f
d
if
f
ic
u
ltie
s
;
o
v
er
f
itti
n
g
p
r
o
b
lem
s
r
ep
r
esen
t
o
n
e
o
f
t
h
e
m
o
s
t
d
i
f
f
icu
lt
p
r
o
b
lem
s
w
h
e
n
u
s
in
g
s
m
al
l
d
atasets
.
T
h
is
p
r
o
b
lem
r
es
u
lt
s
f
r
o
m
t
h
e
e
n
o
r
m
o
u
s
n
u
m
b
er
s
o
f
d
ee
p
n
et
w
o
r
k
p
ar
am
eter
s
s
in
ce
t
h
e
y
h
av
e
m
u
ltip
le
la
y
er
s
a
n
d
th
o
u
s
a
n
d
s
o
f
n
o
d
es.
Mo
s
t
o
f
th
e
s
tu
d
ie
s
o
n
ag
e
esti
m
atio
n
h
av
e
o
cc
u
r
r
ed
r
ec
en
tl
y
,
s
o
w
e
f
i
n
d
th
at
m
o
s
t
o
f
t
h
e
d
atab
ases
b
u
ilt
f
o
r
th
is
p
u
r
p
o
s
e
ar
e
s
m
a
ll
in
s
ize.
T
o
s
o
lv
e
th
e
o
v
er
f
itti
n
g
p
r
o
b
le
m
,
w
e
b
u
ilt o
u
r
p
r
o
p
o
s
ed
p
ed
estrian
ag
e
esti
m
a
tio
n
s
y
s
te
m
b
ased
o
n
a
d
ee
p
C
NN
m
o
d
el,
w
h
ic
h
tr
ain
ed
o
n
a
h
u
g
e
d
atab
ase
[
1
5
]
.
T
h
e
m
ai
n
co
n
tr
ib
u
tio
n
o
f
th
i
s
p
ap
er
is
to
ac
h
iev
e
th
e
b
est
p
er
f
o
r
m
an
ce
f
o
r
h
u
m
a
n
ag
e
e
s
ti
m
atio
n
t
h
r
o
u
g
h
t
h
e
co
m
b
in
a
tio
n
o
f
V
GG
-
Face
an
d
R
es
Net
-
5
0
m
o
d
els.
Ma
n
y
r
elate
d
s
t
u
d
ies
h
av
e
b
e
en
p
u
b
lis
h
ed
o
n
u
s
in
g
v
id
eo
d
ata
to
in
v
e
s
ti
g
ate
p
ed
estrian
s
cr
o
s
s
i
n
g
b
eh
av
io
u
r
s
.
A
n
in
v
es
tig
a
tio
n
o
f
p
ed
estrian
'
s
lo
ca
tio
n
v
io
lati
o
n
s
i
n
[
1
6
]
,
a
s
tu
d
y
o
n
p
r
ed
ictio
n
o
f
p
ed
estria
n
s
r
ed
-
lig
h
t
cr
o
s
s
in
g
in
te
n
tio
n
s
b
ased
o
n
th
e
ap
p
ea
r
an
ce
ch
ar
ac
ter
is
tics
:
g
e
n
d
er
,
ag
e
an
d
h
ea
d
d
ir
ec
tio
n
in
[
1
7
]
,
a
SOR
T
tr
ac
k
in
g
m
o
d
el
i
n
[
1
8
]
,
an
d
a
s
tu
d
y
b
ased
o
n
r
e
g
io
n
-
b
a
s
ed
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
(R
-
C
NN)
o
b
j
ec
t
d
etec
t
io
n
m
o
d
el
h
as
b
ee
n
p
r
o
p
o
s
ed
in
[
1
9
]
.
A
lt
h
o
u
g
h
t
h
ese
p
r
e
v
io
u
s
s
t
u
d
ies
w
er
e
b
ased
o
n
m
ac
h
in
e
lear
n
in
g
m
o
d
els,
th
e
y
w
er
e
n
o
t
ef
f
icien
t
in
d
etec
tin
g
th
e
r
elatio
n
s
h
ip
s
i
n
th
e
s
er
ies
o
f
d
ata
f
o
r
f
u
tu
r
e
p
r
ed
ictio
n
s
.
Ou
r
w
o
r
k
i
n
p
ed
estrian
d
etec
tio
n
is
b
ased
o
n
ag
e
esti
m
at
io
n
,
a
s
t
u
d
y
b
ased
o
n
d
ee
p
ex
p
ec
t
o
f
v
is
ib
le
a
g
e
f
r
o
m
a
s
i
n
g
le
i
m
a
g
e,
w
h
ic
h
is
p
er
f
o
r
m
ed
u
s
in
g
C
NN
ar
ch
itect
u
r
e
p
r
esen
ted
i
n
[
2
0
]
.
T
h
e
p
r
o
b
lem
o
f
th
i
s
m
e
th
o
d
is
ap
p
r
o
ac
h
ed
as
a
class
i
f
icatio
n
p
r
o
b
lem
w
i
th
1
0
1
ag
e
class
es.
I
n
s
ec
tio
n
2
o
f
th
is
w
o
r
k
,
w
e
d
escr
ib
ed
th
e
p
r
o
p
o
s
ed
m
o
d
el.
T
h
e
ex
p
er
im
e
n
tal
m
o
d
el
tr
ai
n
in
g
an
d
r
esu
lts
h
a
v
e
b
ee
n
p
r
o
p
o
s
ed
in
s
ec
tio
n
3.
Fin
all
y
,
t
h
e
c
o
n
cl
u
s
io
n
s
ar
e
r
ev
ie
w
ed
in
s
ec
tio
n
4
.
2.
T
H
E
P
RO
P
O
SE
D
M
O
DE
L
Ou
r
p
r
o
p
o
s
ed
m
eth
o
d
f
o
r
p
ed
estrian
ag
e
e
s
ti
m
atio
n
f
o
c
u
s
ed
o
n
tr
ain
in
g
a
ge
-
d
ep
en
d
en
t
f
ac
e
r
ep
r
esen
tatio
n
to
en
h
a
n
ce
s
y
s
t
e
m
p
er
f
o
r
m
an
ce
,
an
d
it
is
b
ased
o
n
tw
o
p
o
p
u
lar
p
r
e
-
tr
ain
ed
m
o
d
el
s
(
R
esNet
-
50
an
d
VGG
-
Face
)
to
ex
p
lo
it
a
g
e
in
f
o
r
m
atio
n
f
o
r
i
m
p
r
o
v
in
g
t
h
e
d
etec
tio
n
s
y
s
te
m
.
A
t
f
ir
s
t,
th
e
s
y
s
te
m
d
etec
t
s
f
ac
es
f
r
o
m
th
e
i
n
p
u
t
i
m
ag
e
s
an
d
en
clo
s
i
n
g
t
h
e
m
b
y
a
b
o
u
n
d
in
g
b
o
x
b
ased
o
n
Haa
r
-
ca
s
ca
d
e
m
et
h
o
d
[
2
1
]
.
Haar
-
ca
s
ca
d
e
is
a
m
ac
h
i
n
e
le
ar
n
in
g
al
g
o
r
ith
m
f
o
r
o
b
j
ec
t
d
etec
tio
n
.
I
n
t
h
is
m
et
h
o
d
th
e
c
ascad
e
f
u
n
ctio
n
i
s
tr
ain
ed
b
y
a
lo
t
o
f
p
o
s
itiv
e
a
n
d
n
eg
at
iv
e
i
m
a
g
es,
w
h
er
e
th
e
o
b
j
ec
t
th
at
w
a
n
t
to
b
e
d
etec
ted
is
ex
is
t
i
n
th
e
p
o
s
itiv
e
i
m
a
g
es,
w
h
ile
t
h
e
n
e
g
ati
v
e
i
m
ag
e
s
ar
e
t
h
o
s
e
w
h
er
e
it
is
n
o
t.
T
h
is
m
et
h
o
d
h
as
b
ee
n
u
s
ed
to
d
etec
t
f
ac
es
i
n
i
m
a
g
es.
T
h
en
,
a
n
ag
e
esti
m
ato
r
m
o
d
u
le
h
as
b
ee
n
u
s
ed
to
p
r
ed
icate
th
e
a
g
e
o
f
p
ed
estria
n
s
au
to
m
at
icall
y
.
Fig
u
r
e
2
s
h
o
w
s
a
d
iag
r
a
m
o
f
t
h
e
m
o
d
el
ar
ch
it
ec
tu
r
e.
2
.
1
.
T
he
pre
-
t
ra
ined deep
CNNs
m
o
del
s
B
o
th
R
esNet
-
5
0
co
n
v
o
lu
tio
n
n
et
w
o
r
k
m
o
d
el
s
u
g
g
es
ted
in
[
2
2
]
,
an
d
VGG
-
Face
co
n
v
o
l
u
ti
o
n
n
et
w
o
r
k
m
o
d
el,
s
u
g
g
e
s
ted
in
[
2
3
]
,
[
2
4
]
h
av
e
b
ee
n
u
s
ed
in
t
h
i
s
w
o
r
k
t
o
ac
h
iev
e
s
o
m
e
o
f
th
e
b
es
t
p
er
f
o
r
m
a
n
ce
in
a
n
ag
e
esti
m
atio
n
tas
k
.
R
esNe
t
-
5
0
is
a
d
ee
p
C
NNs
b
ased
o
n
r
es
id
u
al
n
eu
r
o
n
al
n
e
t
w
o
r
k
ar
ch
itect
u
r
e.
T
h
is
n
et
w
o
r
k
i
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
5
4
8
-
1
5
5
5
1550
d
is
tin
g
u
is
h
ed
b
y
ad
d
i
n
g
a
s
h
o
r
tcu
t
co
n
n
ec
tio
n
b
et
w
ee
n
b
lo
ck
s
,
w
h
ic
h
s
w
itc
h
es
th
e
co
n
v
o
l
u
tio
n
n
et
w
o
r
k
to
a
r
esid
u
al
n
e
u
r
al
n
et
w
o
r
k
v
er
s
i
o
n
.
T
h
e
n
et
w
o
r
k
co
n
s
i
s
ts
o
f
3
4
r
esid
u
al
la
y
er
s
f
o
r
tr
ain
i
n
g
a
n
d
4
1
,
1
9
2
,
9
5
1
p
ar
am
eter
s
;
it
h
a
s
b
ee
n
ad
ap
ted
to
ag
e
esti
m
at
io
n
to
b
e
u
s
ed
in
th
is
s
t
u
d
y
.
T
h
e
R
e
s
Net
-
5
0
m
o
d
el
s
tr
u
ct
u
r
e
is
s
h
o
w
n
in
F
ig
u
r
e
3
.
T
h
e
VGG
-
Face
n
et
w
o
r
k
is
an
o
th
er
d
ee
p
C
NNs
b
ased
o
n
VGG
-
1
6
n
et
w
o
r
k
ar
ch
itectu
r
e,
it
co
n
s
is
ts
o
f
1
1
la
y
er
s
,
8
b
lo
ck
s
o
f
co
n
v
o
l
u
tio
n
a
l
la
y
er
s
,
a
n
d
th
r
ee
f
u
ll
y
-
co
n
n
ec
ted
la
y
er
s
w
it
h
r
ec
tif
ied
li
n
ea
r
u
n
i
t
(
R
eL
U)
ac
t
iv
at
io
n
s
,
t
h
e
n
et
wo
r
k
co
n
f
i
g
u
r
atio
n
i
s
s
h
o
w
n
in
T
ab
le
1
.
VGG
-
Face
i
s
ap
p
lie
d
as
a
f
ac
ial
f
ea
t
u
r
e
ex
tr
ac
to
r
f
o
r
ag
e
esti
m
atio
n
p
u
r
p
o
s
es f
r
o
m
th
e
f
ac
e
i
m
ag
e
s
.
Fig
u
r
e
2
.
T
h
e
p
e
d
estrian
ag
e
e
s
ti
m
atio
n
m
o
d
el
ar
ch
itect
u
r
e
Fig
u
r
e
3
.
R
esNet
-
5
0
m
o
d
el
s
tr
u
ctu
r
e
T
ab
le
1
.
VGG
-
Face
la
y
er
s
s
tr
u
ctu
r
e
[
2
2
]
L
a
y
e
r
t
y
p
e
n
a
me
S
u
p
p
o
r
t
F
i
l
t
d
i
m
N
u
m
f
i
l
t
S
t
r
i
d
e
P
a
d
L
a
y
e
r
t
y
p
e
n
a
me
S
u
p
p
o
r
t
F
i
l
t
d
i
m
N
u
m
f
i
l
t
S
t
r
i
d
e
P
a
d
0
i
n
p
u
t
-
-
-
-
-
1
9
r
e
l
u
r
e
l
u
4
_
1
1
-
-
1
0
1
c
o
n
v
c
o
n
v
1
_
1
3
3
64
1
1
2
0
c
o
n
v
c
o
n
v
4
_
2
3
5
1
2
5
1
2
1
1
2
r
e
l
u
r
e
l
u
1
_
1
1
-
-
1
0
2
1
r
e
l
u
r
e
l
u
4
_
2
1
-
-
1
0
3
c
o
n
v
c
o
n
v
1
_
2
3
64
64
1
1
2
2
c
o
n
v
c
o
n
v
4
_
3
3
5
1
2
5
1
2
1
1
4
r
e
l
u
r
e
l
u
1
_
2
1
-
-
1
0
2
3
r
e
l
u
r
e
l
u
4
_
3
1
-
-
1
0
5
m
p
o
o
l
p
o
o
l
1
2
-
-
2
0
2
4
mp
o
o
l
p
o
o
l
4
2
-
-
2
0
6
c
o
n
v
c
o
n
v
2
_
1
3
64
1
2
8
1
1
2
5
c
o
n
v
c
o
n
v
5
_
1
3
5
1
2
5
1
2
1
1
7
r
e
l
u
r
e
l
u
2
_
1
1
-
-
1
0
2
6
r
e
l
u
r
e
l
u
5
_
1
1
-
-
1
0
8
c
o
n
v
c
o
n
v
2
_
2
3
1
2
8
1
2
8
1
1
2
7
c
o
n
v
c
o
n
v
5
_
2
3
5
1
2
5
1
2
1
1
9
r
e
l
u
r
e
l
u
2
_
2
1
-
-
1
0
2
8
r
e
l
u
r
e
l
u
5
_
2
1
-
-
1
0
1
0
mp
o
o
l
p
o
o
l
2
2
-
-
2
0
3
0
r
e
l
u
r
e
l
u
5
_
3
1
-
-
1
0
1
1
c
o
n
v
c
o
n
v
3
_
2
3
1
2
8
2
5
6
1
1
3
1
mp
o
o
l
p
o
o
l
5
2
-
-
2
0
1
2
r
e
l
u
r
e
l
u
3
_
1
1
-
-
1
0
3
2
c
o
n
v
f
c
6
7
5
1
2
4
0
9
6
1
0
1
3
c
o
n
v
c
o
n
v
3
_
2
3
2
5
6
2
5
6
1
1
3
3
r
e
l
u
r
e
l
u
6
1
-
-
1
0
1
4
r
e
l
u
r
e
l
u
3
_
2
1
-
-
1
0
3
4
c
o
n
v
f
c
7
1
4
0
9
6
4
0
9
6
1
0
1
5
c
o
n
v
c
o
n
v
3
_
3
3
2
5
6
2
5
6
1
1
3
5
r
e
l
u
r
e
l
u
7
1
-
-
1
0
1
6
r
e
l
u
r
e
l
u
3
_
3
1
-
-
1
0
3
6
c
o
n
v
f
c
8
1
4
0
9
6
2
6
2
2
1
0
1
7
mp
o
o
l
p
o
o
l
3
2
-
-
2
0
3
7
so
f
t
mx
p
r
o
b
1
-
-
1
0
1
8
c
o
n
v
c
o
n
v
4
_
1
3
2
5
6
5
1
2
1
1
2
.
2
.
Ass
e
m
ble
lea
rning
f
o
r
a
g
e
esti
m
a
t
io
n
T
h
e
p
r
o
p
o
s
ed
ag
e
esti
m
a
to
r
is
b
u
ilt
b
ased
o
n
f
i
n
e
-
t
u
n
i
n
g
th
e
p
r
e
-
tr
ain
ed
m
o
d
els
R
es
Net
-
5
0
an
d
VGG
-
Face
to
lev
er
a
g
in
g
a
n
e
n
s
e
m
b
le
ap
p
r
o
ac
h
s
in
ce
th
e
y
ar
e
ac
h
iev
i
n
g
g
r
ea
ter
d
is
cr
i
m
i
n
atio
n
r
ate
s
w
h
e
n
co
m
p
ar
e
t
h
e
m
w
i
th
o
th
er
s
i
m
i
lar
m
o
d
els.
A
v
ec
to
r
o
f
w
ei
g
h
ts
is
o
b
tain
ed
f
r
o
m
th
e
s
ec
o
n
d
to
th
e
last
la
y
er
o
f
th
e
co
n
v
o
lu
tio
n
al
m
o
d
el,
w
h
i
ch
ca
n
b
e
u
s
ed
as
f
ea
t
u
r
e
v
ec
to
r
s
.
T
h
e
w
ei
g
h
t
s
f
ea
tu
r
e
v
ec
t
o
r
s
co
n
s
id
er
ed
th
e
laten
t
r
ep
r
esen
tatio
n
f
o
r
th
e
in
p
u
t
i
m
ag
e
w
h
ich
ea
ch
m
o
d
e
l
lear
n
ed
.
I
n
t
h
e
p
r
o
p
o
s
ed
m
e
th
o
d
,
w
e
co
m
b
in
ed
th
ese
late
n
t
r
ep
r
esen
ta
tio
n
s
b
y
co
n
ca
te
n
ati
n
g
t
h
e
f
ea
tu
r
e
v
ec
to
r
s
to
f
o
r
m
a
n
o
v
er
all
f
ea
t
u
r
e
v
ec
to
r
in
p
u
tted
in
to
lo
g
i
s
tic
r
eg
r
es
s
io
n
m
o
d
els
f
o
r
th
e
f
i
n
al
a
g
e
p
r
ed
ictio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
ed
estr
ia
n
a
g
e
esti
ma
tio
n
b
a
s
ed
o
n
d
ee
p
lea
r
n
in
g
(
N
a
w
a
l Y
o
u
n
is
A
b
d
u
lla
h
)
1551
2
.
3
.
Da
t
a
ba
s
e
VGGFac
e2
d
ataset
u
s
ed
in
th
i
s
w
o
r
k
,
p
r
o
v
id
es
a
d
ataset
o
f
an
n
o
tatio
n
s
o
f
3
.
3
1
m
illi
o
n
f
a
ce
im
a
g
es
b
elo
n
g
to
9
1
3
1
s
u
b
j
ec
ts
an
d
ab
o
u
t
3
6
2
.
6
im
a
g
es
f
o
r
ea
ch
s
u
b
j
ec
t.
VGGFac
e2
h
as
b
ee
n
d
ev
elo
p
ed
f
o
r
th
e
ad
v
an
ce
d
s
t
u
d
y
o
f
f
ac
e
r
ec
o
g
n
itio
n
s
y
s
te
m
s
,
i
t
h
a
s
a
lar
g
e
v
ar
iet
y
a
n
d
av
a
ilab
le
lab
els
f
o
r
ag
e
an
d
p
o
s
e.
T
h
e
VGGFac
e2
d
ataset's
an
n
o
tati
o
n
s
ar
e
m
ad
e
o
n
f
ac
e
s
s
a
m
p
l
es
tak
en
f
r
o
m
Go
o
g
le
I
m
a
g
e
[
2
5
]
.
A
lth
o
u
g
h
th
e
b
lack
an
d
w
h
ite
f
ac
e
i
m
a
g
es
ar
e
r
o
b
u
s
t
in
f
ac
e
r
ec
o
g
n
itio
n
b
u
t
th
e
y
ar
e
ex
cl
u
d
ed
,
T
h
e
VGGFac
e2
d
ataset
f
o
cu
s
ed
o
n
f
ac
ial
an
d
i
m
ag
e
v
ar
iatio
n
d
u
e
to
co
lo
r
p
r
o
ce
s
s
in
g
a
s
s
h
o
w
n
i
n
Fig
u
r
e
4
.
Fiv
e
ag
e
class
es
h
a
v
e
b
ee
n
in
cl
u
d
ed
in
th
is
s
t
u
d
y
{(
0
0
-
1
0
)
,
(
1
1
-
2
0
)
,
(
2
1
-
3
5
)
,
(
3
6
-
5
4
)
,
(
5
5
-
9
0
)
}.
Fig
u
r
e
4
.
A
s
a
m
p
le
o
f
f
ac
e
i
m
ag
e
f
r
o
m
VGGFac
e2
d
ataset
2
.
4
.
P
re
pro
ce
s
s
ing
da
t
a
s
et
T
h
e
en
v
ir
o
n
m
e
n
t
o
f
th
i
s
w
o
r
k
is
b
ased
o
n
T
en
s
o
r
Flo
w
,
w
h
i
ch
r
ep
r
esen
t
th
e
co
r
e
o
p
en
-
s
o
u
r
ce
lib
r
ar
y
to
h
elp
r
esear
ch
er
d
ev
elo
p
an
d
tr
ain
ML
m
o
d
els.
A
b
o
u
t
3
0
k
i
m
a
g
es
o
f
VGGFac
e2
h
a
v
e
b
ee
n
u
s
ed
.
B
ef
o
r
e
lo
ad
in
g
i
n
p
u
t
i
m
a
g
es,
t
h
e
wo
r
k
in
g
en
v
ir
o
n
m
e
n
t
n
ee
d
s
o
m
e
p
r
ep
ar
atio
n
s
u
c
h
as
i
n
s
t
allin
g
t
h
e
r
eq
u
ir
ed
lib
r
ar
ies,
th
en
s
tar
t
to
lo
ad
th
e
d
ataset.
A
p
r
ep
r
o
ce
s
s
in
g
d
ata
,
is
r
eq
u
ir
ed
at
th
is
s
tag
e
i
n
cl
u
d
in
g
clea
n
s
i
n
g
th
e
d
ataset
f
r
o
m
lo
w
q
u
ali
t
y
i
m
a
g
es
w
h
ich
w
ill co
n
f
u
s
e
th
e
tr
ain
in
g
m
o
d
el,
A
r
elab
eli
n
g
ag
e
g
r
o
u
p
i
m
a
g
e
s
to
f
i
v
e
ca
teg
o
r
ies
{(
0
0
-
1
0
)
,
(
1
1
-
2
0
)
,
(
2
1
-
3
5
)
,
(
3
6
-
5
4
)
,
(
5
5
-
9
0
)
},
s
p
litt
in
g
d
ataset
to
8
0
%
tr
ain
,
1
0
%
test
an
d
1
0
%
v
alid
atio
n
a
n
d
f
i
n
all
y
,
s
et
th
e
b
atch
s
ize.
T
h
e
ag
e
esti
m
atio
n
m
o
d
el
is
tr
ai
n
ed
w
it
h
t
h
r
ee
p
r
e
-
tr
ain
ed
-
w
e
ig
h
t,
an
d
t
w
o
d
ee
p
lear
n
i
n
g
al
g
o
r
i
th
m
s
(
VGG
-
Face
an
d
R
es
Net5
0
)
as it illu
s
tr
ated
in
t
h
e
ex
p
er
i
m
en
tal
w
o
r
k
.
3.
E
XP
E
R
I
M
E
NT
A
L
WO
RK
AND
RE
SUL
T
S
T
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
's
ex
p
e
r
i
m
en
tal
w
o
r
k
w
as
d
esi
g
n
ed
to
tr
ain
an
au
to
m
atic
p
ed
estrian
ag
e
esti
m
ato
r
u
s
i
n
g
f
ac
e
i
m
a
g
es
d
ataset
VGG
Face
2
an
d
an
al
y
ze
th
e
ef
f
ec
t
o
f
u
s
in
g
t
h
e
t
w
o
d
ee
p
lear
n
in
g
m
o
d
els,
R
esNet
-
5
0
an
d
VGG
-
Face
.
T
o
ev
alu
ate
th
e
s
y
s
te
m
p
er
f
o
r
m
a
n
ce
an
d
ef
f
icien
c
y
,
m
a
n
y
e
x
p
er
i
m
en
ts
h
a
v
e
b
ee
n
p
er
f
o
r
m
ed
(
E
x
p
er
i
m
en
t
1
)
;
th
e
R
esNet
-
5
0
m
o
d
el
is
lo
ad
ed
w
it
h
t
h
e
p
r
e
-
tr
ain
ed
w
ei
g
h
t
o
f
"
I
m
ag
eNe
t"
,
to
u
s
e
i
t
as
a
b
ase
m
o
d
el.
T
h
e
I
m
ag
eNe
t
lar
g
e
s
ca
le
v
is
u
al
r
ec
o
g
n
itio
n
ch
alle
n
g
e
(
I
L
SV
R
C
)
is
a
lar
g
e
v
is
u
al
d
atab
ase
co
n
tain
in
g
m
o
r
e
t
h
an
1
4
m
illi
o
n
i
m
a
g
es i
n
cl
u
d
ed
with
a
to
tal
o
f
2
0
,
0
0
0
ca
teg
o
r
ies,
w
h
ic
h
i
s
u
s
ed
f
o
r
o
b
j
ec
t
ca
teg
o
r
y
class
i
f
icat
io
n
an
d
d
etec
tio
n
[
26
]
-
[
2
9
]
.
T
h
e
r
esu
lt
f
r
o
m
R
esNet
-
5
0
p
r
esen
ts
th
at
b
o
th
tr
ain
an
d
test
p
er
f
o
r
m
a
n
ce
ar
e
less
t
h
an
5
0
%
as
s
h
o
w
n
in
F
ig
u
r
e
5
,
th
i
s
is
d
u
e
to
th
e
I
m
ag
e
Net
is
n
o
t
s
p
ec
if
ied
f
o
r
f
ac
e
i
m
a
g
es
a
n
d
co
n
tai
n
s
v
ar
io
u
s
i
m
a
g
e
ca
te
g
o
r
ies.
Fi
g
u
r
e
5
ill
u
s
tr
ate
th
e
ac
c
u
r
ac
y
o
f
R
es
N
et
-
5
0
,
th
e
c
u
r
v
e
o
f
ac
cu
r
ac
y
o
n
t
h
e
tr
ain
in
g
d
ata
(
ac
c)
w
h
ile
(
v
al_
ac
c)
is
t
h
e
cu
r
v
e
o
f
ac
cu
r
ac
y
o
n
t
h
e
v
alid
ati
o
n
d
ata.
(
E
x
p
er
i
m
en
t
2
)
; T
h
e
VGGFac
e2
d
ataset
is
d
iv
id
ed
in
to
th
e
t
r
ain
in
g
s
et,
v
alid
atio
n
s
et,
an
d
test
s
et,
to
tr
ain
b
o
th
VGG
-
Face
an
d
R
esNet
-
5
0
p
r
e
-
tr
ain
ed
m
o
d
els
as
th
e
b
aselin
e,
w
h
ic
h
r
esu
lted
f
r
o
m
a
h
ig
h
er
p
er
f
o
r
m
a
n
ce
o
f
m
o
r
e
t
h
a
n
9
0
%.
Mo
r
eo
v
er
,
T
h
e
R
esNet
-
50
r
esu
lted
i
n
a
h
i
g
h
er
ac
cu
r
ac
y
p
er
f
o
r
m
an
ce
th
a
n
VGG
-
Face
,
as
s
h
o
w
n
i
n
Fi
g
u
r
e
6
.
Fin
all
y
,
(
3
)
,
to
in
cr
ea
s
e
th
e
s
y
s
te
m
e
f
f
icien
c
y
a
n
d
en
h
a
n
ce
th
e
p
er
f
o
r
m
an
ce
w
e
co
m
b
in
ed
th
e
t
w
o
D
C
NN
m
o
d
el
s
i
n
t
h
i
s
e
x
p
er
i
m
e
n
t,
t
h
e
s
y
s
te
m
ac
c
u
r
ac
y
in
cr
ea
s
ed
b
o
th
i
n
tr
ai
n
(
0
.
9
6
8
9
)
an
d
test
s
et
(
0
.
8
8
0
1
)
.
T
h
e
s
y
s
t
e
m
tr
ain
ed
u
s
i
n
g
1
5
0
e
p
o
ch
s
.
Ag
e
p
r
ed
ictio
n
ac
cu
r
ac
y
b
ec
o
m
es
(
0
.
8
6
7
8
)
o
n
th
e
v
alid
atio
n
s
et
a
t
ep
o
ch
5
0
.
W
e
u
s
e
t
h
e
A
d
ad
elta
o
p
ti
m
izatio
n
an
d
s
p
ar
s
e
ca
te
g
o
r
ical
cr
o
s
s
-
en
tr
o
p
y
,
t
h
e
m
o
d
el
i
m
p
le
m
en
ted
u
s
i
n
g
T
en
s
o
r
Flo
w
lib
r
ar
y
.
T
h
e
s
y
s
te
m
p
er
f
o
r
m
an
ce
f
o
r
ex
p
er
i
m
en
t
s
3
illu
s
tr
a
ted
in
Fig
u
r
e
7
.
I
n
th
is
s
t
u
d
y
,
th
e
class
d
ata
is
i
m
b
alan
ce
d
.
T
h
er
ef
o
r
e,
a
c
o
n
f
u
s
io
n
m
atr
i
x
w
it
h
n
o
r
m
a
li
za
tio
n
h
as
b
ee
n
u
s
ed
.
T
h
e
b
est
ca
teg
o
r
y
p
r
ed
ictio
n
is
o
n
th
e
ag
e
o
f
0
0
-
1
0
y
ea
r
s
o
ld
,
an
d
th
e
w
o
r
s
t
p
r
ed
ictio
n
is
o
n
55
-
9
0
y
ea
r
s
o
ld
.
T
h
is
is
b
ec
au
s
e
o
f
f
ac
e
s
tr
u
ct
u
r
e;
th
e
ch
ild
r
en
f
ac
e
f
r
o
m
0
0
-
1
0
y
e
ar
s
o
ld
ca
n
ea
s
il
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
5
4
8
-
1
5
5
5
1552
d
is
tin
g
u
i
s
h
f
r
o
m
o
t
h
er
ag
e
g
r
o
u
p
s
.
T
h
er
ef
o
r
e,
th
e
s
y
s
te
m
cla
s
s
i
f
ied
t
h
is
a
g
e
g
r
o
u
p
as
th
e
b
est.
T
h
e
co
n
f
u
s
io
n
m
atr
i
x
is
s
h
o
w
n
in
F
ig
u
r
e
8
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
's
te
s
t
p
er
f
o
r
m
an
ce
i
s
(
8
8
.
0
1
%)
an
d
ca
n
b
e
in
cr
ea
s
ed
if
th
e
p
ed
estrian
f
ac
e
i
m
a
g
e
tak
e
n
i
n
th
e
clo
s
e
s
h
o
t
w
i
ll
b
e
th
e
h
i
g
h
-
q
u
a
lit
y
r
e
s
o
lu
te
i
m
a
g
e,
s
o
m
e
o
f
t
h
e
test
r
e
s
u
lt
i
m
a
g
es a
r
e
s
h
o
w
n
i
n
Fi
g
u
r
e
9
.
Fig
u
r
e
5
.
T
h
e
ac
cu
r
ac
y
o
f
R
es
Net
-
5
0
lo
ad
ed
w
ith
I
m
ag
e
Net
w
ei
g
h
ts
f
o
r
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
(
a)
(
b
)
Fig
u
r
e
6
.
T
h
e
ac
cu
r
ac
y
o
f
s
y
s
t
e
m
tr
ain
in
g
an
d
tes
tin
g
f
o
r
b
o
t
h
VGG
-
Face
a
n
d
R
esNet
-
5
0
m
o
d
el
s
w
it
h
VGGFac
e2
d
ataset
:
(
a)
VGG
-
Face
tr
ain
ed
w
it
h
VGGFac
e2
d
ataset
ac
cu
r
ac
y
.
T
h
e
b
est tr
ain
ac
cu
r
ac
y
: 0
.
9
0
9
2
.
T
h
e
b
est test
ac
cu
r
ac
y
: 0
.
8
6
0
3
an
d
(
b
)
R
esNet
-
5
0
tr
ain
ed
w
it
h
VGGFac
e2
d
ataset
ac
c
u
r
ac
y
.
T
h
e
b
est tr
ain
ac
cu
r
ac
y
: 0
.
9
3
9
2
.
T
h
e
b
est test
ac
cu
r
ac
y
: 0
.
8
7
5
1
Fig
u
r
e
7
.
T
h
e
o
v
er
all
s
y
s
te
m
p
er
f
o
r
m
an
ce
b
y
co
m
b
i
n
i
n
g
t
h
e
t
w
o
D
C
NN
m
o
d
els
Fig
u
r
e
8
.
C
o
n
f
u
s
io
n
m
atr
ix
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
ed
estr
ia
n
a
g
e
esti
ma
tio
n
b
a
s
ed
o
n
d
ee
p
lea
r
n
in
g
(
N
a
w
a
l Y
o
u
n
is
A
b
d
u
lla
h
)
1553
Fig
u
r
e
9
.
So
m
e
o
f
test
r
es
u
lt i
m
ag
e
s
4.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
im
p
lie
s
a
co
m
b
i
n
atio
n
o
f
t
w
o
d
ee
p
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
DC
N
N
m
o
d
el'
s
ar
ch
itect
u
r
e
to
esti
m
ate
t
h
e
p
ed
estrian
s
'
a
g
e
f
r
o
m
t
h
e
p
r
o
v
id
ed
im
a
g
es
o
f
th
e
s
u
r
v
eilla
n
ce
ca
m
er
a
s
y
s
te
m
.
T
h
e
r
esear
ch
co
n
tr
ib
u
te
s
to
r
ed
u
cin
g
ac
cid
en
ts
r
es
u
lti
n
g
f
r
o
m
c
r
o
s
s
in
g
th
e
s
tr
ee
t,
i
n
ad
d
itio
n
to
aler
tin
g
v
e
h
icle
d
r
iv
er
s
to
th
e
p
r
esen
ce
o
f
p
ed
estrian
s
o
n
t
h
e
r
o
ad
.
T
h
e
ex
p
er
i
m
e
n
tal
w
o
r
k
s
h
o
w
s
th
e
ef
f
ic
ien
c
y
an
d
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
.
I
t
is
also
o
b
s
er
v
ed
th
at
t
h
er
e
ar
e
s
o
m
e
d
if
f
ic
u
lt
ies
to
esti
m
ate
t
h
e
eld
er
ly
a
g
e
g
r
o
u
p
,
b
esid
es
th
e
p
o
o
r
im
ag
e
r
eso
l
u
tio
n
p
r
o
b
le
m
.
W
e
w
ill
e
n
h
a
n
ce
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
i
n
th
e
f
u
tu
r
e
to
i
m
p
r
o
v
e
t
h
e
ef
f
icie
n
c
y
o
f
o
th
er
p
ed
estrian
f
ac
e
attr
i
b
u
tes.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
is
r
esear
ch
is
f
u
n
d
ed
b
y
co
o
p
e
r
atio
n
b
et
w
ee
n
d
ee
p
lear
n
in
g
tea
m
i
n
No
r
th
er
n
T
ec
h
n
ical
Un
i
v
er
s
it
y
NT
U
in
I
r
aq
.
W
eb
s
ite: Fo
r
No
r
th
er
n
T
ec
h
n
ical
U
n
iv
er
s
it
y
h
ttp
s
://
www
.
n
t
u
.
ed
u
.
iq
RE
F
E
R
E
NC
E
S
[1
]
W
o
rld
He
a
lt
h
Org
a
n
iza
ti
o
n
,
“
G
lo
b
a
l
sta
tu
s rep
o
rt
o
n
r
o
a
d
sa
f
e
t
y
,
”
W
o
rld
He
a
lt
h
Org
a
n
iza
ti
o
n
,
2
0
1
8
.
[2
]
D.
Du
iv
e
s,
W
.
Da
a
m
e
n
,
a
n
d
S
.
Ho
o
g
e
n
d
o
o
r
n
,
“
M
o
n
it
o
rin
g
th
e
Nu
m
b
e
r
o
f
P
e
d
e
strian
s
in
a
n
A
r
e
a
:
T
h
e
A
p
p
li
c
a
b
il
it
y
o
f
Co
u
n
t
in
g
S
y
ste
m
s
f
o
r
De
n
sity
S
tate
Esti
m
a
ti
o
n
,
”
J
o
u
rn
a
l
o
f
Ad
v
a
n
c
e
d
T
ra
n
s
p
o
rt
a
ti
o
n
,
v
o
l.
2
0
1
8
,
p
p
.
1
-
1
4
,
2
0
1
8
,
d
o
i:
1
0
.
1
1
5
5
/2
0
1
8
/7
3
2
8
0
7
4
.
[3
]
D.
C.
S
c
h
w
e
b
e
l,
A
.
L
.
Da
v
is,
a
n
d
E
.
E.
O'
Ne
a
l,
“
Ch
il
d
p
e
d
e
s
tri
a
n
i
n
ju
ry
:
A
re
v
ie
w
o
f
b
e
h
a
v
io
ra
l
risk
s
a
n
d
p
re
v
e
n
ti
v
e
stra
te
g
ies
,
”
Am
e
ric
a
n
jo
u
rn
a
l
o
f
li
fes
tyle
me
d
icin
e
,
v
o
l.
6
,
n
o
.
4
,
p
p
.
2
9
2
-
3
0
2
,
2
0
1
2
,
d
o
i:
1
0
.
1
1
7
7
/0
8
8
5
0
6
6
6
1
1
4
0
4
8
7
6
.
[4
]
P
.
P
a
tel
a
n
d
A
.
T
h
a
k
k
a
r,
“
T
h
e
u
p
su
rg
e
o
f
d
e
e
p
lea
rn
in
g
f
o
r
c
o
m
p
u
ter
v
isio
n
a
p
p
li
c
a
ti
o
n
s,
”
I
n
ter
n
a
t
i
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
,
v
o
l.
1
0
,
n
o
.
1
,
p
p
.
5
3
8
-
5
4
8
,
2
0
2
0
,
d
o
i:
1
0
.
1
1
5
9
1
/i
jec
e
.
v
1
0
i
1
.
p
p
5
3
8
-
5
4
8
.
[5
]
R.
I.
Be
n
d
ji
l
lali,
M
.
Be
lad
g
h
a
m
,
K.
M
e
rit
,
a
n
d
A
.
T
a
leb
-
A
h
m
e
d
,
“
Ill
u
m
in
a
ti
o
n
-
r
o
b
u
st
f
a
c
e
re
c
o
g
n
it
io
n
b
a
se
d
o
n
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
a
rc
h
it
e
c
tu
re
s,
”
In
d
o
n
e
si
a
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
i
n
e
e
rin
g
a
n
d
Co
mp
u
ter
S
c
ien
c
e
,
v
o
l.
1
8
,
n
o
.
2
,
p
p
.
1
0
1
5
-
1
0
2
7
,
2
0
2
0
,
d
o
i:
1
0
.
1
1
5
9
1
/
ij
e
e
c
s.v
1
8
.
i2
.
p
p
1
0
1
5
-
1
0
2
7
.
[6
]
D.
A
.
Ja
s
m
,
M
.
M
.
M
u
rtad
h
a
,
a
n
d
A
.
T
.
H.
A
lra
w
i,
“
De
e
p
i
m
a
g
e
m
in
in
g
f
o
r
c
o
n
v
o
lu
ti
o
n
n
e
u
ra
l
n
e
tw
o
rk
,
”
In
d
o
n
e
sia
n
J
o
u
r
n
a
l
o
f
E
lec
trica
l
En
g
i
n
e
e
rin
g
a
n
d
Co
mp
u
ter
S
c
ien
c
e
,
v
o
l.
2
0
,
n
o
.
1
,
p
p
.
3
4
7
-
3
5
2
,
2
0
2
0
,
d
o
i:
1
0
.
1
1
5
9
1
/
ij
e
e
c
s.v
2
0
.
i1
.
p
p
3
4
7
-
3
5
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
22
,
No
.
3
,
J
u
n
e
2
0
2
1
:
1
5
4
8
-
1
5
5
5
1554
[7
]
H.
Kim
,
S
.
L
e
e
,
a
n
d
H.
Ju
n
g
,
“
Hu
m
a
n
a
c
ti
v
it
y
re
c
o
g
n
it
io
n
b
y
u
sin
g
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
E
lec
trica
l
a
n
d
Co
mp
u
ter
E
n
g
i
n
e
e
rin
g
,
v
o
l.
9
,
n
o
.
6
,
p
p
.
5
2
7
0
-
76
,
2
0
1
9
,
d
o
i:
1
0
.
1
1
5
9
1
/
ij
e
c
e
.
v
9
i6
.
p
p
5
2
7
0
-
5
2
7
6
.
[8
]
B.
Kim
,
N.
Yu
v
a
ra
j,
K.
S
ri
P
re
e
t
h
a
a
,
R.
S
a
n
th
o
sh
,
a
n
d
A
.
S
a
b
a
ri,
“
En
h
a
n
c
e
d
p
e
d
e
strian
d
e
tec
ti
o
n
u
sin
g
o
p
t
im
ize
d
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
n
e
u
ra
l
n
e
tw
o
rk
f
o
r
s
m
a
rt
b
u
il
d
i
n
g
su
rv
e
il
lan
c
e
,
”
S
o
ft
Co
m
p
u
t
in
g
,
v
o
l.
2
4
,
n
o
.
2
,
p
p
.
1
-
1
2
,
2
0
2
0
,
d
o
i:
1
0
.
1
0
0
7
/s0
0
5
0
0
-
0
2
0
-
0
4
9
9
9
-
1.
[9
]
F
.
H.
K.
Zam
a
n
,
J.
Jo
h
a
ri,
a
n
d
A
.
I.
M
.
Ya
ss
in
,
“
L
e
a
rn
in
g
F
a
c
e
S
im
il
a
rit
ies
f
o
r
F
a
c
e
V
e
rif
ica
ti
o
n
u
sin
g
Hy
b
rid
Co
n
v
o
l
u
ti
o
n
a
l
Ne
u
ra
l
Ne
tw
o
rk
s,
”
In
d
o
n
e
sia
n
J
o
u
r
n
a
l
o
f
El
e
c
tric
a
l
En
g
i
n
e
e
rin
g
a
n
d
Co
mp
u
ter
S
c
ien
c
e
,
v
o
l.
1
6
,
n
o
.
3
,
p
.
1
3
3
3
-
1
3
4
2
,
2
0
1
9
,
d
o
i:
1
0
.
1
1
5
9
1
/i
jee
c
s.v
1
6
.
i
3
.
p
p
1
3
3
3
-
1
3
4
2
.
[1
0
]
R.
Ya
m
a
sh
it
a
,
M
.
Nish
io
,
R.
K
.
G
.
Do
,
a
n
d
K.
T
o
g
a
sh
i,
“
Co
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s:
a
n
o
v
e
rv
ie
w
a
n
d
a
p
p
li
c
a
ti
o
n
i
n
ra
d
i
o
lo
g
y
,
”
In
sig
h
ts
in
to
im
a
g
i
n
g
,
v
o
l.
9
,
n
o
.
4
,
p
p
.
6
1
1
-
6
2
9
,
2
0
1
8
,
d
o
i:
1
0
.
1
0
0
7
/s1
3
2
4
4
-
01
8
-
0
6
3
9
-
9.
[1
1
]
Z.
Ka
d
im
,
M
.
A
.
Zu
lk
if
le
y
,
a
n
d
N.
Ha
m
z
a
h
,
“
De
e
p
-
lea
rn
in
g
b
a
se
d
sin
g
le
o
b
jec
t
trac
k
e
r
f
o
r
n
ig
h
t
su
rv
e
il
lan
c
e
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
&
Co
mp
u
ter
En
g
in
e
e
ri
n
g
,
v
o
l.
1
0
,
n
o
.
4
,
p
p
.
3
3
5
7
6
-
8
7
,
2
0
2
0
,
d
o
i:
1
0
.
1
1
5
9
1
/
ij
e
c
e
.
v
1
0
i4
.
p
p
3
5
7
6
-
3
5
8
7
.
[1
2
]
N.
O'
M
a
h
o
n
y
e
t
a
l.
,
“
De
e
p
lea
rn
i
n
g
v
s.
trad
it
io
n
a
l
c
o
m
p
u
ter
v
isio
n
,
”
in
S
c
ien
c
e
a
n
d
In
f
o
rm
a
ti
o
n
Co
n
fer
e
n
c
e
,
2
0
1
9
,
p
p
.
1
2
8
-
1
4
4
.
[1
3
]
M
.
H.
A
.
Hijaz
i,
S
.
K.
T
.
H
wa
,
A
.
Ba
d
e
,
R.
Ya
a
k
o
b
,
a
n
d
M
.
S
.
J
e
ff
r
e
e
,
“
En
se
m
b
le
d
e
e
p
lea
rn
in
g
f
o
r
tu
b
e
rc
u
lo
si
s
d
e
tec
ti
o
n
u
si
n
g
c
h
e
st
X
-
ra
y
a
n
d
c
a
n
n
y
e
d
g
e
d
e
tec
ted
im
a
g
e
s,
”
In
t
e
rn
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Arti
fi
c
ia
l
I
n
telli
g
e
n
c
e
,
v
o
l.
8
,
n
o
.
4
,
p
p
.
4
2
9
-
4
3
5
,
2
0
1
9
,
d
o
i:
1
0
.
1
1
5
9
1
/
ij
a
i.
v
8
.
i
4
.
p
p
4
2
9
-
4
3
5
.
[1
4
]
A
.
Dh
o
m
n
e
,
R.
K
u
m
a
r,
a
n
d
V
.
B
h
a
n
,
“
G
e
n
d
e
r
re
c
o
g
n
it
io
n
th
r
o
u
g
h
f
a
c
e
u
sin
g
d
e
e
p
lea
rn
in
g
,
”
Pr
o
c
e
d
ia
C
o
mp
u
ter
S
c
ien
c
e
,
v
o
l.
1
3
2
,
p
p
.
2
-
1
0
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/j
.
p
r
o
c
s.2
0
1
8
.
0
5
.
0
5
3
.
[1
5
]
Z.
Qa
w
a
q
n
e
h
,
A
.
A
.
M
a
ll
o
u
h
,
a
n
d
B.
D.
Ba
rk
a
n
a
,
“
De
e
p
c
o
n
v
o
lu
t
io
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
f
o
r
a
g
e
e
sti
m
a
ti
o
n
b
a
se
d
o
n
V
GG
-
fa
c
e
m
o
d
e
l,
”
a
rXiv p
re
p
rin
t
a
rXiv:1
7
0
9
.
0
1
6
6
4
,
2
0
1
7
.
[1
6
]
M
.
H.
Zak
i
a
n
d
T
.
S
a
y
e
d
,
“
A
u
to
m
a
ted
a
n
a
l
y
sis
o
f
p
e
d
e
strian
s'
n
o
n
c
o
n
f
o
rm
in
g
b
e
h
a
v
io
r
a
n
d
d
a
ta
c
o
ll
e
c
ti
o
n
a
t
a
n
u
rb
a
n
c
r
o
ss
in
g
,
”
T
ra
n
sp
o
rta
ti
o
n
re
se
a
rc
h
re
c
o
rd
,
v
o
l.
2
4
4
3
,
n
o
.
1
,
p
p
.
1
2
3
-
1
3
3
,
2
0
1
4
,
d
o
i:
1
0
.
3
1
4
1
/
2
4
4
3
-
14
.
[1
7
]
D.
Ka
,
D.
L
e
e
,
S
.
Kim
,
a
n
d
H.
Y
e
o
,
“
St
u
d
y
o
n
t
h
e
f
ra
m
e
w
o
rk
o
f
in
ters
e
c
ti
o
n
p
e
d
e
strian
c
o
ll
isio
n
w
a
rn
in
g
sy
ste
m
c
o
n
sid
e
ri
n
g
p
e
d
e
strian
c
h
a
ra
c
teristics
,
”
T
ra
n
sp
o
rta
ti
o
n
re
se
a
rc
h
re
c
o
rd
,
v
o
l.
2
6
7
3
,
n
o
.
5
,
p
p
.
7
4
7
-
7
5
8
,
2
0
1
9
,
d
o
i:
1
0
.
1
1
7
7
/0
3
6
1
1
9
8
1
1
9
8
3
8
5
1
9
.
[1
8
]
N.
W
o
jk
e
,
A
.
Be
w
le
y
,
a
n
d
D.
P
a
u
lu
s,
“
S
im
p
le
o
n
li
n
e
a
n
d
re
a
lt
ime
trac
k
in
g
w
it
h
a
d
e
e
p
a
ss
o
c
iatio
n
m
e
tri
c
,
”
2
0
1
7
IEE
E
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
ima
g
e
p
ro
c
e
ss
in
g
(
ICIP)
,
2
0
1
7
,
p
p
.
3
6
4
5
-
3
6
4
9
,
d
o
i:
1
0
.
1
1
0
9
/ICI
P
.
2
0
1
7
.
8
2
9
6
9
6
2
.
[1
9
]
R.
G
irsh
ick
,
J.
Do
n
a
h
u
e
,
T
.
Da
rre
ll
,
a
n
d
J.
M
a
li
k
,
“
Rich
f
e
a
tu
re
h
iera
rc
h
ies
f
o
r
a
c
c
u
ra
te
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
se
m
a
n
ti
c
se
g
m
e
n
tatio
n
,
”
Pro
c
e
e
d
in
g
s
o
f
th
e
I
EE
E
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
ter
v
isio
n
a
n
d
p
a
tt
e
rn
re
c
o
g
n
it
io
n
,
2
0
1
4
,
p
p
.
5
8
0
-
587
,
d
o
i:
1
0
.
1
1
0
9
/C
VP
R.
2
0
1
4
.
8
1
.
[2
0
]
R.
Ro
t
h
e
,
R.
T
im
o
f
te,
a
n
d
L
.
V
a
n
G
o
o
l,
“
De
x
:
De
e
p
e
x
p
e
c
ta
ti
o
n
o
f
a
p
p
a
re
n
t
a
g
e
f
ro
m
a
sin
g
le
im
a
g
e
,
”
in
Pro
c
e
e
d
in
g
s
o
f
t
h
e
IEE
E
i
n
t
e
rn
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
ter
v
isio
n
wo
rk
sh
o
p
s
,
2
0
1
5
,
p
p
.
1
0
-
15
,
d
o
i:
1
0
.
1
1
0
9
/ICC
V
W
.
2
0
1
5
.
4
1
.
[2
1
]
P
.
V
io
la
a
n
d
M
.
Jo
n
e
s,
“
Ra
p
id
o
b
jec
t
d
e
tec
ti
o
n
u
sin
g
a
b
o
o
ste
d
c
a
sc
a
d
e
o
f
si
m
p
le
f
e
a
tu
re
s,
”
Pr
o
c
e
e
d
in
g
s
o
f
th
e
2
0
0
1
I
EE
E
C
o
mp
u
ter
S
o
c
iety
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
Vi
si
o
n
a
n
d
P
a
tt
e
rn
Rec
o
g
n
it
io
n
.
CVP
R
2
0
0
1
,
Ka
u
a
i,
HI,
USA
,
2
0
0
1
,
d
o
i:
1
0
.
1
1
0
9
/C
VP
R.
2
0
0
1
.
9
9
0
5
1
7
.
[2
2
]
K.
He
,
X
.
Z
h
a
n
g
,
S
.
Re
n
,
a
n
d
J.
S
u
n
,
“
De
e
p
re
sid
u
a
l
lea
rn
in
g
f
o
r
im
a
g
e
re
c
o
g
n
it
io
n
,
”
Pr
o
c
e
e
d
in
g
s
o
f
t
h
e
IEE
E
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
ter
v
isio
n
a
n
d
p
a
tt
e
rn
re
c
o
g
n
it
io
n
,
2
0
1
6
,
p
p
.
7
7
0
-
7
7
8
,
d
o
i:
1
0
.
1
1
0
9
/C
VP
R.
2
0
1
6
.
9
0
.
[2
3
]
O.
M
.
P
a
rk
h
i,
A
.
V
e
d
a
ld
i,
a
n
d
A
.
Zi
ss
e
r
m
a
n
,
“
De
e
p
f
a
c
e
re
c
o
g
n
it
io
n
,
”
Visu
a
l
G
e
o
m
e
tr
y
G
ro
u
p
De
p
a
rtm
e
n
t
o
f
En
g
in
e
e
rin
g
S
c
ien
c
e
Un
iv
e
r
sity
o
f
Ox
f
o
rd
,
2
0
1
5
.
[2
4
]
A
.
A
.
M
o
u
sta
f
a
,
A
.
El
n
a
k
ib
,
a
n
d
N.
F
.
A
re
e
d
,
“
Op
ti
m
iza
ti
o
n
o
f
d
e
e
p
lea
rn
in
g
fe
a
tu
re
s
f
o
r
a
g
e
-
in
v
a
rian
t
f
a
c
e
re
c
o
g
n
it
io
n
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
&
Co
mp
u
ter
E
n
g
i
n
e
e
rin
g
,
v
o
l.
1
0
,
n
o
.
2
,
p
p
.
1
8
3
3
-
4
1
,
2
0
2
0
,
d
o
i:
1
0
.
1
1
5
9
1
/
ij
e
c
e
.
v
1
0
i2
.
p
p
1
8
3
3
-
1
8
4
1
.
[2
5
]
Q.
Ca
o
,
L
.
S
h
e
n
,
W
.
X
ie,
O.
M
.
P
a
rk
h
i,
a
n
d
A
.
Zi
ss
e
r
m
a
n
,
“
V
g
g
fa
c
e
2
:
A
d
a
tas
e
t
f
o
r
re
c
o
g
n
isin
g
f
a
c
e
s
a
c
ro
ss
p
o
se
a
n
d
a
g
e
,
”
2
0
1
8
1
3
th
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
A
u
to
m
a
ti
c
Fa
c
e
&
Ge
stu
re
Rec
o
g
n
it
io
n
(
FG
2
0
1
8
)
,
2
0
1
8
,
p
p
.
6
7
-
74
,
d
o
i
:
1
0
.
1
1
0
9
/F
G
.
2
0
1
8
.
0
0
0
2
0
.
[2
6
]
O.
Ru
ss
a
k
o
v
sk
y
e
t
a
l.
,
“
Im
a
g
e
n
e
t
larg
e
sc
a
le
v
isu
a
l
re
c
o
g
n
it
io
n
c
h
a
ll
e
n
g
e
,
”
In
ter
n
a
ti
o
n
a
l
jo
u
rn
a
l
o
f
c
o
mp
u
ter
v
isio
n
,
v
o
l
.
1
1
5
,
n
o
.
3
,
p
p
.
2
1
1
-
2
5
2
,
2
0
1
5
.
[2
7
]
H.
S
o
f
ian
,
J.
T
.
C.
M
in
g
,
S
.
M
u
h
a
m
m
a
d
,
a
n
d
N.
M
.
No
o
r,
“
C
a
lcif
ic
a
ti
o
n
d
e
tec
ti
o
n
u
si
n
g
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
a
rc
h
it
e
c
tu
re
s
in
i
n
trav
a
sc
u
lar
u
lt
ra
so
u
n
d
im
a
g
e
s,
”
In
d
o
n
e
sia
n
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
En
g
in
e
e
rin
g
a
n
d
Co
mp
u
ter
S
c
ien
c
e
,
v
o
l.
1
7
,
n
o
.
3
,
p
p
.
1
3
1
3
-
1
3
2
1
,
2
0
2
0
,
d
o
i:
1
0
.
1
1
5
9
1
/i
jee
c
s.v
1
7
.
i3
.
p
p
1
3
1
3
-
1
3
2
1
.
[2
8
]
R.
F
.
Ra
h
m
a
t,
O
.
S
.
S
it
o
m
p
u
l
,
S
.
P
u
rn
a
m
a
w
a
ti
,
a
n
d
R.
Bu
d
iarto
,
“
A
d
v
e
rti
se
m
e
n
t
b
il
l
b
o
a
rd
d
e
tec
ti
o
n
a
n
d
g
e
o
tag
g
in
g
s
y
ste
m
w
it
h
in
d
u
c
ti
v
e
tran
sfe
r
lea
rn
in
g
in
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
,
”
T
EL
KOM
NIKA
T
e
lec
o
mm
u
n
ica
ti
o
n
C
o
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
.
5
,
p
p
.
2
6
5
9
-
2
6
6
6
,
2
0
1
9
,
d
o
i:
1
0
.
1
2
9
2
8
/t
e
lk
o
m
n
ik
a
.
v
1
7
i5
.
1
1
2
7
6
.
[2
9
]
M.
G
h
a
z
a
l,
N.
W
a
isi,
a
n
d
N.
A
b
d
u
ll
a
h
,
“
T
h
e
d
e
tec
ti
o
n
o
f
h
a
n
d
g
u
n
s
f
ro
m
li
v
e
-
v
id
e
o
in
re
a
l
-
ti
m
e
b
a
se
d
o
n
d
e
e
p
lea
rn
in
g
,”
T
EL
KOM
NIKA
T
e
lec
o
mm
u
n
ica
ti
o
n
C
o
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
.
6
,
p
p
.
3
0
2
6
-
3
0
3
2
,
2
0
2
0
,
d
o
i:
1
0
.
1
2
9
2
8
/t
e
lk
o
m
n
ik
a
.
v
1
8
i
6
.
1
6
1
7
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
P
ed
estr
ia
n
a
g
e
esti
ma
tio
n
b
a
s
ed
o
n
d
ee
p
lea
r
n
in
g
(
N
a
w
a
l Y
o
u
n
is
A
b
d
u
lla
h
)
1555
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Na
w
a
l
Yo
u
n
is
Abd
u
ll
a
h
o
b
tain
e
d
h
e
r
M
.
S
c
.
d
e
g
re
e
f
ro
m
Co
m
p
u
ter
En
g
in
e
e
rin
g
T
e
c
h
n
o
lo
g
y
,
No
rth
e
rn
T
e
c
h
n
ica
l
Un
iv
e
rsity
,
M
o
su
l
,
Ira
q
in
2
0
1
4
.
He
r
M
.
S
c
.
t
h
e
sis
e
n
ti
tl
e
d
:
"
F
P
GA
Ba
s
e
d
V
id
e
o
S
c
e
n
e
Bo
u
n
d
a
ries
De
tec
ti
o
n
Us
in
g
E
n
h
a
n
c
e
d
S
o
b
e
l
F
il
ter
"
a
n
d
h
e
r
c
u
rre
n
t
re
se
a
rc
h
in
tere
sts in
c
lu
d
e
th
e
d
e
e
p
lea
rn
i
n
g
CNN
s
m
o
d
e
ls an
d
o
b
jec
t
d
e
tec
t
io
n
.
M
o
h
a
m
m
e
d
Ta
la
l
G
h
a
z
a
l
o
b
tain
e
d
h
is
M
.
S
c
.
d
e
g
re
e
f
ro
m
Co
m
p
u
ter
En
g
in
e
e
rin
g
T
e
c
h
n
o
lo
g
y
,
No
rth
e
rn
T
e
c
h
n
ica
l
Un
iv
e
rsity
,
M
o
su
l,
Ira
q
in
2
0
1
6
.
His
M
.
S
c
.
t
h
e
sis
e
n
ti
tl
e
d
:
"
W
h
e
e
l
c
h
a
ir
Ro
b
o
t
Co
n
tro
l
Us
in
g
EOG
si
g
n
a
ls"
.
His
re
s
e
a
rc
h
in
tere
sts
in
c
lu
d
e
th
e
d
e
sig
n
o
f
f
a
c
e
re
c
o
g
n
it
io
n
a
lg
o
rit
h
m
s,
d
e
e
p
lea
rn
in
g
CNN
s m
o
d
e
ls an
d
o
b
jec
t
d
e
tec
ti
o
n
.
Na
jw
a
n
Zu
h
a
ir
W
a
isi
o
b
tain
e
d
h
e
r
M
.
S
c
.
d
e
g
re
e
f
ro
m
Co
m
p
u
ter
S
c
ien
c
e
De
p
a
rtme
n
t,
M
o
u
sl
Un
iv
e
rsit
y
,
M
o
su
l,
Ira
q
i
n
2
0
1
4
.
He
r
M
.
S
c
.
th
e
sis
e
n
ti
tl
e
d
:
"
De
sig
n
a
n
d
Im
p
le
m
e
n
tatio
n
o
f
Cli
e
n
t
Ho
n
e
y
p
o
t"
a
n
d
h
e
r
c
u
rre
n
t
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
t
h
e
d
e
e
p
lea
rn
in
g
CNN
s
m
o
d
e
ls
a
n
d
o
b
jec
t
d
e
tec
ti
o
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.