I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
3
9
,
No
.
1
,
Ju
ly
2
0
2
5
,
p
p
.
634
~
6
4
3
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
3
9
.i
1
.
pp
634
-
6
4
3
634
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
Lo
w
-
reso
lution i
ma
g
e quality
enh
a
ncement
using
e
nha
nced
super
-
reso
lu
tion
co
nv
o
lutiona
l net
wo
rk and su
pe
r
-
r
eso
lution
residua
l net
wo
rk
M
o
ha
m
m
a
d F
a
is
a
l R
if
t
ia
rr
a
s
y
id
1
,
Rico
H
a
lim
1
,
Andi
en
Dwi
No
v
ik
a
1
,
A
m
a
lia
Z
a
hra
2
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
S
c
h
o
o
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
B
i
n
a
N
u
sa
n
t
a
r
a
U
n
i
v
e
r
si
t
y
,
J
a
k
a
r
t
a
,
I
n
d
o
n
e
s
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
B
I
N
U
S
G
r
a
d
u
a
t
e
P
r
o
g
r
a
m
-
M
a
st
e
r
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
B
i
n
a
N
u
s
a
n
t
a
r
a
U
n
i
v
e
r
si
t
y
,
Jak
a
r
t
a
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
May
11
,
2
0
2
4
R
ev
is
ed
Mar
12
,
2
0
2
5
Acc
ep
ted
Mar
26
,
2
0
2
5
Th
is
re
se
a
rc
h
e
x
p
lo
re
s
th
e
i
n
teg
ra
ti
o
n
o
f
e
n
h
a
n
c
e
d
su
p
e
r
-
re
so
lu
ti
o
n
c
o
n
v
o
lu
ti
o
n
a
l
n
e
two
r
k
(ES
P
CN
)
a
n
d
su
p
e
r
-
re
so
lu
t
io
n
re
sid
u
a
l
n
e
two
r
k
(S
RRe
sN
e
t)
to
e
n
h
a
n
c
e
ima
g
e
q
u
a
li
t
y
c
a
p
tu
re
d
b
y
l
o
w
-
re
so
lu
t
io
n
(
LR)
c
a
m
e
ra
s
a
n
d
in
i
n
tern
e
t
o
f
th
in
g
s
(Io
T)
d
e
v
ice
s.
F
o
c
u
sin
g
o
n
f
a
c
e
m
a
sk
p
re
d
ictio
n
m
o
d
e
ls,
th
e
st
u
d
y
a
c
h
i
e
v
e
s
a
su
b
sta
n
ti
a
l
imp
r
o
v
e
m
e
n
t,
a
tt
a
in
in
g
a
p
e
a
k
s
ig
n
a
l
-
to
-
n
o
ise
ra
ti
o
(P
S
NR)
o
f
2
8
.
5
1
4
2
d
B
a
n
d
a
n
e
x
e
c
u
ti
o
n
ti
m
e
o
f
0
.
3
4
7
0
4
6
3
8
se
c
o
n
d
s.
Th
e
in
t
e
g
ra
ti
o
n
o
f
su
p
e
r
-
re
so
lu
t
io
n
t
e
c
h
n
iq
u
e
s
sig
n
ifi
c
a
n
t
ly
b
o
o
sts
th
e
v
is
u
a
l
g
e
o
m
e
try
g
r
o
u
p
-
1
6
(
VG
G
1
6
)
m
o
d
e
l
’
s
p
e
rfo
rm
a
n
c
e
,
e
lev
a
ti
n
g
c
las
sifica
ti
o
n
a
c
c
u
ra
c
y
fro
m
7
1
.
3
0
%
t
o
9
6
.
3
0
%
.
Th
e
se
fin
d
in
g
s
h
i
g
h
l
ig
h
t
t
h
e
p
o
ten
ti
a
l
o
f
su
p
e
r
-
re
so
lu
ti
o
n
i
n
o
p
ti
m
izi
n
g
ima
g
e
q
u
a
li
ty
f
o
r
l
o
w
-
p
e
rfo
r
m
a
n
c
e
d
e
v
ice
s
a
n
d
e
n
c
o
u
ra
g
e
fu
rth
e
r
e
x
p
lo
ra
ti
o
n
a
c
ro
ss
d
i
v
e
rse
a
p
p
li
c
a
ti
o
n
s
in
ima
g
e
p
ro
c
e
ss
in
g
a
n
d
p
a
tt
e
rn
re
c
o
g
n
it
i
o
n
with
i
n
I
o
T
a
n
d
b
e
y
o
n
d
.
K
ey
w
o
r
d
s
:
E
SP
C
N
I
m
ag
e
p
r
o
ce
s
s
in
g
SR
R
esNet
Su
p
er
r
eso
lu
tio
n
VGG1
6
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
:
An
d
ien
Dwi
No
v
ik
a
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
,
B
I
NUS
Gr
ad
u
ate
Pro
g
r
a
m
-
Ma
s
ter
o
f
C
o
m
p
u
ter
Scien
ce
B
in
a
Nu
s
an
tar
a
Un
iv
er
s
ity
J
ak
ar
ta
,
1
1
4
8
0
,
I
n
d
o
n
esia
E
m
ail: a
n
d
ien
.
n
o
v
ik
a@
b
in
u
s
.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
I
n
r
ec
e
n
t
y
ea
r
s
,
in
ter
n
et
o
f
th
i
n
g
s
(
I
o
T
)
h
as
b
ec
o
m
e
a
h
o
t
to
p
ic
f
o
r
r
esear
c
h
in
in
d
u
s
tr
ies,
esp
ec
ially
to
p
ics
o
n
i
n
teg
r
atio
n
o
f
d
ee
p
l
ea
r
n
in
g
im
ag
e
class
if
icatio
n
o
r
d
etec
tio
n
in
I
o
T
tec
h
n
o
lo
g
ie
s
[
1
]
.
W
ith
in
t
h
o
s
e
r
esear
ch
,
ca
m
er
a
will
alwa
y
s
b
e
p
r
esen
t
as
th
e
m
ain
s
en
s
o
r
to
g
ath
er
in
p
u
t
f
o
r
th
e
s
y
s
tem
.
Mo
d
er
n
ca
m
er
a
wh
ich
is
b
ein
g
u
tili
ze
d
in
I
o
T
s
y
s
tem
wil
l
ca
p
tu
r
e
lig
h
t
wh
ich
is
g
o
in
g
to
war
d
s
it,
th
en
co
n
s
tr
u
ct
a
d
ig
ital
p
ictu
r
e
f
r
o
m
th
e
lig
h
t
d
ata
r
ec
eiv
ed
b
y
th
e
s
en
s
o
r
.
C
am
e
r
a
will
p
er
ce
iv
e
d
ata
s
u
c
h
as
co
lo
r
in
f
o
r
m
at
io
n
,
s
atu
r
atio
n
,
an
d
b
r
ig
h
tn
ess
f
r
o
m
in
co
m
in
g
lig
h
t
a
n
d
th
en
ass
ig
n
th
o
s
e
d
ata
in
to
ea
ch
tin
y
p
ar
t
with
in
a
p
ictu
r
e
ca
lled
p
ix
el.
T
h
o
s
e
s
et
o
f
p
ix
e
ls
ar
e
g
o
in
g
to
b
e
ass
em
b
led
in
to
o
n
e
an
d
p
r
o
jecte
d
as a
p
ict
u
r
e.
T
h
e
m
o
r
e
p
ix
el
with
in
a
p
ictu
r
e,
th
e
s
h
ar
p
e
r
th
at
p
ictu
r
e
g
ets
o
r
it
ca
n
also
b
e
s
aid
th
at
th
e
p
ictu
r
e
co
n
tain
s
m
o
r
e
in
f
o
r
m
atio
n
.
T
h
e
am
o
u
n
t
o
f
p
ix
el
ca
n
b
e
ca
p
tu
r
ed
b
y
ca
m
er
a
m
ea
n
s
m
o
r
e
in
f
o
r
m
atio
n
ca
n
b
e
s
to
r
e
d
with
in
a
p
ictu
r
e.
I
t m
ea
n
s
,
lar
g
er
r
eso
lu
tio
n
im
a
g
e
co
n
tain
s
m
o
r
e
in
f
o
r
m
atio
n
wh
ich
c
an
b
e
p
r
o
ce
s
s
ed
f
u
r
th
e
r
.
B
ef
o
r
e
u
s
in
g
ca
m
er
a
as
an
in
p
u
t,
o
n
es
m
u
s
t
co
n
s
id
er
its
l
im
i
tatio
n
an
d
ca
p
ab
ilit
ies.
E
ac
h
ca
m
er
a
h
as
th
eir
o
wn
lim
itatio
n
s
,
wh
ich
is
th
e
am
o
u
n
t
o
f
p
ix
el
th
at
co
u
l
d
b
e
ca
p
tu
r
e
d
in
a
s
in
g
le
p
ictu
r
e
o
r
th
e
am
o
u
n
t
o
f
p
o
wer
n
ee
d
e
d
to
o
p
e
r
ate
it
[
2
]
.
C
am
er
a
ca
p
ab
ilit
y
is
o
f
ten
s
h
o
wn
in
th
e
am
o
u
n
t
o
f
m
eg
a
p
ix
el
o
f
in
f
o
r
m
atio
n
ca
n
h
an
d
led
.
E
v
en
th
o
u
g
h
m
o
r
e
p
ix
el
s
ee
m
s
lik
e
th
e
b
est
o
p
tio
n
,
it
n
ee
d
s
s
ig
n
if
ican
tly
lar
g
er
s
p
ac
e
to
s
to
r
e
th
e
im
ag
e
an
d
it
n
ee
d
s
m
o
r
e
c
o
m
p
u
tin
g
p
o
wer
to
f
u
r
th
er
p
r
o
ce
s
s
it.
I
n
r
ec
en
t
s
tu
d
ies
wh
i
ch
u
tili
ze
ca
m
er
a
as
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Lo
w
-
r
eso
lu
tio
n
ima
g
e
q
u
a
lity
en
h
a
n
ce
me
n
t u
s
in
g
en
h
a
n
ce
d
…
(
Mo
h
a
mma
d
F
a
is
a
l R
ifti
a
r
r
a
s
yid
)
635
th
e
m
ain
s
o
u
r
ce
o
f
in
p
u
t,
r
e
s
ea
r
ch
er
f
o
cu
s
es
o
n
im
p
r
o
v
i
n
g
p
er
f
o
r
m
an
ce
o
r
ef
f
icien
c
y
o
f
th
eir
p
r
o
t
o
ty
p
e.
I
n
o
r
d
er
to
im
p
r
o
v
e
p
er
f
o
r
m
an
ce
an
d
ef
f
icien
cy
,
r
ec
e
n
t
s
tu
d
ies
s
u
g
g
est
lo
wer
in
g
th
e
i
m
ag
e
r
eso
lu
tio
n
f
o
r
o
b
ject
d
etec
tio
n
[
3
]
.
B
y
u
s
in
g
m
in
im
izin
g
r
eso
lu
tio
n
,
r
esear
ch
er
aim
s
to
allo
ca
te
co
m
p
u
ti
n
g
p
o
wer
to
o
t
h
er
p
r
o
ce
s
s
es
lik
e
d
ee
p
lear
n
in
g
[
4
]
o
r
m
in
im
ize
b
atter
y
co
n
s
u
m
p
tio
n
o
f
r
e
m
o
te
co
n
tr
o
lled
d
ev
ices
lik
e
d
r
o
n
es
[
5
]
.
Fu
r
th
er
m
o
r
e
b
y
u
s
in
g
lo
w
er
r
eso
l
u
tio
n
(
L
R
)
,
m
in
im
izin
g
ch
an
ce
s
o
f
u
n
wan
ted
i
n
f
o
r
m
atio
n
to
ap
p
ea
r
[
6
]
.
Oth
er
th
in
g
s
to
c
o
n
s
id
er
is
th
e
co
s
t
to
b
u
y
ca
m
er
a
wh
ic
h
ca
p
tu
r
e
h
ig
h
r
eso
lu
tio
n
(
HR
)
im
ag
e
is
s
ig
n
if
ican
tly
m
o
r
e
ex
p
en
s
iv
e,
s
o
p
e
o
p
le
n
ee
d
s
to
co
n
s
id
er
wh
ic
h
ca
m
er
a
i
s
b
est s
u
ited
f
o
r
i
ts
in
ten
d
ed
f
u
n
ctio
n
ality
.
Mo
d
er
n
ar
tific
ial
in
tellig
en
ce
b
r
ea
k
th
r
o
u
g
h
o
p
en
its
im
ag
e
g
en
er
atin
g
ab
ilit
ies,
in
wh
ich
o
p
en
s
m
o
r
e
p
o
s
s
ib
ilit
ies
to
tr
ain
d
ee
p
lear
n
in
g
m
o
d
el
to
g
e
n
er
ate
im
ag
e
with
lar
g
er
r
eso
lu
tio
n
f
r
o
m
its
o
r
ig
in
al
r
eso
lu
tio
n
[
7
]
.
Su
p
er
r
eso
lu
tio
n
im
a
g
e
e
n
h
an
ce
m
en
t
aim
s
to
ex
p
a
n
d
im
ag
e
r
eso
lu
tio
n
a
n
d
en
r
ich
its
i
n
f
o
r
m
atio
n
c
o
n
ten
t
b
ased
o
n
th
e
o
r
ig
in
al
im
ag
e.
Su
p
er
r
eso
lu
tio
n
will
ex
p
a
n
d
th
e
r
eso
lu
tio
n
s
ize
o
f
th
e
o
r
i
g
in
al
im
ag
e
b
y
a
n
u
p
s
ca
le
f
ac
to
r
th
en
p
r
e
d
ict
th
e
ex
p
an
d
e
d
p
ix
els
b
ased
f
r
o
m
in
f
o
r
m
atio
n
o
f
th
e
ad
jace
n
t
o
r
ig
in
al
p
ix
el.
So
m
e
s
tu
d
ies
in
d
icate
s
th
at
b
y
im
p
r
o
v
in
g
q
u
ality
o
f
im
ag
es
h
as
a
s
ig
n
if
ican
t
im
p
ac
t
in
d
ee
p
lear
n
in
g
p
r
ed
ictio
n
r
esu
lts
[
8
]
,
[
9
]
.
I
t
m
ea
n
s
b
y
u
tili
zin
g
s
u
p
er
-
r
eso
lu
tio
n
to
p
r
e
-
p
r
o
ce
s
s
im
ag
e
co
u
l
d
im
p
r
o
v
e
p
er
f
o
r
m
an
ce
o
f
d
eep
lear
n
in
g
in
teg
r
ate
d
s
y
s
t
em
.
T
h
is
is
v
er
y
b
e
n
ef
icial,
esp
ec
ially
with
lo
w
p
er
f
o
r
m
a
n
ce
m
ac
h
i
n
e
f
r
o
m
p
r
ev
io
u
s
g
en
e
r
atio
n
.
B
y
im
p
r
o
v
in
g
s
o
f
twar
e
with
o
u
t a
n
y
n
ee
d
s
to
im
p
r
o
v
e
its
h
ar
d
war
e
s
u
ch
as c
am
er
a
as th
e
m
ain
s
en
s
o
r
s
o
f
its
s
y
s
tem
wh
ich
ca
n
b
e
v
er
y
ex
p
en
s
iv
e
to
u
p
g
r
ad
e.
Nev
e
r
th
eless
,
th
er
e
is
s
til
l
y
et
an
y
r
esear
ch
u
tili
ze
en
h
an
ce
d
s
u
p
er
-
r
eso
lu
tio
n
co
n
v
o
lu
tio
n
al
n
etwo
r
k
(
E
SP
C
N)
an
d
s
u
p
er
-
r
eso
lu
tio
n
r
esid
u
al
n
etwo
r
k
(
SR
R
esNet)
s
u
p
er
r
eso
lu
tio
n
to
p
r
e
-
p
r
o
ce
s
s
im
ag
e
u
s
in
g
lo
w
p
er
f
o
r
m
an
ce
d
e
v
ices
lik
e
I
o
T
d
ev
ices.
T
h
is
r
esear
ch
tr
ies to
ex
p
lo
r
e
t
h
e
id
ea
o
f
im
p
r
o
v
i
n
g
im
ag
e
ca
p
tu
r
ed
b
y
lo
w
p
er
f
o
r
m
a
n
ce
d
e
v
ices b
y
u
s
in
g
lo
w
r
eso
lu
tio
n
ca
m
er
a
with
s
u
p
er
r
eso
lu
tio
n
h
as
s
ig
n
if
ican
t
im
p
ac
t
in
class
if
y
in
g
m
ask
p
r
ed
ictio
n
d
ee
p
lear
n
in
g
m
o
d
el
(
v
is
u
al
g
e
o
m
etr
y
g
r
o
u
p
-
1
6
(
VGG1
6
)
,
Mo
b
ileNetV2
,
an
d
R
es
N
et
5
0
)
.
T
h
is
r
esear
ch
tr
ies
to
ex
p
l
o
r
e
r
esear
ch
es
wh
ich
h
as
r
elate
d
to
p
ics.
T
h
is
is
d
o
n
e
b
y
s
ea
r
ch
in
g
th
e
Go
o
g
le
Sch
o
lar
p
ag
e
with
th
e
k
ey
wo
r
d
s
“
s
u
p
e
r
r
eso
lu
tio
n
d
ee
p
lea
r
n
in
g
”
,
“
SR
R
es
N
et
”
,
“
E
SP
C
N
”
an
d
“
s
u
p
e
r
r
eso
l
u
tio
n
d
ee
p
lear
n
in
g
class
if
icatio
n
”
wh
ich
wer
e
p
u
b
lis
h
ed
b
etwe
e
n
2
0
1
9
a
n
d
2
0
2
3
.
I
n
o
r
d
er
to
m
ain
tain
th
e
r
elev
an
ce
o
f
t
h
e
r
esear
ch
to
p
ic
an
d
f
ill
g
ap
s
in
ex
is
tin
g
r
esear
ch
.
Stu
d
y
o
n
s
in
g
le
im
ag
e
s
u
p
er
-
r
eso
lu
tio
n
(
SISR
)
was
f
ir
s
t
co
n
d
u
cted
b
y
[
1
0
]
.
T
h
e
p
r
esen
t
s
tu
d
y
em
p
lo
y
s
s
u
p
er
-
r
eso
lu
tio
n
g
e
n
er
ativ
e
a
d
v
er
s
ar
i
al
n
etwo
r
k
(
SR
GAN
)
an
d
SR
R
esNet
to
p
r
o
d
u
ce
s
u
p
er
-
r
eso
lu
tio
n
im
ag
es
(
SI
SR
)
f
o
r
laser
co
n
f
o
ca
l
im
a
g
es
o
f
th
e
r
o
o
t
ce
lls
o
f
So
lan
u
m
n
ig
r
u
m
,
a
h
y
p
er
ac
c
u
m
u
lato
r
[
1
1
]
in
t
h
e
y
ea
r
2
0
2
3
.
T
h
e
ev
al
u
atio
n
m
e
th
o
d
s
em
p
lo
y
e
d
in
th
is
s
tu
d
y
ar
e
p
ea
k
s
ig
n
al
-
to
-
n
o
is
e
r
atio
(
PS
NR
)
,
s
tr
u
ctu
r
al
s
im
ilar
ity
in
d
ex
(
SS
I
M)
,
an
d
m
ea
n
o
p
in
io
n
s
co
r
e
(
MO
S).
T
h
e
f
in
d
in
g
s
o
f
th
i
s
wo
r
k
in
d
icate
th
at
b
o
th
t
h
e
p
r
im
ar
y
r
ec
o
n
s
tr
u
ctio
n
an
d
th
e
s
u
b
s
eq
u
en
t
r
ec
o
n
s
tr
u
cti
o
n
o
f
SR
GAN
an
d
SR
R
esNet
h
av
e
d
em
o
n
s
tr
ated
en
h
an
ce
d
r
eso
lu
tio
n
ca
p
ab
ilit
ies f
o
r
laser
co
n
f
o
ca
l p
ictu
r
es.
I
n
o
r
d
er
to
en
h
a
n
ce
th
e
in
tr
ic
ate
f
ea
tu
r
es
o
f
tex
tu
r
e,
th
e
au
t
h
o
r
s
o
f
[
1
2
]
p
r
o
p
o
s
e
th
e
em
p
l
o
y
m
en
t
o
f
SR
GAN,
a
f
r
am
ewo
r
k
th
at
d
em
o
n
s
tr
ates th
e
ab
ilit
y
to
g
en
er
ate
p
h
o
to
-
r
ea
lis
tic
n
atu
r
al
im
a
g
es with
a
f
o
u
r
-
f
o
ld
in
cr
ea
s
e
in
r
eso
lu
tio
n
.
T
h
e
SR
GAN
m
o
d
el
em
p
lo
y
s
a
f
r
a
m
ewo
r
k
b
ased
o
n
g
en
e
r
ativ
e
ad
v
er
s
ar
ial
n
etwo
r
k
s
(
GANs)
,
co
m
p
r
is
in
g
a
g
en
er
a
to
r
n
etwo
r
k
an
d
a
d
is
cr
im
in
ato
r
n
etwo
r
k
.
T
h
e
g
en
e
r
ato
r
n
e
two
r
k
is
tr
ain
ed
to
p
r
o
d
u
ce
HR
im
ag
es
th
at
e
x
h
ib
it
v
is
u
al
s
im
ilar
ity
to
t
h
e
g
r
o
u
n
d
tr
u
th
HR
p
h
o
to
s
.
C
o
n
v
er
s
ely
,
th
e
d
is
cr
im
in
ato
r
n
etwo
r
k
is
tr
ai
n
ed
to
d
is
ce
r
n
b
etwe
en
th
e
g
en
er
ated
im
ag
es
a
n
d
th
e
a
u
th
en
t
ic
HR
p
h
o
to
g
r
ap
h
s
[
1
3
]
.
T
h
r
o
u
g
h
th
e
u
tili
za
tio
n
o
f
a
p
e
r
ce
p
tu
al
lo
s
s
f
u
n
ctio
n
th
at
am
alg
am
ates
an
a
d
v
er
s
ar
ial
lo
s
s
an
d
a
c
o
n
ten
t
lo
s
s
,
SR
GA
N
d
em
o
n
s
tr
ates
t
h
e
ca
p
a
b
ilit
y
to
g
en
er
ate
s
u
p
er
-
r
eso
lv
ed
im
ag
es
th
at
ex
h
i
b
it
n
o
t
o
n
ly
elev
ated
PS
NR
,
b
u
t
also
ef
f
ec
tiv
ely
c
ap
tu
r
e
in
tr
icate
tex
tu
r
e
f
ea
tu
r
es,
r
esu
ltin
g
in
v
is
u
ally
a
p
p
ea
lin
g
o
u
t
p
u
ts
f
o
r
h
u
m
an
o
b
s
er
v
e
r
s
.
T
h
e
e
v
alu
at
io
n
m
eth
o
d
s
em
p
lo
y
ed
i
n
th
is
s
tu
d
y
ar
e
PS
NR
,
SS
I
M,
an
d
MO
S.
T
h
e
f
in
d
i
n
g
s
o
f
[
1
2
]
an
d
[
1
4
]
o
n
t
h
e
a
p
p
licatio
n
o
f
SR
GAN
in
p
ictu
r
e
s
u
p
e
r
-
r
eso
lu
tio
n
d
em
o
n
s
t
r
ated
th
at
S
R
GAN
e
x
h
i
b
i
t
e
d
s
u
p
e
r
i
o
r
p
e
r
f
o
r
m
a
n
c
e
c
o
m
p
a
r
e
d
t
o
o
t
h
e
r
a
d
v
a
n
c
e
d
t
e
c
h
n
i
q
u
e
s
,
a
s
d
e
t
e
r
m
i
n
e
d
t
h
r
o
u
g
h
t
h
e
u
t
i
l
i
z
a
ti
o
n
o
f
t
h
e
M
O
S
e
v
a
l
u
a
ti
o
n
m
e
t
h
o
d
,
s
p
e
c
i
f
i
c
a
ll
y
i
n
t
e
r
m
s
o
f
p
e
r
c
e
p
tu
a
l
q
u
a
l
i
t
y
.
I
n
t
e
r
m
s
o
f
q
u
a
n
t
i
ta
t
i
v
e
p
e
r
f
o
r
m
a
n
c
e
,
S
R
R
es
N
et
e
x
h
i
b
it
s
s
u
p
e
r
i
o
r
r
e
s
u
l
t
s
i
n
c
o
m
p
a
r
is
o
n
t
o
SR
GA
N
,
a
s
m
e
as
u
r
e
d
b
y
m
et
r
i
cs
s
u
c
h
as
P
S
NR
a
n
d
S
S
I
M.
F
u
r
t
h
e
r
m
o
r
e
,
R
es
N
et
s
h
a
v
e
b
ee
n
s
h
o
w
n
t
o
g
e
n
e
r
a
t
e
s
h
a
r
p
e
r
s
u
p
e
r
r
e
s
o
l
u
t
i
o
n
i
m
a
g
e
o
u
t
p
u
t
[
1
5
]
.
E
SP
C
N
wa
s
p
r
o
p
o
s
ed
b
y
[
1
6
]
,
wh
er
e
t
h
e
n
etwo
r
k
p
r
im
ar
i
ly
f
o
cu
s
es
o
n
en
h
an
cin
g
th
e
r
eso
lu
t
io
n
f
r
o
m
L
R
to
HR
in
th
e
latter
s
tag
es
o
f
th
e
n
etwo
r
k
.
I
t
ac
h
iev
es
th
is
b
y
s
u
p
er
-
r
eso
lv
in
g
HR
d
ata
u
s
in
g
L
R
f
ea
tu
r
e
m
ap
s
.
T
h
is
o
b
v
iates
th
e
n
ec
ess
ity
o
f
d
o
in
g
th
e
m
aj
o
r
ity
o
f
t
h
e
SR
o
p
er
atio
n
with
in
th
e
s
ig
n
if
ican
tl
y
HR
.
T
h
e
p
r
esen
t
wo
r
k
em
p
lo
y
s
th
e
E
SP
C
N
tech
n
iq
u
e
f
o
r
th
e
p
u
r
p
o
s
e
o
f
r
ea
l
-
tim
e
s
u
p
er
-
r
eso
lu
tio
n
o
f
1
0
8
0
p
f
ilm
s
.
T
h
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
in
d
icate
th
at
th
e
E
SP
C
N
m
o
d
el
o
u
tp
e
r
f
o
r
m
s
th
e
SR
C
NN
m
o
d
el
in
ter
m
s
o
f
p
er
f
o
r
m
an
ce
.
E
SP
C
N
ex
ce
l
in
r
u
n
n
in
g
tim
es
with
d
ec
en
t
PS
NR
im
ag
e
o
u
tp
u
t
[
1
7
]
,
th
is
m
ea
n
s
th
at
E
SP
C
N
i
s
b
est s
u
ited
to
r
ea
l
-
tim
e
s
u
p
er
r
eso
lu
tio
n
p
r
o
ce
s
s
in
g
th
at
f
o
cu
s
es o
n
p
r
o
ce
s
s
in
g
s
p
ee
d
.
2.
M
E
T
H
O
D
T
h
e
o
b
jectiv
e
o
f
th
is
s
tu
d
y
is
to
d
eter
m
in
e
th
e
p
r
o
ce
s
s
o
f
g
en
er
ati
n
g
a
HR
im
ag
e
b
ased
o
n
a
p
r
o
v
id
e
d
L
R
im
ag
e.
T
h
e
in
itial
s
tag
e
o
f
th
is
r
esear
ch
in
v
o
lv
es
p
r
e
-
p
r
o
ce
s
s
in
g
,
s
p
ec
if
ically
th
e
r
ed
u
ctio
n
o
f
th
e
HR
im
ag
e
to
cr
ea
te
th
e
L
R
im
ag
e.
T
h
e
tr
ain
in
g
ap
p
r
o
a
ch
u
tili
ze
d
two
d
atasets
,
n
am
e
ly
B
SD5
0
0
an
d
6
0
0
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
634
-
6
4
3
636
web
s
cr
ap
p
ed
p
h
o
to
s
.
T
h
e
E
SP
C
N
-
SR
R
esNe
t
m
o
d
el
will
u
tili
ze
L
R
im
ag
es
a
s
it
s
i
n
p
u
t.
T
h
e
E
SP
C
N
-
SR
R
esNet
m
o
d
el
is
tr
ain
ed
t
o
ac
cu
r
ately
p
r
ed
ict
HR
im
ag
es.
T
h
e
av
er
ag
i
n
g
o
f
lay
er
s
f
r
o
m
E
SP
C
N
an
d
SR
R
esNet
will
b
e
p
er
f
o
r
m
ed
,
f
o
llo
w
ed
b
y
t
h
e
ca
lcu
latio
n
o
f
th
e
PS
NR
.
Hy
p
er
p
ar
am
eter
tu
n
in
g
is
also
co
n
d
u
cte
d
to
ascer
tain
th
e
o
p
ti
m
al
m
o
d
el.
T
h
e
test
in
g
p
r
o
ce
d
u
r
e
em
p
l
o
y
ed
th
e
f
ac
e
m
ask
d
ataset
s
u
p
p
lied
f
r
o
m
Kag
g
le.
T
h
e
E
SP
C
N
-
SR
R
esNet
m
o
d
el
is
u
tili
ze
d
f
o
r
th
e
p
u
r
p
o
s
e
o
f
s
u
p
er
r
eso
lu
tio
n
p
r
ed
ictio
n
.
T
h
e
d
ataset
co
n
s
is
ts
o
f
s
u
p
er
r
eso
lu
tio
n
im
ag
es
o
f
f
ac
e
m
ask
s
,
alo
n
g
with
th
eir
co
r
r
esp
o
n
d
in
g
o
r
ig
in
al
p
h
o
to
s
.
T
h
ese
im
ag
es
ar
e
s
u
b
s
eq
u
en
tly
u
tili
ze
d
as in
p
u
t
f
o
r
th
r
ee
d
if
f
er
e
n
t
C
NN
m
o
d
el
s
: V
GG1
6
,
Mo
b
ileNet,
an
d
R
esNet
5
0
.
T
h
er
e
ar
e
th
r
ee
m
o
d
el
s
th
at
ar
e
co
m
m
o
n
ly
em
p
lo
y
ed
f
o
r
th
e
p
u
r
p
o
s
e
o
f
ev
al
u
atin
g
th
e
p
er
f
o
r
m
an
ce
o
f
E
SP
C
N
-
SR
R
esNet.
T
h
e
s
tu
d
y
f
lo
wch
a
r
t is d
ep
icted
in
Fig
u
r
e
1
.
Fig
u
r
e
1
.
Flo
wch
ar
t
o
f
th
is
s
tu
d
y
2
.
1
.
E
SPCN
T
h
e
ac
r
o
n
y
m
E
SP
C
N
r
ep
r
ese
n
ts
ef
f
icien
t
s
u
b
-
p
ix
el
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
.
Sh
i
et
a
l.
[
1
6
]
in
tr
o
d
u
ce
s
a
u
n
iq
u
e
C
NN
ar
ch
itectu
r
e
d
esig
n
ed
s
p
ec
if
ically
f
o
r
th
e
r
ea
l
-
tim
e
en
h
a
n
ce
m
en
t
o
f
1
0
8
0
p
v
id
e
o
s
th
r
o
u
g
h
s
u
p
e
r
-
r
eso
lu
tio
n
tech
n
iq
u
es.
T
h
e
p
r
i
m
ar
y
co
n
ce
p
t
u
n
d
e
r
ly
in
g
E
SP
C
N
in
v
o
lv
es
th
e
ex
t
r
ac
tio
n
o
f
f
ea
tu
r
e
m
a
p
s
with
in
th
e
L
R
s
p
ac
e
[
1
8
]
,
as
o
p
p
o
s
ed
to
th
e
ty
p
ical
s
tr
ateg
y
o
f
u
tili
zin
g
th
e
HR
s
p
ac
e.
T
h
is
f
ac
ilit
ates
en
h
an
ce
d
co
m
p
u
ta
tio
n
al
ef
f
icien
c
y
a
n
d
m
itig
at
es
th
e
co
m
p
lex
ity
ass
o
ciate
d
with
t
h
e
s
u
p
e
r
-
r
eso
lu
tio
n
p
r
o
ce
d
u
r
e.
Fu
r
th
er
m
o
r
e
,
E
SP
C
N
p
r
esen
ts
a
h
ig
h
ly
ef
f
ec
tiv
e
s
u
b
-
p
ix
el
co
n
v
o
lu
ti
o
n
al
lay
er
.
T
h
e
p
r
esen
t
lay
er
is
d
esig
n
ed
t
o
ac
q
u
ir
e
k
n
o
wled
g
e
r
e
g
ar
d
in
g
a
n
ass
o
r
tm
en
t
o
f
u
p
s
ca
lin
g
f
ilter
s
[
1
9
]
,
wh
ic
h
ar
e
s
u
b
s
eq
u
e
n
tly
u
tili
ze
d
to
u
p
s
ca
le
th
e
u
ltima
te
L
R
f
ea
tu
r
e
m
ap
s
in
to
th
e
HR
o
u
tp
u
t.
B
y
s
u
b
s
titu
tin
g
th
e
m
an
u
ally
d
esig
n
e
d
b
icu
b
ic
f
ilter
with
th
ese
tr
ain
ed
u
p
s
ca
lin
g
f
ilter
s
,
th
e
E
SP
C
N
m
o
d
el
d
em
o
n
s
tr
ates
th
e
c
ap
ab
ilit
y
to
p
r
o
d
u
ce
im
ag
es
with
en
h
an
ce
d
s
h
ar
p
n
ess
an
d
in
cr
ea
s
ed
co
n
tr
ast.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
e
x
h
ib
its
s
u
p
er
io
r
p
e
r
f
o
r
m
an
ce
in
ter
m
s
o
f
b
o
th
r
ec
o
n
s
tr
u
ctio
n
ac
cu
r
ac
y
a
n
d
c
o
m
p
u
tatio
n
al
ef
f
icien
cy
co
m
p
ar
ed
to
ea
r
lier
m
eth
o
d
s
b
ased
o
n
C
NNs.
I
n
g
e
n
er
al,
E
SP
C
N
r
ef
er
s
to
an
ar
c
h
itectu
r
al
d
esig
n
with
in
th
e
C
NN
f
r
am
ew
o
r
k
th
at
ef
f
ec
tiv
ely
u
tili
ze
s
L
R
f
ea
tu
r
e
m
ap
s
an
d
a
s
u
b
-
p
ix
el
co
n
v
o
l
u
tio
n
lay
e
r
to
ac
co
m
p
lis
h
r
e
al
-
tim
e
s
u
p
er
-
r
eso
lu
tio
n
o
f
f
ilm
s
with
a
r
eso
lu
tio
n
o
f
1
0
8
0
p
.
T
h
e
en
h
an
ce
m
en
t
o
f
s
u
p
er
-
r
eso
lv
ed
im
ag
e
q
u
ality
i
s
ac
h
iev
ed
with
a
s
im
u
ltan
eo
u
s
r
ed
u
ctio
n
in
co
m
p
u
tin
g
co
m
p
le
x
ity
.
T
h
e
ar
c
h
itectu
r
al
d
esig
n
o
f
E
SP
C
N
ca
n
b
e
o
b
s
er
v
ed
in
Fig
u
r
e
2
.
2
.
2
.
S
up
er
-
re
s
o
lutio
n r
esid
u
a
l net
wo
rk
T
h
e
SR
R
es
Net
alg
o
r
ith
m
is
em
p
lo
y
ed
f
o
r
t
h
e
p
u
r
p
o
s
e
o
f
e
n
h
an
cin
g
th
e
r
eso
lu
tio
n
o
f
d
i
g
ital
co
r
e
im
ag
es
th
r
o
u
g
h
a
p
r
o
ce
s
s
k
n
o
wn
as
s
u
p
er
-
r
eso
lu
ti
o
n
r
ec
o
n
s
tr
u
ctio
n
.
T
h
e
p
r
o
p
o
s
ed
a
p
p
r
o
ac
h
u
tili
z
es
d
ee
p
lear
n
in
g
tech
n
iq
u
es
in
o
r
d
er
t
o
im
p
r
o
v
e
th
e
r
eso
lu
tio
n
an
d
o
v
er
all
q
u
ality
o
f
p
h
o
to
s
th
at
h
av
e
a
p
o
o
r
lev
el
o
f
d
etail.
T
h
e
n
etwo
r
k
a
r
ch
itectu
r
e
o
f
SR
R
esNet
co
m
p
r
is
es
a
g
en
er
ato
r
n
etwo
r
k
th
at
is
tr
ai
n
ed
to
estab
lis
h
th
e
r
elatio
n
s
h
ip
b
etwe
en
LR
an
d
H
R
p
ictu
r
es [
2
0
]
.
T
h
e
d
if
f
icu
lty
o
f
u
tili
zin
g
lim
ited
d
ata
f
r
o
m
s
m
all
s
am
p
les
f
o
r
s
u
p
er
-
r
eso
lu
tio
n
r
ec
o
n
s
tr
u
ctio
n
is
ef
f
ec
tiv
ely
ad
d
r
ess
ed
b
y
SR
R
esNet
with
th
e
im
p
lem
en
tatio
n
o
f
tr
an
s
f
er
lear
n
in
g
.
T
r
a
n
s
f
er
lear
n
in
g
is
a
wid
ely
em
p
lo
y
e
d
ap
p
r
o
ac
h
in
th
e
f
ield
o
f
m
ac
h
in
e
lear
n
in
g
,
wh
er
ein
th
e
p
ar
am
eter
s
o
f
a
p
r
e
-
ex
is
tin
g
m
o
d
el
ar
e
lev
er
ag
e
d
an
d
r
e
p
u
r
p
o
s
ed
f
o
r
a
d
if
f
er
e
n
t
jo
b
[
2
0
]
.
I
n
t
h
e
co
n
tex
t
o
f
SR
R
esNet,
th
e
p
r
e
-
ex
is
tin
g
m
o
d
el
h
as
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Lo
w
-
r
eso
lu
tio
n
ima
g
e
q
u
a
lity
en
h
a
n
ce
me
n
t u
s
in
g
en
h
a
n
ce
d
…
(
Mo
h
a
mma
d
F
a
is
a
l R
ifti
a
r
r
a
s
yid
)
637
ac
q
u
ir
ed
s
h
ar
e
d
ch
ar
ac
te
r
is
tics
f
r
o
m
an
e
x
ten
s
iv
e
d
ataset.
T
h
r
o
u
g
h
th
e
u
s
e
o
f
tr
an
s
f
er
lear
n
in
g
,
th
e
SR
R
esNe
t
tech
n
iq
u
e
h
as
th
e
ca
p
ab
ilit
y
to
ef
f
ec
tiv
ely
tr
ain
d
is
tin
ctiv
e
f
ea
tu
r
es
with
in
lim
ited
s
am
p
le
d
ata
[
2
1
]
,
h
en
c
e
d
im
in
is
h
in
g
th
e
n
ec
ess
ity
f
o
r
an
ex
ten
s
iv
e
t
r
ain
in
g
d
ataset
an
d
allev
iatin
g
th
e
is
s
u
es
o
f
o
v
er
f
itti
n
g
th
at
ar
is
e
d
u
e
to
in
a
d
eq
u
ate
d
ata.
I
n
th
e
d
o
m
ain
o
f
s
u
p
er
-
r
eso
lu
tio
n
r
ec
o
n
s
tr
u
ctio
n
o
f
d
ig
ital
co
r
e
im
ag
es,
th
e
u
tili
za
tio
n
o
f
SR
R
es
Net
with
tr
an
s
f
er
lear
n
in
g
e
n
co
m
p
ass
es
a
two
-
s
tep
p
r
o
ce
s
s
[
1
5
]
.
T
h
e
in
itial
p
a
r
t
o
f
p
r
e
-
tr
ain
i
n
g
in
v
o
lv
es
u
tili
zin
g
th
e
p
lu
n
g
er
s
am
p
le
im
ag
e
an
d
th
e
LR
im
a
g
e
ac
q
u
ir
ed
f
r
o
m
d
o
wn
s
am
p
lin
g
as
th
e
tr
ain
in
g
d
ataset.
T
h
is
s
tag
e
s
er
v
es
to
illu
s
tr
ate
th
e
ca
p
ab
il
ity
o
f
SR
R
esNet
in
ac
h
iev
in
g
s
u
p
er
-
r
eso
lu
tio
n
r
ec
o
n
s
tr
u
ctio
n
.
Fu
r
th
e
r
m
o
r
e
,
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
lay
er
o
f
th
e
p
r
e
-
tr
ain
ed
m
o
d
el
is
im
m
o
b
i
lized
,
an
d
th
e
m
o
d
el
u
n
d
er
g
o
es
f
in
e
-
tu
n
i
n
g
b
y
u
tili
zin
g
r
ea
l
p
lu
n
g
er
s
am
p
les
an
d
s
u
b
s
am
p
le
im
a
g
es
as
th
e
f
r
esh
tr
ain
in
g
d
ataset.
T
h
e
u
tili
za
tio
n
o
f
tr
an
s
f
er
lear
n
in
g
in
th
e
SR
R
esNet
m
e
th
o
d
en
ab
les
th
e
r
ec
o
n
s
tr
u
ctio
n
o
f
f
in
e
r
d
etails,
in
clu
d
in
g
tex
tu
r
e
an
d
p
o
r
es,
d
esp
ite
th
e
p
r
esen
ce
o
f
a
lim
it
ed
n
u
m
b
er
o
f
tr
ain
in
g
ex
am
p
les.
I
n
b
r
ief
,
th
e
SR
R
esNet
ap
p
r
o
a
ch
is
d
esig
n
ed
f
o
r
th
e
p
u
r
p
o
s
e
o
f
s
u
p
er
-
r
eso
lu
tio
n
r
ec
o
n
s
tr
u
ctio
n
[
2
2
]
.
I
t
tack
les
th
e
is
s
u
e
o
f
lim
it
ed
d
ata
r
esu
ltin
g
f
r
o
m
s
m
al
l
s
am
p
le
s
izes
b
y
em
p
lo
y
in
g
t
r
an
s
f
er
lear
n
in
g
tech
n
iq
u
es.
T
h
is
in
v
o
lv
es
le
v
er
ag
in
g
p
r
e
-
e
x
is
tin
g
m
o
d
els
an
d
tr
ain
i
n
g
s
p
ec
if
ic
f
ea
tu
r
es
u
s
in
g
tin
y
s
am
p
le
d
atasets
.
T
h
e
ar
ch
itectu
r
al
d
es
ig
n
o
f
SR
R
esNet
ca
n
b
e
o
b
s
er
v
ed
in
Fig
u
r
e
3
.
Fig
u
r
e
2
.
E
SP
C
N
a
r
ch
itectu
r
e
[
1
6
]
Fig
u
r
e
3
.
SR
R
esNet
a
r
ch
itectu
r
e
[
2
0
]
2
.
3
.
Sig
na
l
-
to
-
no
is
e
ra
t
io
PS
NR
is
a
co
m
m
o
n
ly
u
s
ed
m
etr
ic
to
ev
alu
ate
th
e
q
u
ality
o
f
r
ec
o
n
s
tr
u
cte
d
p
ictu
r
es
in
th
e
f
ield
o
f
s
u
p
er
-
r
eso
lu
tio
n
r
ec
o
n
s
tr
u
ctio
n
.
T
h
e
r
elev
a
n
ce
o
f
t
h
is
r
esid
es
in
its
ca
p
ac
ity
to
q
u
an
t
if
y
th
e
co
r
r
elatio
n
b
etwe
en
th
e
s
ig
n
al
’
s
m
ax
im
u
m
p
o
te
n
tial
p
o
wer
an
d
th
e
lev
el
o
f
n
o
is
e
th
at
i
m
p
ac
ts
th
e
ac
cu
r
ac
y
o
f
t
h
e
s
ig
n
al.
PS
NR
i
s
cr
u
cial
in
ev
alu
atin
g
th
e
ac
cu
r
ac
y
o
f
s
u
p
e
r
-
r
eso
l
u
tio
n
r
ec
o
n
s
tr
u
ctio
n
b
y
m
ea
s
u
r
in
g
th
e
s
im
ilar
ity
b
etwe
en
th
e
r
ec
o
n
s
tr
u
cte
d
HR
im
ag
e
an
d
th
e
ac
tu
al
HR
im
ag
e.
T
h
e
PS
NR
m
ea
s
u
r
e
p
r
o
v
id
es
a
n
u
m
e
r
ical
ev
alu
atio
n
o
f
th
e
q
u
ality
o
f
t
h
e
r
ec
o
n
s
tr
u
cted
im
a
g
e.
A
h
i
g
h
er
PS
NR
v
alu
e
in
d
icate
s
a
clo
s
er
m
atch
to
th
e
o
r
ig
in
al
im
ag
e
an
d
h
en
ce
f
ew
er
d
is
to
r
tio
n
s
.
T
h
e
p
r
im
ar
y
ad
v
an
tag
e
o
f
PS
NR
is
its
d
ep
en
d
ab
ilit
y
in
ca
tch
in
g
s
u
b
tl
e
in
tr
icac
ies
an
d
g
en
er
al
s
im
ilar
ity
b
etwe
en
th
e
r
ec
o
n
s
tr
u
cted
an
d
o
r
ig
in
al
HR
im
ag
es,
r
en
d
er
in
g
it
an
in
d
is
p
en
s
ab
le
to
o
l f
o
r
co
m
p
let
e
ev
alu
atio
n
o
f
im
ag
e
q
u
ality
[
2
3
]
.
T
h
e
PS
NR
is
a
wid
ely
u
s
ed
m
etr
ic
f
o
r
ev
alu
atin
g
th
e
q
u
ality
o
f
c
o
m
p
r
ess
ed
o
r
d
is
to
r
ted
im
ag
es
[
2
4
]
.
I
t
m
ea
s
u
r
es
th
e
d
is
cr
ep
an
cy
b
etwe
en
th
e
o
r
ig
in
al
a
n
d
d
is
to
r
ted
im
ag
es
in
ter
m
s
o
f
p
ix
el
v
a
lu
es.
PS
NR
,
wh
ich
is
m
ea
s
u
r
ed
in
d
ec
ib
els
(
d
B
)
,
is
d
eter
m
in
ed
b
y
ca
lcu
latin
g
th
e
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
b
etwe
en
co
r
r
esp
o
n
d
in
g
p
ix
els
in
two
p
ictu
r
es
a
n
d
ex
p
r
ess
in
g
it
as
th
e
r
atio
o
f
p
ea
k
s
ig
n
al
p
o
wer
(
h
ig
h
est
p
ix
el
v
alu
e)
to
MSE
.
T
h
e
PS
NR
m
etr
ic
i
s
cr
u
cial
in
ass
ess
in
g
im
ag
e
q
u
ality
,
s
in
ce
a
g
r
ea
ter
v
alu
e
in
d
icate
s
s
u
p
er
i
o
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
634
-
6
4
3
638
im
ag
e
q
u
ality
,
wh
ile
a
lo
wer
v
alu
e
s
ig
n
als
p
o
o
r
er
q
u
a
lity
.
T
h
e
p
r
im
ar
y
ad
v
an
tag
e
o
f
PS
NR
is
its
d
ep
en
d
a
b
ilit
y
as
a
q
u
an
titativ
e
m
etr
ic,
o
f
f
er
in
g
a
d
is
tin
ct
in
d
icato
r
o
f
th
e
q
u
ality
o
f
t
h
e
r
ec
o
n
s
tr
u
cted
o
r
co
m
p
r
ess
ed
im
ag
e.
I
ts
wid
esp
r
ea
d
u
s
ag
e
s
tem
s
f
r
o
m
its
ca
p
ac
ity
to
ac
cu
r
ately
m
ea
s
u
r
e
th
e
s
im
ilar
ity
in
p
ix
el
v
alu
es
b
etwe
en
th
e
o
r
ig
i
n
al
a
n
d
d
is
to
r
ted
im
ag
es,
p
r
o
v
id
i
n
g
a
clea
r
a
n
d
u
n
d
e
r
s
tan
d
ab
le
ev
alu
atio
n
o
f
im
ag
e
q
u
ality
[
2
5
]
.
T
h
e
f
o
r
m
u
la
f
o
r
PS
NR
i
s
elu
cid
ated
in
(
1
)
.
=
20
.
l
og
10
(
√
)
(
1
)
2
.
4
.
M
a
s
k
c
l
a
s
s
if
ica
t
io
n
Af
ter
im
ag
e
en
h
a
n
ce
m
en
t
h
as
b
ee
n
b
u
ilt
a
n
d
f
in
e
-
tu
n
ed
,
th
i
s
r
esear
ch
will
h
a
v
e
to
p
r
o
v
e
its
im
p
ac
t
o
n
f
u
tu
r
e
r
esear
c
h
esp
ec
ially
in
co
m
p
u
ter
s
cien
ce
s
u
b
jects
.
T
o
d
o
it,
th
is
r
esear
ch
em
p
lo
y
d
ee
p
lear
n
in
g
to
m
ea
s
u
r
e
im
ag
e
en
h
an
ce
m
en
t
ef
f
ec
t
o
n
im
ag
e
class
if
icati
o
n
.
T
o
d
eter
m
in
e
wh
eth
er
i
m
ag
e
en
h
an
ce
m
e
n
t
im
p
r
o
v
e
o
r
wo
r
s
en
d
ee
p
lear
n
in
g
p
er
f
o
r
m
an
ce
,
th
is
r
esear
ch
will
f
o
cu
s
to
m
ea
s
u
r
e
th
e
ac
cu
r
ac
y
o
f
test
ed
m
o
d
els.
Acc
u
r
ac
y
is
ch
o
s
en
a
s
m
etr
ics
o
f
p
er
f
o
r
m
a
n
ce
b
ec
au
s
e
it
m
ea
s
u
r
e
m
o
d
el
p
er
f
o
r
m
an
ce
o
n
co
r
r
ec
tly
class
if
y
in
g
d
ata.
Acc
u
r
ac
y
is
with
(
2
).
=
100%
(
2
)
2
.
5
.
Da
t
a
s
et
T
h
e
r
esear
ch
u
tili
ze
d
t
h
r
ee
d
a
tasets
:
5
0
0
p
h
o
to
s
s
o
u
r
ce
d
f
r
o
m
B
SDS5
0
0
,
6
0
0
im
a
g
es
o
b
ta
in
ed
f
r
o
m
p
ex
els.co
m
,
an
d
3
,
4
4
4
f
ac
e
m
ask
im
ag
es
e
x
tr
ac
ted
f
r
o
m
th
e
f
ac
e
m
ask
d
ataset.
T
h
e
B
SDS5
0
0
an
d
6
0
0
d
atasets
co
n
s
is
t
o
f
a
co
m
b
in
ed
to
tal
o
f
1
,
1
0
0
p
h
o
t
o
s
.
T
h
ese
d
atasets
aim
to
en
co
m
p
ass
a
wid
e
r
an
g
e
o
f
s
u
b
jects,
in
clu
d
in
g
b
u
il
d
in
g
s
,
an
im
als,
h
u
m
an
s
,
n
atu
r
e,
an
d
o
th
e
r
r
elev
a
n
t
ca
teg
o
r
ies.
T
h
e
in
ten
t
io
n
b
e
h
in
d
th
is
co
m
p
r
eh
e
n
s
iv
e
co
llectio
n
is
to
en
s
u
r
e
th
at
all
asp
ec
ts
o
f
im
ag
er
y
in
s
o
ciety
ar
e
well
r
e
p
r
esen
ted
.
I
n
th
e
in
itial
tr
ain
i
n
g
p
h
ase,
a
to
tal
o
f
4
5
0
p
h
o
to
g
r
ap
h
s
wer
e
u
tili
ze
d
f
o
r
tr
ain
in
g
p
u
r
p
o
s
es,
wh
ile
an
ad
d
itio
n
al
4
5
0
im
ag
es
wer
e
allo
ca
ted
f
o
r
test
in
g
.
Fu
r
th
er
m
o
r
e,
an
a
d
d
itio
n
al
s
et
o
f
2
0
0
im
ag
es
wer
e
d
esig
n
ated
f
o
r
v
alid
atio
n
i
n
o
r
d
er
to
o
b
tain
th
e
m
o
s
t
o
p
tim
al
m
o
d
el
f
o
r
s
u
p
er
r
eso
lu
tio
n
.
T
h
e
s
ec
o
n
d
tr
ain
i
n
g
s
ess
io
n
u
tili
ze
d
a
d
ataset
co
n
s
is
tin
g
o
f
3
,
4
4
4
p
h
o
t
o
s
s
p
ec
if
ically
d
esig
n
ed
f
o
r
f
ac
e
m
ask
d
etec
tio
n
.
T
h
is
d
ataset
was
f
u
r
th
er
d
iv
id
e
d
in
to
s
u
b
s
ets,
with
2
,
7
5
4
im
ag
es
allo
ca
ted
f
o
r
tr
ai
n
in
g
p
u
r
p
o
s
es,
4
8
5
im
ag
es
f
o
r
v
alid
atio
n
,
an
d
2
0
5
im
a
g
es
f
o
r
test
in
g
.
I
llu
s
tr
atio
n
s
d
ep
ictin
g
s
am
p
le
im
ag
es
f
o
r
th
e
in
i
tial
an
d
s
u
b
s
eq
u
en
t
tr
ain
in
g
s
ess
io
n
s
ar
e
p
r
esen
ted
in
Fig
u
r
e
4
.
Fig
u
r
e
4
(
a)
s
h
o
ws
s
ev
er
al
ex
am
p
les
o
f
B
SD5
0
0
d
ataset,
also
s
ev
er
al
ex
am
p
les
o
f
6
0
0
s
cr
ap
p
ed
im
ag
es.
Fig
u
r
e
4
(
b
)
s
h
o
ws
s
ev
er
al
ex
am
p
les
o
f
f
a
ce
m
ask
d
etec
tio
n
d
ataset.
(
a)
(
b
)
Fig
u
r
e
4
.
E
x
am
p
les o
f
d
ataset
(
a)
B
SD5
0
0
an
d
6
0
0
s
cr
ap
p
e
d
im
ag
es
an
d
(
b
)
f
ac
e
m
ask
d
et
ec
tio
n
d
ataset
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
im
p
lem
en
tatio
n
o
f
th
e
s
u
p
er
-
r
eso
lu
tio
n
m
eth
o
d
o
lo
g
y
r
eq
u
ir
ed
d
o
in
g
m
u
ltip
le
iter
ati
o
n
s
o
f
test
s
th
at
in
v
o
lv
ed
m
er
g
in
g
th
e
E
S
PC
N
an
d
SR
R
es
N
et
tech
n
iq
u
e
s
.
T
h
e
co
n
d
u
cted
ex
p
e
r
im
en
t
s
en
co
m
p
ass
ed
th
e
p
r
o
ce
s
s
o
f
h
y
p
er
p
a
r
am
eter
o
p
tim
izatio
n
,
em
p
lo
y
in
g
b
o
th
m
an
u
al
an
d
g
r
id
s
ea
r
ch
m
eth
o
d
o
lo
g
ies.
E
x
p
er
im
en
ts
wer
e
p
er
f
o
r
m
e
d
u
s
in
g
d
if
f
er
e
n
t
h
y
p
e
r
p
ar
am
ete
r
co
n
f
ig
u
r
atio
n
s
,
in
clu
d
in
g
m
o
d
if
icatio
n
s
in
b
atch
s
ize,
lear
n
in
g
r
ate,
E
SP
C
N
s
tr
u
ctu
r
e,
an
d
SR
R
es
N
et
s
tr
u
ctu
r
e.
T
h
e
m
an
u
al
a
p
p
r
o
ac
h
en
co
m
p
ass
ed
th
e
p
r
o
ce
s
s
o
f
o
p
tim
izin
g
th
e
q
u
a
n
tity
o
f
lay
er
s
an
d
f
ilter
s
in
b
o
th
m
o
d
els.
I
n
th
e
s
tu
d
y
co
n
d
u
cted
,
f
o
u
r
d
if
f
er
en
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Lo
w
-
r
eso
lu
tio
n
ima
g
e
q
u
a
lity
en
h
a
n
ce
me
n
t u
s
in
g
en
h
a
n
ce
d
…
(
Mo
h
a
mma
d
F
a
is
a
l R
ifti
a
r
r
a
s
yid
)
639
s
tr
u
ctu
r
al
v
ar
ian
ts
wer
e
ex
am
in
ed
in
th
e
E
SP
C
N
m
o
d
el.
T
h
e
s
e
v
ar
iatio
n
s
in
clu
d
ed
(
6
4
,
6
4
,
3
2
)
,
(
6
4
,
6
4
,
6
4
,
3
2
)
,
(
1
2
8
,
6
4
,
6
4
,
3
2
)
,
an
d
(
1
2
8
,
1
2
8
,
6
4
,
3
2
)
.
Ad
d
itio
n
ally
,
t
wo
d
if
f
e
r
en
t
lear
n
in
g
r
ate
v
ar
i
atio
n
s
wer
e
test
ed
,
s
p
ec
if
ically
0
.
0
0
1
,
0
.
0
0
0
1
,
an
d
0
.
0
0
0
0
1
.
Am
o
u
n
t
o
f
SR
R
es
Net
lay
er
s
ex
am
in
e
d
in
th
is
s
t
u
d
y
ar
e
(
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
)
an
d
(
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
)
.
T
h
at
v
ar
iatio
n
s
also
u
s
e
d
f
o
r
g
r
i
d
s
ea
r
ch
.
T
h
e
test
r
es
u
lts
ar
e
p
r
esen
ted
in
T
ab
le
1
.
T
ab
le
1
.
Ma
n
u
al
h
y
p
er
p
a
r
am
e
ter
tu
n
in
g
N
u
mb
e
r
ESP
C
N
S
R
R
e
sN
e
t
Le
a
r
n
i
n
g
r
a
t
e
P
S
N
R
p
r
e
d
i
c
t
(
d
b
)
Ex
e
c
u
t
i
o
n
t
i
m
e
(
s)
1
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
4
3
0
.
1
0
6
2
6
4
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
4
0
.
3
8
1
3
6
4
,
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
5
1
0
.
2
7
2
4
6
4
,
6
4
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
3
8
0
.
1
2
7
5
1
2
8
,
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
3
9
0
.
3
7
7
6
1
2
8
,
6
4
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
3
5
0
.
4
5
2
7
1
2
8
,
1
2
8
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
5
1
0
.
3
4
7
8
1
2
8
,
1
2
8
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
5
2
0
.
9
2
9
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
1
2
7
.
7
0
.
8
2
10
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
0
9
0
.
8
2
6
11
6
4
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
1
3
0
.
4
6
1
12
6
4
,
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
0
5
0
.
4
2
7
13
6
4
,
6
4
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
2
4
0
.
3
4
8
14
1
2
8
,
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
0
5
0
.
4
8
7
15
1
2
8
,
6
4
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
1
4
0
.
6
9
3
16
1
2
8
,
1
2
8
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
1
2
0
.
3
4
7
17
1
2
8
,
1
2
8
,
6
4
,
3
2
6
4
,
[
7
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
0
1
2
8
.
1
4
0
.
7
6
6
Fro
m
T
ab
le
1
,
it
ca
n
b
e
s
ee
n
th
at
th
e
h
ig
h
est
PS
NR
o
b
tain
ed
b
y
th
e
8
th
ex
p
er
im
en
t
an
d
th
e
s
ec
o
n
d
h
ig
h
est
PS
NR
o
b
tain
ed
b
y
th
e
7
th
ex
p
er
im
en
t.
T
h
e
PS
NR
v
alu
es
o
b
tain
ed
b
y
th
at
two
e
x
p
er
im
en
ts
ar
e
n
o
t
m
u
ch
d
if
f
er
e
n
t,
b
u
t
th
er
e
ar
e
q
u
ite
s
ig
n
if
ican
t
d
if
f
er
en
ce
s
i
n
ex
ec
u
tio
n
tim
e.
T
h
e
7
th
ex
p
er
im
en
t
r
u
n
6
2
.
2
8
8
%
f
aster
th
an
th
e
8
th
ex
p
er
im
en
t,
h
en
ce
th
e
7
th
tr
ial
in
y
ield
ed
th
e
o
p
tim
al
r
esu
lt.
T
h
e
r
esu
lts
o
f
th
is
ex
p
er
im
en
t
d
em
o
n
s
tr
ate
th
at
E
SC
PC
N
h
as
ex
ce
p
tio
n
al
tim
e
e
f
f
icie
n
cy
in
co
m
p
u
tatio
n
al
task
s
.
T
h
er
e
is
o
n
l
y
o
n
e
ex
p
er
im
en
t th
at
u
s
e
0
.
0
0
1
lear
n
in
g
r
ate
an
d
th
at
is
th
e
9
th
ex
p
er
im
en
t.
I
ts
b
ec
au
s
e
0
.
0
0
1
le
ar
n
in
g
r
ate
p
r
o
d
u
ce
lo
wer
PS
NR
an
d
lo
n
g
er
ex
ec
u
tio
n
tim
e.
Dee
p
er
E
SP
C
N
a
n
d
SR
R
es
Net
lay
er
s
tak
es
lo
n
g
er
ex
ec
u
tio
n
tim
e,
co
u
ld
b
e
s
ee
n
in
ex
p
er
im
en
t
8
t
h
an
d
1
7
th
.
Gr
id
s
ea
r
ch
is
also
u
s
ed
in
th
i
s
s
tu
d
y
to
f
i
n
d
th
e
b
est
h
y
p
er
p
ar
am
eter
f
o
r
im
a
g
e
e
n
h
an
ce
m
en
t.
R
esu
lt
s
h
o
wn
th
at
co
m
p
ar
ed
to
th
e
b
est
m
o
d
el
f
r
o
m
m
an
u
al
tu
n
i
n
g
(
ex
p
e
r
im
en
t
7
)
,
g
r
id
s
ea
r
ch
h
as
a
litt
le
lo
wer
PS
NR
.
T
h
e
r
esu
lt
co
u
ld
b
e
s
e
en
in
T
ab
le
2
.
As
s
ee
n
in
T
ab
l
e
2
,
n
u
m
b
er
o
f
SR
R
esNet
lay
er
an
d
lea
r
n
in
g
r
ate
f
r
o
m
g
r
id
s
ea
r
ch
an
d
m
an
u
al
tu
n
in
g
h
as
th
e
s
am
e
v
alu
e.
Ho
wev
er
,
m
o
d
el
f
r
o
m
g
r
id
s
ea
r
ch
h
as
less
f
ilter
in
th
e
s
ec
o
n
d
lay
er
th
a
n
m
o
d
el
f
r
o
m
m
an
u
al
tu
n
in
g
.
E
x
ec
u
tio
n
tim
e
o
f
g
r
id
s
ea
r
ch
an
d
m
an
u
al
tu
n
in
g
h
as
n
o
s
ig
n
if
ican
t d
if
f
er
e
n
ce
.
T
ab
le
2
.
Hy
p
er
p
ar
a
m
eter
tu
n
i
n
g
with
g
r
i
d
s
ea
r
ch
Ex
p
e
r
i
m
e
n
t
ESP
C
N
S
R
R
e
sN
e
t
Le
a
r
n
i
n
g
r
a
t
e
P
S
N
R
p
r
e
d
i
c
t
(
d
b
)
Ex
e
c
u
t
i
o
n
t
i
m
e
(
s)
M
a
n
u
a
l
T
u
n
i
n
g
1
2
8
,
1
2
8
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
5
1
4
2
0
.
3
4
7
G
r
i
d
S
e
a
r
c
h
1
2
8
,
6
4
,
6
4
,
3
2
6
4
,
[
5
*
6
4
]
,
6
4
,
2
5
6
0
.
0
0
1
2
8
.
5
1
2
5
0
.
3
8
7
As
p
r
ev
io
u
s
ly
m
en
tio
n
ed
,
th
er
e
ar
e
th
r
ee
class
if
icatio
n
m
o
d
el
u
s
ed
in
th
is
r
esear
ch
.
VGG1
6
as
m
o
d
el
n
u
m
b
er
o
n
e
,
Mo
b
ileNe
tV2
as
m
o
d
el
n
u
m
b
e
r
two
,
a
n
d
R
esNet5
0
as
m
o
d
el
n
u
m
b
er
th
r
ee
.
E
ac
h
m
o
d
els
was
tr
ain
ed
u
s
in
g
tr
an
s
f
er
lear
n
in
g
m
eth
o
d
b
y
f
r
ee
zi
n
g
its
lay
e
r
an
d
co
n
n
ec
tin
g
1
lay
er
o
f
f
u
lly
-
co
n
n
ec
ted
n
o
d
es
an
d
clo
s
ed
o
f
with
o
u
tp
u
t
lay
er
wh
ich
r
ep
r
esen
ts
ea
ch
class
es
in
lab
el.
Fig
u
r
e
5
s
h
o
ws
m
o
d
el
tr
ain
in
g
r
esu
lt.
Mo
d
el
tr
ain
in
g
lo
s
s
co
u
ld
b
e
s
ee
n
in
Fig
u
r
e
5
(
a)
an
d
m
o
d
el
tr
ain
in
g
ac
cu
r
ac
y
co
u
ld
b
e
s
ee
n
i
n
F
ig
u
r
e
5
(
b
)
.
Mo
d
el
1
s
tan
d
s
f
o
r
VGG1
6
,
m
o
d
el
2
s
tan
d
s
f
o
r
Mo
b
ileNetV2
,
m
o
d
el
3
s
tan
d
s
f
o
r
R
esNet5
0
.
I
n
th
e
test
in
g
p
h
ase,
in
p
u
t
f
r
o
m
th
e
m
ask
class
if
icatio
n
m
o
d
el
is
u
p
s
ca
led
with
s
u
p
er
-
r
eso
lu
tio
n
f
r
o
m
th
e
im
ag
e
en
h
an
ce
m
e
n
t
m
o
d
el.
I
m
a
g
e
en
h
an
ce
m
e
n
t
m
o
d
el
wi
th
th
e
g
r
ea
test
p
er
f
o
r
m
a
n
ce
with
o
p
tim
al
PS
NR
an
d
ex
ec
u
tio
n
tim
e
r
esu
lt
was
ch
o
s
en
as
th
e
m
o
d
el
wh
ich
an
d
in
teg
r
ated
to
th
e
m
ask
class
i
f
icatio
n
.
B
ased
o
n
T
a
b
l
e
1
t
h
e
m
o
s
t
o
p
t
i
m
a
l
m
o
d
e
l
w
i
t
h
t
h
e
g
r
e
a
t
es
t
P
S
NR
v
a
l
u
e
a
n
d
m
i
n
i
m
a
l
e
x
e
c
u
t
i
o
n
t
i
m
e
i
s
m
o
d
e
l
n
u
m
b
e
r
7.
A
n
d
b
a
s
e
d
o
n
t
h
e
g
r
i
d
s
e
a
r
c
h
m
e
t
h
o
d
r
e
s
u
l
t
e
d
i
n
PS
NR
v
a
l
u
e
o
f
2
8
.
5
1
2
5
a
n
d
e
x
e
c
u
t
i
o
n
t
i
m
e
0
.
3
8
7
2
8
7
1
0
9
2
m
s
.
I
n
teg
r
atio
n
r
esu
lts
is
p
r
esen
ted
as
co
m
p
ar
is
o
n
b
etwe
en
t
h
r
ee
p
o
p
u
lar
class
if
icatio
n
ar
ch
itectu
r
e
wh
ich
is
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
634
-
6
4
3
640
VGG1
6
,
Mo
b
ilen
etV,
an
d
R
esNet5
0
.
Acc
u
r
ac
y
was
u
s
ed
to
m
ea
s
u
r
e
p
er
f
o
r
m
a
n
ce
im
p
ac
t
o
f
th
o
s
e
th
r
ee
class
if
icatio
n
m
o
d
els,
a
co
m
p
a
r
is
o
n
b
ef
o
r
e
an
d
af
ter
s
u
p
e
r
r
e
s
o
lu
tio
n
will b
e
p
r
esen
ted
.
(
a)
(
b
)
Fig
u
r
e
5
.
Mo
d
el
tr
ain
in
g
(
a)
l
o
s
s
an
d
(
b
)
ac
cu
r
ac
y
Data
f
r
o
m
T
ab
le
3
s
h
o
w
n
an
in
cr
ea
s
e
in
ac
cu
r
ac
y
b
y
u
s
in
g
s
u
p
er
r
eso
lu
tio
n
to
p
r
e
-
p
r
o
ce
s
s
in
p
u
t
im
ag
es.
VGG1
6
h
as
th
e
m
o
s
t
s
ig
n
if
ican
t
im
p
r
o
v
e
m
en
t
with
2
5
%
in
cr
ea
s
e
in
ac
cu
r
ac
y
,
w
h
ich
is
f
o
llo
wed
b
y
Mo
b
ileNetV2
with
5
% in
cr
ea
s
e
in
ac
cu
r
ac
y
,
an
d
R
esNet5
0
wh
ich
h
as th
e
least im
p
r
o
v
e
m
en
t in
ac
cu
r
ac
y
with
less
th
an
0
.
0
0
1
%
in
cr
ea
s
e.
Fu
r
th
er
an
aly
s
is
in
d
ee
p
lear
n
in
g
class
if
icatio
n
m
o
d
el
s
h
o
ws
th
at
VGG1
6
ar
ch
itectu
r
e
h
as
1
6
lay
er
s
o
f
n
eu
r
al
n
etwo
r
k
[
2
6
]
,
Mo
b
ileNetV2
h
as
1
1
lay
er
s
[
2
7
]
,
an
d
R
esNet5
0
h
as
5
0
lay
er
s
o
f
n
eu
r
al
n
etwo
r
k
with
r
esid
u
al
n
etwo
r
k
[
2
6
]
.
T
h
is
in
d
icate
s
th
at
in
cr
ea
s
in
g
m
o
d
e
l
co
m
p
lex
ity
lay
er
r
ed
u
ce
s
th
e
im
p
ac
t
o
f
s
u
p
e
r
-
r
eso
lu
tio
n
.
Nev
er
t
h
eless
,
th
is
f
in
d
in
g
also
s
u
p
p
o
r
ts
th
is
r
esear
ch
p
u
r
p
o
s
es
wh
ich
is
im
p
r
o
v
in
g
r
esu
lts
o
n
d
ee
p
l
ea
r
n
in
g
m
o
d
el
in
a
lo
w
p
er
f
o
r
m
in
g
d
e
v
ice.
Data
f
r
o
m
T
ab
l
e
4
s
h
o
wn
th
a
t
f
in
e
-
tu
n
in
g
im
ag
e
e
n
h
an
ce
m
en
t
m
o
d
el
with
g
r
id
s
ea
r
ch
a
n
d
m
an
u
al
f
in
e
-
tu
n
in
g
ac
h
iev
e
s
im
ilar
r
e
s
u
lts
.
VGG1
6
m
o
d
el
test
r
esu
lted
in
9
6
.
3
0
%
ac
cu
r
ac
y
u
s
in
g
m
an
u
al
a
n
d
g
r
id
s
ea
r
ch
f
in
e
-
tu
n
i
n
g
.
f
o
llo
wed
b
y
Mo
b
ileNetV2
wh
o
s
e
ac
cu
r
a
cy
s
co
r
ed
7
6
.
3
0
%
in
b
o
th
m
an
u
al
an
d
g
r
id
s
ea
r
c
h
f
in
e
-
tu
n
in
g
.
a
n
d
R
esNet5
0
ac
h
iev
ed
7
1
.
3
0
%
ac
cu
r
ac
y
i
n
b
o
th
m
a
n
u
al
a
n
d
g
r
id
s
ea
r
ch
.
All
th
e
test
ed
m
o
d
els
s
h
o
wed
a
m
in
u
s
cu
le
d
if
f
e
r
en
ce
with
less
th
an
0
.
0
0
1
%
d
if
f
er
en
ce
i
n
ac
c
u
r
ac
y
b
etwe
en
th
e
two
f
in
e
t
u
n
in
g
m
eth
o
d
s
.
T
ab
le
3
.
C
lass
if
icatio
n
ac
cu
r
a
cy
co
m
p
a
r
is
o
n
M
o
d
e
l
C
l
a
s
si
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
W
i
t
h
s
u
p
e
r
r
e
s
o
l
u
t
i
o
n
W
i
t
h
o
u
t
s
u
p
e
r
r
e
s
o
l
u
t
i
o
n
V
G
G
1
6
9
6
.
3
0
%
7
1
.
3
0
%
M
o
b
i
l
e
N
e
t
V
2
7
6
.
3
0
%
7
1
.
3
0
%
R
e
sN
e
t
5
0
7
1
.
3
0
%
7
1
.
3
0
%
T
ab
le
4
.
Ma
n
u
al
tu
n
in
g
an
d
g
r
id
s
ea
r
ch
im
p
ac
t o
n
m
ask
clas
s
if
icatio
n
M
o
d
e
l
H
y
p
e
r
p
a
r
a
me
t
e
r
t
u
n
i
n
g
M
a
n
u
a
l
t
u
n
i
n
g
G
r
i
d
s
e
a
r
c
h
V
G
G
1
6
9
6
.
3
0
%
9
6
.
3
0
%
M
o
b
i
l
e
N
e
t
V
2
7
6
.
3
0
%
7
6
.
3
0
%
R
e
sN
e
t
5
0
7
1
.
3
0
%
7
1
.
3
0
%
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Lo
w
-
r
eso
lu
tio
n
ima
g
e
q
u
a
lity
en
h
a
n
ce
me
n
t u
s
in
g
en
h
a
n
ce
d
…
(
Mo
h
a
mma
d
F
a
is
a
l R
ifti
a
r
r
a
s
yid
)
641
4.
CO
NCLU
SI
O
N
T
h
is
r
esear
ch
d
em
o
n
s
tr
ates
th
e
ef
f
ec
tiv
e
in
te
g
r
atio
n
o
f
E
SP
C
N
with
SR
R
esNe
t,
r
esu
ltin
g
in
a
s
u
b
s
tan
tial e
n
h
an
ce
m
en
t o
f
p
i
ctu
r
e
q
u
ality
f
o
r
f
ac
e
m
ask
cla
s
s
if
ier
s
.
T
h
e
h
an
d
twea
k
in
g
tr
ials
y
ield
ed
an
id
ea
l
co
n
f
ig
u
r
atio
n
with
a
PS
NR
o
f
2
8
.
5
1
4
2
d
B
an
d
an
e
x
ec
u
ti
o
n
tim
e
o
f
0
.
3
4
7
0
4
6
3
8
s
ec
o
n
d
s
.
T
h
ese
f
in
d
in
g
s
s
h
o
wca
s
e
th
e
s
u
p
r
em
ac
y
o
f
th
is
tech
n
iq
u
e
in
ef
f
ec
tiv
ely
p
r
o
d
u
cin
g
ex
ce
p
tio
n
al
o
u
tco
m
es
th
r
o
u
g
h
m
eticu
lo
u
s
h
an
d
twea
k
in
g
.
T
h
e
e
x
p
er
im
en
tal
r
esu
lts
f
u
r
th
er
em
p
h
asize
th
e
s
ig
n
if
ican
t
n
ee
d
o
f
m
a
n
u
ally
ad
ju
s
tin
g
t
h
e
h
y
p
er
p
ar
am
eter
s
to
attain
o
p
tim
al
o
u
tco
m
es.
Ad
d
itio
n
al
ly
,
th
e
in
teg
r
atio
n
o
f
s
u
p
er
r
eso
lu
tio
n
g
r
ea
tly
en
h
an
ce
d
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
VGG1
6
m
o
d
el,
r
esu
ltin
g
in
a
n
o
ta
b
le
in
cr
ea
s
e
i
n
clas
s
if
icatio
n
ac
cu
r
ac
y
.
Sp
ec
if
ically
,
th
e
ac
cu
r
ac
y
im
p
r
o
v
ed
f
r
o
m
7
1
.
3
0
%
with
o
u
t su
p
er
r
eso
lu
tio
n
to
9
6
.
3
0
% with
s
u
p
er
r
eso
lu
tio
n
.
Su
b
s
eq
u
en
tly
,
th
is
r
esear
ch
p
r
esen
ts
th
e
p
o
s
s
ib
ilit
y
o
f
i
n
teg
r
atin
g
E
SP
C
N
an
d
SR
R
esNet
in
to
alter
n
ativ
e
d
ee
p
lear
n
i
n
g
f
r
a
m
ewo
r
k
s
.
I
n
co
r
p
o
r
atin
g
th
es
e
two
s
u
p
er
-
r
eso
lu
tio
n
ap
p
r
o
ac
h
es
in
to
a
m
o
r
e
co
m
p
r
eh
e
n
s
iv
e
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
ca
n
s
ig
n
if
ican
tl
y
en
h
an
ce
p
ictu
r
e
q
u
ality
in
m
an
y
ap
p
licatio
n
s
ce
n
ar
io
s
,
ex
ten
d
in
g
b
e
y
o
n
d
th
e
s
co
p
e
o
f
f
ac
e
m
ask
class
if
ier
s
.
Hen
ce
,
th
e
f
o
r
th
co
m
i
n
g
o
b
jectiv
es
in
v
o
lv
e
in
v
esti
g
atin
g
th
e
p
r
o
s
p
ec
tiv
e
u
s
es
o
f
E
SP
C
N
an
d
S
R
R
e
s
Net
in
ad
d
itio
n
al
d
ee
p
lear
n
i
n
g
f
r
am
ewo
r
k
s
,
s
o
ex
p
an
d
i
n
g
th
e
b
e
n
ef
icial
in
f
lu
en
ce
o
f
th
is
s
tu
d
y
ac
r
o
s
s
m
an
y
f
ield
s
o
f
im
a
g
e
p
r
o
c
ess
in
g
an
d
p
atter
n
r
ec
o
g
n
itio
n
.
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
e
r
esear
ch
p
r
esen
ted
in
th
i
s
p
ap
er
was
m
ad
e
p
o
s
s
ib
le
th
r
o
u
g
h
th
e
f
in
a
n
cial
s
u
p
p
o
r
t
p
r
o
v
id
ed
b
y
B
in
a
Nu
s
an
tar
a
Un
iv
er
s
ity
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Mo
h
am
m
ad
Fais
al
R
if
tiar
r
asy
id
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
R
ico
Halim
✓
✓
✓
✓
✓
✓
✓
An
d
ien
Dwi
No
v
ik
a
✓
✓
✓
✓
✓
✓
Am
alia
Z
ah
r
a
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
atasets
u
s
ed
in
th
is
s
tu
d
y
ar
e
p
u
b
licly
av
ailab
le
an
d
ca
n
b
e
ac
ce
s
s
ed
as f
o
llo
ws:
−
Face
m
ask
d
etec
tio
n
d
ataset
:
t
h
e
d
ataset
f
o
r
f
ac
e
m
ask
d
etec
tio
n
,
co
n
tain
in
g
an
n
o
tated
i
m
ag
es
o
f
th
r
e
e
class
es
:
in
d
iv
id
u
als
wea
r
in
g
m
ask
,
wea
r
in
g
it
in
co
r
r
ec
tly
a
n
d
with
o
u
t
wea
r
in
g
f
ac
e
m
as
k
s
,
is
av
ailab
le
o
n
Kag
g
le
at
h
ttp
s
://www.
k
ag
g
le.
co
m
/d
atasets
/an
d
r
ewm
v
d
/f
ac
e
-
m
ask
-
d
etec
tio
n
.
−
Su
p
e
r
-
r
eso
lu
tio
n
d
ataset:
th
e
co
r
e
d
ataset
f
o
r
im
a
g
e
s
u
p
er
-
r
eso
lu
tio
n
is
th
e
B
er
k
eley
Seg
m
en
tatio
n
Data
s
et
5
0
0
(
B
SD5
0
0
d
ataset)
in
tr
o
d
u
ce
d
b
y
Ma
r
tin
et
al.
(
2
0
0
1
)
,
wh
ich
ca
n
b
e
ac
ce
s
s
ed
th
r
o
u
g
h
th
e
B
e
r
k
e
l
e
y
C
o
m
p
u
t
e
r
V
i
s
i
o
n
P
r
o
j
e
c
t
w
e
b
s
i
t
e
:
h
t
t
p
s
:
/
/
w
w
w
2
.
e
e
c
s
.
b
e
r
k
e
l
e
y
.
e
d
u
/
R
e
s
e
a
r
c
h
/
P
r
o
j
e
c
t
s
/
C
S
/
v
i
s
i
o
n
/
b
s
d
s
/
.
−
I
n
ad
d
itio
n
,
th
e
s
tu
d
y
in
co
r
p
o
r
ates
6
0
0
s
u
p
p
lem
en
tar
y
im
ag
es
r
an
d
o
m
l
y
s
o
u
r
ce
d
f
r
o
m
th
e
f
r
ee
s
to
ck
p
h
o
to
web
s
ite
Pex
els (
h
ttp
s
://www.
p
ex
els.co
m
)
to
in
c
r
ea
s
e
d
ataset
d
iv
er
s
ity
an
d
r
o
b
u
s
tn
e
s
s
.
All d
ata
wer
e
u
s
ed
i
n
ac
co
r
d
a
n
ce
with
th
e
ter
m
s
an
d
co
n
d
itio
n
s
s
p
ec
if
ied
b
y
th
e
r
esp
ec
tiv
e
d
ata
s
o
u
r
ce
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
5
2
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
9
,
No
.
1
,
Ju
ly
20
25
:
634
-
6
4
3
642
RE
F
E
R
E
NC
E
S
[
1
]
K
.
A
l
me
z
h
g
h
w
i
,
S
.
S
e
r
t
e
,
a
n
d
F
.
A
l
-
Tu
r
j
m
a
n
,
“
C
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
t
h
e
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
c
h
e
st
X
-
r
a
y
s
i
n
t
h
e
I
o
T
e
r
a
,
”
M
u
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
0
,
n
o
.
1
9
,
p
p
.
2
9
0
5
1
–
2
9
0
6
5
,
Ju
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
0
4
2
-
021
-
1
0
9
0
7
-
y.
[
2
]
M
.
P
r
e
t
i
,
F
.
V
e
r
h
e
g
g
e
n
,
a
n
d
S
.
A
n
g
e
l
i
,
“
I
n
sec
t
p
e
s
t
mo
n
i
t
o
r
i
n
g
w
i
t
h
c
a
m
e
r
a
-
e
q
u
i
p
p
e
d
t
r
a
p
s:
st
r
e
n
g
t
h
s
a
n
d
l
i
mi
t
a
t
i
o
n
s
,
”
J
o
u
rn
a
l
o
f
Pe
st
S
c
i
e
n
c
e
,
v
o
l
.
9
4
,
n
o
.
2
,
p
p
.
2
0
3
–
2
1
7
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
3
4
0
-
0
2
0
-
0
1
3
0
9
-
4.
[
3
]
L.
F
u
,
F
.
G
a
o
,
J.
W
u
,
R
.
Li
,
M
.
K
a
r
k
e
e
,
a
n
d
Q
.
Z
h
a
n
g
,
“
A
p
p
l
i
c
a
t
i
o
n
o
f
c
o
n
s
u
mer
R
G
B
-
D
c
a
mer
a
s
f
o
r
f
r
u
i
t
d
e
t
e
c
t
i
o
n
a
n
d
l
o
c
a
l
i
z
a
t
i
o
n
i
n
f
i
e
l
d
:
a
c
r
i
t
i
c
a
l
r
e
v
i
e
w
,
”
C
o
m
p
u
t
e
rs
a
n
d
E
l
e
c
t
r
o
n
i
c
s
i
n
Ag
r
i
c
u
l
t
u
re
,
v
o
l
.
1
7
7
,
p
.
1
0
5
6
8
7
,
O
c
t
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
a
g
.
2
0
2
0
.
1
0
5
6
8
7
.
[
4
]
Y
.
X
i
e
t
a
l
.
,
“
D
R
L
-
G
A
N
:
d
u
a
l
-
s
t
r
e
a
m
r
e
p
r
e
s
e
n
t
a
t
i
o
n
l
e
a
r
n
i
n
g
G
A
N
f
o
r
l
o
w
-
r
e
so
l
u
t
i
o
n
i
ma
g
e
c
l
a
ssi
f
i
c
a
t
i
o
n
i
n
U
A
V
a
p
p
l
i
c
a
t
i
o
n
s,
”
I
EEE
J
o
u
r
n
a
l
o
f
S
e
l
e
c
t
e
d
T
o
p
i
c
s
i
n
A
p
p
l
i
e
d
E
a
rt
h
O
b
ser
v
a
t
i
o
n
s
a
n
d
Re
m
o
t
e
S
e
n
si
n
g
,
v
o
l
.
1
4
,
p
p
.
1
7
0
5
–
1
7
1
6
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
JS
TA
R
S
.
2
0
2
0
.
3
0
4
3
1
0
9
.
[
5
]
V
.
M
a
g
o
u
l
i
a
n
i
t
i
s,
D
.
A
t
a
l
o
g
l
o
u
,
A
.
D
i
mo
u
,
D
.
Za
r
p
a
l
a
s,
a
n
d
P
.
D
a
r
a
s,
“
D
o
e
s
d
e
e
p
su
p
e
r
-
r
e
s
o
l
u
t
i
o
n
e
n
h
a
n
c
e
UAV
d
e
t
e
c
t
i
o
n
?
,
”
i
n
2
0
1
9
1
6
t
h
I
EEE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
A
d
v
a
n
c
e
d
V
i
d
e
o
a
n
d
S
i
g
n
a
l
Ba
s
e
d
S
u
rve
i
l
l
a
n
c
e
(
AV
S
S
)
,
S
e
p
.
2
0
1
9
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
A
V
S
S
.
2
0
1
9
.
8
9
0
9
8
6
5
.
[
6
]
H
.
Y
a
o
,
R
.
Q
i
n
,
a
n
d
X
.
C
h
e
n
,
“
U
n
m
a
n
n
e
d
a
e
r
i
a
l
v
e
h
i
c
l
e
f
o
r
r
e
m
o
t
e
se
n
si
n
g
a
p
p
l
i
c
a
t
i
o
n
s
-
a
r
e
v
i
e
w
,
”
R
e
m
o
t
e
S
e
n
si
n
g
,
v
o
l
.
1
1
,
n
o
.
1
2
,
p
.
1
4
4
3
,
Ju
n
.
2
0
1
9
,
d
o
i
:
1
0
.
3
3
9
0
/
r
s1
1
1
2
1
4
4
3
.
[
7
]
Y
.
K
.
O
o
i
a
n
d
H
.
I
b
r
a
h
i
m,
“
D
e
e
p
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms
f
o
r
si
n
g
l
e
i
m
a
g
e
su
p
e
r
-
r
e
s
o
l
u
t
i
o
n
:
a
sy
s
t
e
m
a
t
i
c
r
e
v
i
e
w
,
”
E
l
e
c
t
r
o
n
i
c
s
(
S
w
i
t
zer
l
a
n
d
)
,
v
o
l
.
1
0
,
n
o
.
7
,
p
.
8
6
7
,
A
p
r
.
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s
1
0
0
7
0
8
6
7
.
[
8
]
J.
S
h
e
r
me
y
e
r
a
n
d
A
.
V
a
n
Et
t
e
n
,
“
T
h
e
e
f
f
e
c
t
s
o
f
s
u
p
e
r
-
r
e
so
l
u
t
i
o
n
o
n
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
p
e
r
f
o
r
ma
n
c
e
i
n
sa
t
e
l
l
i
t
e
i
ma
g
e
r
y
,
”
i
n
2
0
1
9
I
EEE/
C
VF
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
Pa
t
t
e
rn
R
e
c
o
g
n
i
t
i
o
n
W
o
rks
h
o
p
s
(
C
VP
RW
)
,
Ju
n
.
2
0
1
9
,
v
o
l
.
2
0
1
9
-
Ju
n
e
,
p
p
.
1
4
3
2
–
1
4
4
1
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
W
.
2
0
1
9
.
0
0
1
8
4
.
[
9
]
A
.
D
e
s
h
p
a
n
d
e
,
V
.
V
.
Es
t
r
e
l
a
,
a
n
d
P
.
P
a
t
a
v
a
r
d
h
a
n
,
“
T
h
e
D
C
T
-
C
N
N
-
R
e
sN
e
t
5
0
a
r
c
h
i
t
e
c
t
u
r
e
t
o
c
l
a
ssi
f
y
b
r
a
i
n
t
u
mo
r
s
w
i
t
h
su
p
e
r
-
r
e
so
l
u
t
i
o
n
,
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
,
a
n
d
t
h
e
R
e
sN
e
t
5
0
,
”
N
e
u
ro
sci
e
n
c
e
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
1
,
n
o
.
4
,
p
.
1
0
0
0
1
3
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
r
i
.
2
0
2
1
.
1
0
0
0
1
3
.
[
1
0
]
C.
-
Y
.
Y
a
n
g
,
C
.
M
a
,
a
n
d
M
.
-
H
.
Y
a
n
g
,
“
S
i
n
g
l
e
-
i
m
a
g
e
su
p
e
r
-
r
e
s
o
l
u
t
i
o
n
:
a
b
e
n
c
h
mar
k
,
”
i
n
L
e
c
t
u
r
e
N
o
t
e
s
i
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
(
i
n
c
l
u
d
i
n
g
su
b
ser
i
e
s
L
e
c
t
u
re
N
o
t
e
s
i
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
re
N
o
t
e
s
i
n
B
i
o
i
n
f
o
rm
a
t
i
c
s)
,
v
o
l
.
8
6
9
2
LN
C
S
,
n
o
.
P
A
R
T
4
,
S
p
r
i
n
g
e
r
I
n
t
e
r
n
a
t
i
o
n
a
l
P
u
b
l
i
sh
i
n
g
,
2
0
1
4
,
p
p
.
3
7
2
–
3
8
6
.
[
1
1
]
W
.
Li
,
D
.
H
e
,
Y
.
L
i
u
,
F
.
W
a
n
g
,
a
n
d
F
.
H
u
a
n
g
,
“
S
u
p
e
r
-
r
e
so
l
u
t
i
o
n
r
e
c
o
n
s
t
r
u
c
t
i
o
n
,
r
e
c
o
g
n
i
t
i
o
n
,
a
n
d
e
v
a
l
u
a
t
i
o
n
o
f
l
a
ser
c
o
n
f
o
c
a
l
i
ma
g
e
s
o
f
h
y
p
e
r
a
c
c
u
m
u
l
a
t
o
r
S
o
l
a
n
u
m
n
i
g
r
u
m
e
n
d
o
c
y
t
o
si
s
v
e
si
c
l
e
s
b
a
s
e
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
:
c
o
m
p
a
r
a
t
i
v
e
st
u
d
y
o
f
S
R
G
A
N
a
n
d
S
R
R
e
sN
e
t
,
”
Fro
n
t
i
e
rs
i
n
P
l
a
n
t
S
c
i
e
n
c
e
,
v
o
l
.
1
4
,
M
a
r
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
8
9
/
f
p
l
s.2
0
2
3
.
1
1
4
6
4
8
5
.
[
1
2
]
C
.
L
e
d
i
g
e
t
a
l
.
,
“
P
h
o
t
o
-
r
e
a
l
i
st
i
c
s
i
n
g
l
e
i
m
a
g
e
s
u
p
e
r
-
r
e
so
l
u
t
i
o
n
u
si
n
g
a
g
e
n
e
r
a
t
i
v
e
a
d
v
e
r
sar
i
a
l
n
e
t
w
o
r
k
,
”
i
n
2
0
1
7
I
EEE
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
(
C
VP
R)
,
J
u
l
.
2
0
1
7
,
v
o
l
.
2
0
1
7
-
Jan
u
a
,
p
p
.
1
0
5
–
1
1
4
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
7
.
1
9
.
[
1
3
]
Y
.
Lu
a
n
d
Y
.
W
a
n
g
,
“
R
e
s
e
a
r
c
h
o
n
su
p
e
r
-
r
e
s
o
l
u
t
i
o
n
i
ma
g
e
p
r
o
c
e
ssi
n
g
b
a
s
e
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
,
”
i
n
2
0
2
3
6
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
N
e
t
w
o
rk,
El
e
c
t
ro
n
i
c
a
n
d
A
u
t
o
m
a
t
i
o
n
(
I
C
C
N
EA)
,
S
e
p
.
2
0
2
3
,
p
p
.
2
0
7
–
2
1
0
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
N
EA
6
0
1
0
7
.
2
0
2
3
.
0
0
0
5
2
.
[
1
4
]
W
.
Z
h
a
n
g
,
Y
.
L
i
u
,
C
.
D
o
n
g
,
a
n
d
Y
.
Q
i
a
o
,
“
R
a
n
k
S
R
G
A
N
:
s
u
p
e
r
r
e
s
o
l
u
t
i
o
n
g
e
n
e
r
a
t
i
v
e
a
d
v
e
r
sari
a
l
n
e
t
w
o
r
k
s
w
i
t
h
l
e
a
r
n
i
n
g
t
o
r
a
n
k
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
Pa
t
t
e
rn
A
n
a
l
y
si
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
4
,
n
o
.
1
0
,
p
p
.
7
1
4
9
–
7
1
6
6
,
O
c
t
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
t
p
a
m
i
.
2
0
2
1
.
3
0
9
6
3
2
7
.
[
1
5
]
L.
D
o
n
g
a
n
d
K
.
I
n
o
u
e
,
“
S
u
p
e
r
-
r
e
so
l
u
t
i
o
n
r
e
c
o
n
st
r
u
c
t
i
o
n
b
a
se
d
o
n
t
w
o
-
s
t
a
g
e
r
e
s
i
d
u
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
w
i
t
h
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
6
,
p
.
1
0
0
1
6
2
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
ml
w
a
.
2
0
2
1
.
1
0
0
1
6
2
.
[
1
6
]
W
.
S
h
i
e
t
a
l
.
,
“
R
e
a
l
-
t
i
me
si
n
g
l
e
i
ma
g
e
a
n
d
v
i
d
e
o
s
u
p
e
r
-
r
e
so
l
u
t
i
o
n
u
s
i
n
g
a
n
e
f
f
i
c
i
e
n
t
s
u
b
-
p
i
x
e
l
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k,
”
i
n
2
0
1
6
I
E
EE
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
(
C
VPR)
,
J
u
n
.
2
0
1
6
,
v
o
l
.
2
0
1
6
-
D
e
c
e
m,
p
p
.
1
8
7
4
–
1
8
8
3
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
2
0
7
.
[
1
7
]
F
.
C
a
o
a
n
d
H
.
L
i
u
,
“
S
i
n
g
l
e
i
m
a
g
e
s
u
p
e
r
-
r
e
so
l
u
t
i
o
n
v
i
a
m
u
l
t
i
-
sc
a
l
e
r
e
si
d
u
a
l
c
h
a
n
n
e
l
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
,
”
N
e
u
r
o
c
o
m
p
u
t
i
n
g
,
v
o
l
.
3
5
8
,
p
p
.
4
2
4
–
4
3
6
,
S
e
p
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.
2
0
1
9
.
0
5
.
0
6
6
.
[
1
8
]
J.
D
u
,
Q
.
Li
u
,
a
n
d
K
.
C
h
e
n
,
“
L
i
c
e
n
s
e
p
l
a
t
e
s
u
p
e
r
-
r
e
so
l
u
t
i
o
n
r
e
c
o
n
st
r
u
c
t
i
o
n
b
a
se
d
o
n
i
m
p
r
o
v
e
d
ESP
C
N
n
e
t
w
o
r
k
,
”
i
n
2
0
1
9
I
EEE
3
rd
A
d
v
a
n
c
e
d
I
n
f
o
rm
a
t
i
o
n
M
a
n
a
g
e
m
e
n
t
,
C
o
m
m
u
n
i
c
a
t
e
s,
El
e
c
t
r
o
n
i
c
a
n
d
Au
t
o
m
a
t
i
o
n
C
o
n
t
r
o
l
C
o
n
f
e
re
n
c
e
(
I
M
C
EC
)
,
O
c
t
.
2
0
1
9
,
p
p
.
1
2
2
–
1
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
M
C
E
C
4
6
7
2
4
.
2
0
1
9
.
8
9
8
4
0
5
0
.
[
1
9
]
Z.
C
h
u
e
t
a
l
.
,
“
A
g
e
n
e
r
a
l
i
z
a
b
l
e
sam
p
l
e
r
e
s
o
l
u
t
i
o
n
a
u
g
me
n
t
a
t
i
o
n
me
t
h
o
d
f
o
r
me
c
h
a
n
i
c
a
l
f
a
u
l
t
d
i
a
g
n
o
si
s
b
a
se
d
o
n
ESP
C
N
,
”
J
o
u
rn
a
l
o
f
S
e
n
s
o
rs
,
v
o
l
.
2
0
2
1
,
n
o
.
1
,
Ja
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
1
/
7
4
9
6
0
0
7
.
[
2
0
]
Y
.
W
a
n
g
,
F
.
N
i
u
,
X
.
Z
h
a
n
g
,
J.
X
i
a
o
,
a
n
d
C
.
X
u
,
“
S
u
p
e
r
r
e
so
l
u
t
i
o
n
r
e
c
o
n
s
t
r
u
c
t
i
o
n
o
f
d
i
g
i
t
a
l
c
o
r
e
i
m
a
g
e
b
a
s
e
d
o
n
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
En
e
r
g
y
Re
p
o
rt
s
,
v
o
l
.
8
,
p
p
.
8
7
–
9
4
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
g
y
r
.
2
0
2
2
.
1
0
.
0
6
9
.
[
2
1
]
Y
.
Z
h
a
n
g
,
W
.
Y
u
,
a
n
d
D
.
Z
h
u
,
“
Te
r
r
a
i
n
f
e
a
t
u
r
e
-
a
w
a
r
e
d
e
e
p
l
e
a
r
n
i
n
g
n
e
t
w
o
r
k
f
o
r
d
i
g
i
t
a
l
e
l
e
v
a
t
i
o
n
mo
d
e
l
su
p
e
r
r
e
so
l
u
t
i
o
n
,
”
I
S
PR
S
J
o
u
rn
a
l
o
f
Ph
o
t
o
g
r
a
m
m
e
t
r
y
a
n
d
Re
m
o
t
e
S
e
n
si
n
g
,
v
o
l
.
1
8
9
,
p
p
.
1
4
3
–
1
6
2
,
Ju
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
s
p
r
sj
p
r
s.
2
0
2
2
.
0
4
.
0
2
8
.
[
2
2
]
S
.
U
l
l
a
h
a
n
d
S
.
H
.
S
o
n
g
,
“
S
R
R
e
sN
e
t
p
e
r
f
o
r
m
a
n
c
e
e
n
h
a
n
c
e
m
e
n
t
u
si
n
g
p
a
t
c
h
i
n
p
u
t
s
a
n
d
p
a
r
t
i
a
l
c
o
n
v
o
l
u
t
i
o
n
-
b
a
se
d
p
a
d
d
i
n
g
,
”
C
o
m
p
u
t
e
rs
,
M
a
t
e
r
i
a
l
s
a
n
d
C
o
n
t
i
n
u
a
,
v
o
l
.
7
4
,
n
o
.
2
,
p
p
.
2
9
9
9
–
3
0
1
4
,
2
0
2
3
,
d
o
i
:
1
0
.
3
2
6
0
4
/
c
mc
.
2
0
2
3
.
0
3
2
3
2
6
.
[
2
3
]
D
.
Z
h
o
u
,
R
.
D
u
a
n
,
L
.
Z
h
a
o
,
a
n
d
X
.
C
h
a
i
,
“
S
i
n
g
l
e
i
m
a
g
e
s
u
p
e
r
-
r
e
s
o
l
u
t
i
o
n
r
e
c
o
n
st
r
u
c
t
i
o
n
b
a
se
d
o
n
mu
l
t
i
-
sca
l
e
f
e
a
t
u
r
e
ma
p
p
i
n
g
a
d
v
e
r
sar
i
a
l
n
e
t
w
o
r
k
,
”
S
i
g
n
a
l
Pro
c
e
ss
i
n
g
,
v
o
l
.
1
6
6
,
p
.
1
0
7
2
5
1
,
J
a
n
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
si
g
p
r
o
.
2
0
1
9
.
1
0
7
2
5
1
.
[
2
4
]
D
.
K
e
sr
a
r
a
t
,
K
.
Th
a
k
u
l
s
u
k
a
n
a
n
t
,
a
n
d
V
.
P
a
t
a
n
a
v
i
j
i
t
,
“
A
n
o
v
e
l
e
l
e
me
n
t
a
r
y
sp
a
t
i
a
l
e
x
p
a
n
d
i
n
g
sc
h
e
m
e
f
o
r
m
o
n
S
I
S
R
m
e
t
h
o
d
w
i
t
h
mo
d
i
f
y
i
n
g
G
e
ma
n
&
M
c
C
l
u
r
e
f
u
n
c
t
i
o
n
,
”
T
e
l
k
o
m
n
i
k
a
(
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
C
o
m
p
u
t
i
n
g
E
l
e
c
t
r
o
n
i
c
s
a
n
d
C
o
n
t
r
o
l
)
,
v
o
l
.
1
7
,
n
o
.
5
,
p
p
.
2
5
5
4
–
2
5
6
0
,
O
c
t
.
2
0
1
9
,
d
o
i
:
1
0
.
1
2
9
2
8
/
TEL
K
O
M
N
I
K
A
.
v
1
7
i
5
.
1
2
7
9
9
.
[
2
5
]
Z.
W
a
n
g
,
A
.
C
.
B
o
v
i
k
,
H
.
R
.
S
h
e
i
k
h
,
a
n
d
E.
P
.
S
i
mo
n
c
e
l
l
i
,
“
I
mag
e
q
u
a
l
i
t
y
a
ssessm
e
n
t
:
f
r
o
m
e
r
r
o
r
v
i
s
i
b
i
l
i
t
y
t
o
st
r
u
c
t
u
r
a
l
si
mi
l
a
r
i
t
y
,
”
I
EEE
T
r
a
n
sa
c
t
i
o
n
s
o
n
I
m
a
g
e
Pr
o
c
e
ssi
n
g
,
v
o
l
.
1
3
,
n
o
.
4
,
p
p
.
6
0
0
–
6
1
2
,
A
p
r
.
2
0
0
4
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
P
.
2
0
0
3
.
8
1
9
8
6
1
.
[
2
6
]
S
.
M
a
sc
a
r
e
n
h
a
s
a
n
d
M
.
A
g
a
r
w
a
l
,
“
A
c
o
mp
a
r
i
s
o
n
b
e
t
w
e
e
n
V
G
G
1
6
,
V
G
G
1
9
a
n
d
R
e
sN
e
t
5
0
a
r
c
h
i
t
e
c
t
u
r
e
f
r
a
m
e
w
o
r
k
s
f
o
r
i
m
a
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
i
n
2
0
2
1
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
D
i
sru
p
t
i
v
e
T
e
c
h
n
o
l
o
g
i
e
s
f
o
r
M
u
l
t
i
-
D
i
sc
i
p
l
i
n
a
ry
R
e
se
a
rc
h
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
(
C
EN
T
C
O
N
)
,
N
o
v
.
2
0
2
1
,
p
p
.
9
6
–
9
9
,
d
o
i
:
1
0
.
1
1
0
9
/
C
EN
T
C
O
N
5
2
3
4
5
.
2
0
2
1
.
9
6
8
7
9
4
4
.
[
2
7
]
Y
.
H
a
m
i
d
,
S
.
W
a
n
i
,
A
.
B
.
S
o
o
mr
o
,
A
.
A
.
A
l
w
a
n
,
a
n
d
Y
.
G
u
l
z
a
r
,
“
S
m
a
r
t
s
e
e
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
sy
st
e
m
b
a
s
e
d
o
n
M
o
b
i
l
e
N
e
t
V
2
a
r
c
h
i
t
e
c
t
u
r
e
,
”
i
n
2
0
2
2
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
i
n
g
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
(
I
C
C
I
T
)
,
J
a
n
.
2
0
2
2
,
p
p
.
2
1
7
–
2
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
I
T5
2
4
1
9
.
2
0
2
2
.
9
7
1
1
6
6
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4
7
5
2
Lo
w
-
r
eso
lu
tio
n
ima
g
e
q
u
a
lity
en
h
a
n
ce
me
n
t u
s
in
g
en
h
a
n
ce
d
…
(
Mo
h
a
mma
d
F
a
is
a
l R
ifti
a
r
r
a
s
yid
)
643
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Mo
h
a
m
m
a
d
Fa
is
a
l
Rifti
a
r
r
a
sy
id
is
a
g
ra
d
u
a
te
’
s
stu
d
e
n
t
a
t
Bin
a
Nu
sa
n
tara
Un
iv
e
rsity
Do
c
to
ra
l
P
ro
g
ra
m
,
I
n
d
o
n
e
sia
.
He
m
a
jo
re
d
in
d
o
c
to
r
o
f
c
o
m
p
u
ter
sc
ien
c
e
wit
h
e
x
p
e
rien
c
e
in
c
o
m
p
u
ter
v
isio
n
a
n
d
r
o
b
o
ti
c
s
re
se
a
rc
h
wh
e
n
p
u
rsu
i
n
g
h
is
b
a
c
h
e
lo
r
’
s
d
e
g
re
e
a
n
d
e
n
h
a
n
c
e
h
is
e
x
p
e
rien
c
e
i
n
in
ter
n
e
t
o
f
t
h
in
g
s
w
h
e
n
p
u
rsu
in
g
h
is
m
a
ste
r
’
s
d
e
g
re
e
.
Afte
r
fin
ish
i
n
g
h
is
b
a
c
h
e
lo
r
’
s
d
e
g
re
e
i
n
2
0
1
8
a
n
d
m
a
ste
r
’
s
d
e
g
re
e
in
2
0
2
4
,
h
e
d
e
c
id
e
d
t
o
p
u
rsu
e
h
i
g
h
e
r
e
d
u
c
a
ti
o
n
to
a
c
h
iev
e
a
c
a
re
e
r
in
th
e
e
d
u
c
a
ti
o
n
a
l
field
.
His
re
se
a
rc
h
a
re
a
s
a
r
e
in
tern
e
t
o
f
th
i
n
g
s
a
n
d
ima
g
e
p
r
o
c
e
ss
in
g
.
His
fiel
d
o
f
stu
d
y
is i
n
s
p
ired
b
y
h
is i
n
tere
st i
n
tec
h
n
o
lo
g
ica
l
a
d
v
a
n
c
e
s in
ro
b
o
ti
c
s.
He
e
n
v
isi
o
n
s
t
o
a
c
h
i
e
v
e
a
n
e
x
p
e
rti
se
in
d
r
o
n
e
d
e
v
e
lo
p
m
e
n
t
a
n
d
d
ro
n
e
p
a
t
h
a
u
to
m
a
ti
o
n
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
o
h
a
m
m
a
d
.
rift
iarra
sy
id
@b
in
u
s.a
c
.
id
.
Ric
o
H
a
li
m
is
a
m
a
ste
r
’
s
stu
d
e
n
t
a
t
Bi
n
a
Nu
sa
n
tara
Un
iv
e
rsity
,
In
d
o
n
e
sia
.
He
m
a
jo
re
d
in
c
o
m
p
u
ter
sc
ien
c
e
with
e
x
p
e
rien
c
e
i
n
c
o
m
p
u
ter
v
isi
o
n
s
re
se
a
rc
h
wh
e
n
p
u
rsu
in
g
h
is
b
a
c
h
e
lo
r
’
s
d
e
g
re
e
.
Afte
r
fi
n
i
sh
in
g
h
is
b
a
c
h
e
lo
r
’
s
d
e
g
re
e
in
2
0
1
8
,
h
e
d
e
c
id
e
d
t
o
p
u
rs
u
e
h
ig
h
e
r
e
d
u
c
a
ti
o
n
to
a
c
h
ie
v
e
a
c
a
re
e
r
in
th
e
e
d
u
c
a
ti
o
n
a
l
field
.
His
re
se
a
rc
h
a
re
a
s
a
re
d
e
e
p
lea
rn
in
g
,
a
n
d
c
o
m
p
u
ter
v
isio
n
.
I
n
sp
ired
b
y
h
is
i
n
tere
st
i
n
ima
g
e
s,
he
a
ims
to
p
u
rsu
e
a
c
a
re
e
r
in
e
d
u
c
a
ti
o
n
,
d
ri
v
e
n
b
y
h
is
d
e
d
i
c
a
ti
o
n
t
o
a
d
v
a
n
c
in
g
th
e
se
field
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
rico
.
h
a
li
m
@b
i
n
u
s.a
c
.
i
d
.
And
ien
Dw
i
No
v
ik
a
is
a
lec
tu
re
r
in
th
e
c
o
m
p
u
ter
sc
ien
c
e
p
ro
g
ra
m
a
t
Bin
a
Nu
sa
n
tara
Un
iv
e
rsity
,
h
a
v
in
g
c
o
m
p
lete
d
h
e
r
m
a
ste
r
’
s
d
e
g
re
e
in
c
o
m
p
u
ter
sc
ien
c
e
th
e
re
in
Ju
n
e
2
0
2
4
.
S
h
e
p
re
v
i
o
u
sl
y
e
a
rn
e
d
h
e
r
b
a
c
h
e
lo
r
’
s
d
e
g
re
e
i
n
c
o
m
p
u
ter
sc
ien
c
e
fro
m
Dip
o
n
e
g
o
r
o
Un
iv
e
rsity
,
b
ri
n
g
in
g
a
so
li
d
f
o
u
n
d
a
ti
o
n
i
n
a
c
a
d
e
m
ic
k
n
o
wle
d
g
e
a
n
d
p
ra
c
ti
c
a
l
e
x
p
e
rien
c
e
.
He
r
p
ro
fe
ss
io
n
a
l
b
a
c
k
g
r
o
u
n
d
in
c
lu
d
e
s
ro
les
a
s
a
n
in
teg
ra
ti
o
n
sta
ff
a
t
Nic
e
P
a
y
,
a
fu
ll
sta
c
k
d
e
v
e
lo
p
e
r
in
ter
n
a
t
M
a
v
e
n
Di
g
i
tal
As
ia,
a
n
d
a
b
a
c
k
-
e
n
d
d
e
v
e
l
o
p
e
r
i
n
tern
a
t
G
a
g
a
s
Cip
ta
P
e
rsa
d
a
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
at
e
m
a
il
:
a
n
d
ien
.
n
o
v
ik
a
@b
in
u
s.a
c
.
id
.
Am
a
li
a
Z
a
h
r
a
is
a
lec
tu
re
r
a
t
th
e
m
a
ste
r
o
f
i
n
fo
rm
a
ti
o
n
tec
h
n
o
l
o
g
y
,
B
in
a
Nu
sa
n
tara
Un
iv
e
rsity
,
In
d
o
n
e
sia
.
S
h
e
re
c
e
iv
e
d
h
e
r
b
a
c
h
e
l
o
r
’
s
d
e
g
re
e
in
c
o
m
p
u
ter
sc
ien
c
e
fro
m
th
e
F
a
c
u
lt
y
o
f
C
o
m
p
u
ter
S
c
ien
c
e
,
Un
iv
e
rsit
y
o
f
I
n
d
o
n
e
sia
(UI)
i
n
2
0
0
8
.
S
h
e
d
o
e
s
n
o
t
h
a
v
e
a
m
a
ste
r
’
s
d
e
g
re
e
.
He
r
P
h
.
D.
w
a
s
o
b
tain
e
d
fr
o
m
th
e
S
c
h
o
o
l
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
In
fo
rm
a
ti
c
s,
Un
i
v
e
rsity
C
o
ll
e
g
e
Du
b
li
n
(UCD
),
Ire
lan
d
in
2
0
1
4
.
He
r
re
se
a
r
c
h
in
tere
sts
c
o
v
e
r
v
a
rio
u
s
field
s
i
n
s
p
e
e
c
h
tec
h
n
o
l
o
g
y
,
su
c
h
a
s
sp
e
e
c
h
re
c
o
g
n
it
i
o
n
,
s
p
o
k
e
n
lan
g
u
a
g
e
id
e
n
ti
fica
ti
o
n
,
sp
e
a
k
e
r
v
e
rifi
c
a
ti
o
n
,
sp
e
e
c
h
e
m
o
t
io
n
re
c
o
g
n
i
ti
o
n
,
a
n
d
s
o
o
n
.
Ad
d
it
i
o
n
a
ll
y
,
sh
e
a
lso
h
a
s
in
tere
st
in
n
a
t
u
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
(NLP
),
c
o
m
p
u
tatio
n
a
l
li
n
g
u
isti
c
s,
m
a
c
h
in
e
lea
rn
in
g
,
a
n
d
a
rti
ficia
l
i
n
telli
g
e
n
c
e
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
m
a
li
a
.
z
a
h
ra
@b
in
u
s.e
d
u
.
Evaluation Warning : The document was created with Spire.PDF for Python.