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.
37
,
No
.
2
,
Feb
r
u
ar
y
20
25
,
p
p
.
903
~
912
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
37
.
i
2
.
p
p
903
-
9
1
2
903
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
SRCNN
-
ba
sed im
a
g
e t
ra
nsm
iss
io
n f
o
r auto
no
mo
us v
ehicles in
limited n
etwork
area
s
Anin
dy
a
Af
ina
Ca
rm
ely
a
1
,
Arie
f
Su
ry
a
di Sa
t
y
a
wa
n
2
,3
,
G
a
lura
M
uh
a
m
m
a
d Sura
neg
a
ra
1
,
M
o
k
ha
m
a
m
a
d M
irza
E
t
nis
a
H
a
qiq
i
4
,
H
elf
y
Su
s
ila
wa
t
i
5
,
Niza
r
Ala
m
H
a
m
da
ni
5
,
P
up
ut
Da
ni P
ra
s
et
y
o
Adi
2
1
D
e
p
a
r
t
me
n
t
o
f
Te
l
e
c
o
m
mu
n
i
c
a
t
i
o
n
S
y
st
e
m,
U
n
i
v
e
r
s
i
t
a
s Pe
n
d
i
d
i
k
a
n
I
n
d
o
n
e
si
a
,
P
u
r
w
a
k
a
r
t
a
,
I
n
d
o
n
e
si
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
a
s
N
u
r
t
a
n
i
o
,
B
a
n
d
u
n
g
,
I
n
d
o
n
e
s
i
a
3
N
a
t
i
o
n
a
l
R
e
se
a
r
c
h
a
n
d
I
n
n
o
v
a
t
i
o
n
A
g
e
n
c
y
,
Ja
k
a
r
t
a
,
I
n
d
o
n
e
s
i
a
4
D
e
p
a
r
t
me
n
t
o
f
E
n
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
s
i
t
a
s I
n
d
o
n
e
s
i
a
,
D
e
p
o
k
,
I
n
d
o
n
e
si
a
5
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
a
s G
a
r
u
t
,
G
a
r
u
t
,
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
J
u
l
2
,
2
0
2
4
R
ev
is
ed
Sep
17
,
2
0
2
4
Acc
ep
ted
Sep
30
,
2
0
2
4
Hig
h
-
q
u
a
li
t
y
ima
g
e
s
a
re
c
ru
c
ia
l
fo
r
n
a
v
ig
a
ti
o
n
,
o
b
sta
c
le
d
e
tec
ti
o
n
,
a
n
d
e
n
v
iro
n
m
e
n
tal
u
n
d
e
rsta
n
d
i
n
g
,
b
u
t
tran
sm
it
ti
n
g
h
i
g
h
-
re
so
l
u
ti
o
n
im
a
g
e
s
o
v
e
r
c
o
n
stra
in
e
d
n
e
two
rk
s
p
re
se
n
ts
sig
n
ifi
c
a
n
t
c
h
a
ll
e
n
g
e
s.
T
h
is
stu
d
y
in
tro
d
u
c
e
s
a
n
ima
g
e
tran
sm
issio
n
sy
ste
m
u
sin
g
su
p
e
r
-
re
so
lu
ti
o
n
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
s
(S
RCNN
)
t
o
e
n
h
a
n
c
e
ima
g
e
q
u
a
li
ty
wit
h
o
u
t
i
n
c
re
a
sin
g
b
a
n
d
wi
d
th
re
q
u
irem
e
n
ts
b
y
tran
sm
it
ti
n
g
lo
w
-
re
so
lu
ti
o
n
ima
g
e
s
a
n
d
u
p
sc
a
li
n
g
th
e
m
with
S
RCNN
.
Th
e
f
irst
p
h
a
se
o
f
th
e
r
e
se
a
rc
h
in
v
o
l
v
e
d
d
a
ta
c
o
ll
e
c
ti
o
n
,
in
wh
i
c
h
in
fo
rm
a
ti
o
n
wa
s
a
c
q
u
ired
d
irec
tl
y
fro
m
a
n
a
p
p
ro
p
riate
lo
c
u
s
t
o
p
ro
d
u
c
e
train
in
g
,
v
a
li
d
a
ti
o
n
,
a
n
d
tes
ti
n
g
d
a
tas
e
ts.
Th
e
se
c
o
n
d
,
t
h
re
e
S
RCN
N
m
o
d
e
ls
(9
1
5
,
9
3
5
,
a
n
d
9
5
5
)
we
re
train
e
d
u
sin
g
su
c
h
a
train
in
g
d
a
tas
e
t.
Th
e
las
t
wa
s
a
n
e
v
a
lu
a
ti
o
n
,
in
w
h
ich
m
o
d
e
l
9
1
5
sh
o
we
d
q
u
ick
lea
rn
in
g
a
n
d
sta
b
le
p
e
rfo
rm
a
n
c
e
with
in
it
ial
h
ig
h
lo
s
s,
wh
il
e
m
o
d
e
l
9
3
5
h
a
d
ra
p
i
d
c
o
n
v
e
rg
e
n
c
e
b
u
t
p
o
ten
ti
a
l
o
v
e
rfit
t
in
g
.
M
o
d
e
l
9
5
5
a
c
h
iev
e
d
h
i
g
h
in
it
ial
p
e
rf
o
rm
a
n
c
e
.
Th
re
e
S
RCNN
m
o
d
e
l
c
o
n
f
ig
u
ra
ti
o
n
s
we
re
tailo
re
d
t
o
t
h
e
sp
e
c
ifi
c
n
e
e
d
s
o
f
a
u
to
n
o
m
o
u
s
e
lec
tri
c
v
e
h
icle
s
o
p
e
ra
ti
n
g
i
n
li
m
it
e
d
a
re
a
s,
su
c
h
a
s
th
e
lo
c
u
s.
In
p
u
t
ima
g
e
re
so
lu
ti
o
n
ra
n
g
e
d
fr
o
m
1
2
8
×
1
2
8
p
i
x
e
ls t
o
2
5
6
×
2
5
6
p
i
x
e
ls,
wh
il
e
o
u
t
p
u
t
re
so
l
u
ti
o
n
v
a
ried
fro
m
2
5
6
×
2
5
6
p
ix
e
ls
t
o
5
1
2
×
5
1
2
p
ix
e
ls.
Th
e
se
re
so
lu
ti
o
n
s
c
a
n
b
e
a
c
c
e
p
tab
le
fo
r
e
fficie
n
t
ima
g
e
tran
sm
issio
n
o
v
e
r
IEE
E
8
0
2
.
1
1
a
c
,
b
u
t
o
n
t
h
e
lo
n
g
ra
n
g
e
(
Lo
Ra
)
n
e
two
rk
,
it
stil
l
p
ro
d
u
c
e
s
so
m
e
d
e
lay
.
K
ey
w
o
r
d
s
:
Au
to
n
o
m
o
u
s
v
eh
icles
Hig
h
-
r
eso
lu
tio
n
im
a
g
es
I
m
ag
e
tr
an
s
m
is
s
io
n
Netwo
r
k
ef
f
icien
cy
SR
C
NN
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
in
d
y
a
Af
in
a
C
ar
m
ely
a
Dep
ar
tm
en
t o
f
T
elec
o
m
m
u
n
ic
atio
n
Sy
s
tem
,
Un
iv
er
s
itas
Pen
d
id
ik
an
I
n
d
o
n
esia
Pu
r
wak
ar
ta,
W
est J
av
a,
4
1
1
1
5
,
I
n
d
o
n
esia
E
m
ail: a
n
in
d
y
af
i
n
a@
u
p
i.e
d
u
1.
I
NT
RO
D
UCT
I
O
N
E
f
f
ec
tiv
e
im
ag
e
tr
an
s
m
is
s
io
n
h
as
b
ec
o
m
e
ess
en
tial
with
th
e
q
u
ick
d
ev
el
o
p
m
e
n
t
o
f
au
to
n
o
m
o
u
s
ca
r
tech
n
o
lo
g
y
,
p
ar
ticu
la
r
ly
in
p
la
ce
s
with
s
p
o
tty
n
etwo
r
k
co
v
er
ag
e.
Hig
h
-
q
u
ality
im
ag
es a
r
e
e
s
s
en
tial f
o
r
v
ar
io
u
s
task
s
s
u
ch
as
n
a
v
ig
atio
n
,
o
b
s
tacle
d
etec
tio
n
,
an
d
en
v
ir
o
n
m
e
n
t
u
n
d
er
s
tan
d
in
g
[
1
]
–
[
5
]
.
Ho
wev
er
,
tr
a
n
s
m
itti
n
g
h
ig
h
-
r
eso
lu
tio
n
im
ag
es
o
v
er
co
n
s
tr
ain
ed
n
etwo
r
k
s
p
o
s
es
s
ig
n
if
ican
t
ch
allen
g
es.
T
h
is
co
n
d
itio
n
is
b
ec
au
s
e
th
ese
im
ag
es
co
n
s
u
m
e
m
ass
iv
e
b
an
d
wid
th
tr
a
n
s
m
is
s
io
n
.
Fo
r
in
s
tan
ce
,
in
o
u
r
d
ev
elo
p
m
en
t
o
f
an
au
t
o
n
o
m
o
u
s
v
eh
icle
o
p
er
ated
in
a
lim
ited
ar
ea
u
s
in
g
I
E
E
E
8
0
2
.
1
1
ac
,
wi
th
an
8
0
MH
z
c
h
an
n
el,
2
s
p
atial
s
tr
ea
m
s
,
an
d
in
id
ea
l
co
n
d
itio
n
s
,
th
e
th
eo
r
etica
l
m
ax
im
u
m
b
an
d
wid
th
is
8
6
6
Mb
p
s
.
T
h
ey
ar
e
ac
co
u
n
tin
g
f
o
r
o
v
er
h
ea
d
an
d
ty
p
ical
n
etwo
r
k
co
n
d
itio
n
s
.
T
h
is
co
n
d
itio
n
m
ig
h
t
ac
h
iev
e
6
0
-
7
0
%
o
f
th
e
t
h
eo
r
etica
l
m
a
x
im
u
m
,
o
r
ar
o
u
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
2
,
Feb
r
u
a
r
y
20
2
5
:
9
0
3
-
9
1
2
904
520
-
6
0
6
Mb
p
s
[
6
]
,
[
7
]
.
T
h
is
b
an
d
wid
th
ca
n
ea
s
ily
s
u
p
p
o
r
t
m
u
ltip
le
s
im
u
ltan
eo
u
s
HD
o
r
UHD
v
id
eo
s
tr
ea
m
s
.
An
o
th
er
o
p
tio
n
is
em
p
lo
y
in
g
th
e
lo
n
g
r
an
g
e
(
L
o
R
a
)
s
tan
d
ar
d
,
wh
ich
o
p
er
ates
with
b
an
d
w
id
th
s
r
an
g
in
g
f
r
o
m
1
2
5
t
o
5
0
0
k
Hz.
T
h
e
m
a
x
im
u
m
d
ata
r
ate
o
f
L
o
R
a
is
o
n
ly
5
%
o
f
th
e
m
in
im
u
m
r
eq
u
ir
e
m
en
t
f
o
r
SD
v
id
e
o
(
7
2
0
×
4
8
0
p
ix
els),
5
0
k
b
p
s
[
8
]
–
[
1
1
]
.
Ou
r
wo
r
k
o
f
f
e
r
s
a
co
m
p
r
eh
e
n
s
iv
e
m
eth
o
d
b
ased
o
n
s
u
p
er
-
r
eso
lu
tio
n
co
n
v
o
lu
tio
n
al
n
eu
r
a
l
n
etwo
r
k
s
(
SR
C
N
N)
f
o
r
en
h
an
cin
g
p
i
ctu
r
e
tr
an
s
m
is
s
io
n
in
au
to
n
o
m
o
u
s
ca
r
s
.
T
h
is
tech
n
iq
u
e
m
in
im
izes
d
ata
tr
an
s
m
is
s
io
n
lo
ad
s
wh
ile
m
ain
tain
in
g
im
p
o
r
tan
t
v
is
u
al
in
f
o
r
m
atio
n
b
y
tr
an
s
m
itti
n
g
l
o
w
-
q
u
ality
im
ag
es
an
d
u
p
s
ca
lin
g
th
em
to
h
ig
h
r
eso
l
u
tio
n
o
n
th
e
v
eh
icle'
s
en
d
,
esp
ec
ially
in
lim
ited
o
r
u
n
s
tab
le
n
etwo
r
k
co
n
d
itio
n
s
.
B
etter
co
o
r
d
in
atio
n
a
n
d
d
ec
is
i
o
n
-
m
ak
i
n
g
ar
e
e
n
co
u
r
ag
ed
b
y
th
is
ap
p
r
o
ac
h
,
wh
ich
im
p
r
o
v
es
p
ictu
r
e
q
u
ality
f
o
r
au
to
n
o
m
o
u
s
v
eh
icles
an
d
f
ac
i
litates
co
m
m
u
n
icatio
n
b
etwe
en
v
eh
icles
an
d
ce
n
tr
al
o
p
er
ati
o
n
s
.
Ou
r
m
eth
o
d
,
wh
ich
u
s
e
SR
C
NN
to
u
p
s
ca
le
s
m
aller
in
p
u
t
im
ag
es
in
co
n
tr
ast
to
ea
r
lier
r
esear
ch
[
1
2
]
,
s
h
o
ws
p
r
o
m
is
e
in
a
r
an
g
e
o
f
im
ag
e
tr
a
n
s
m
is
s
io
n
s
ce
n
ar
io
s
with
in
th
e
au
t
o
n
o
m
o
u
s
ca
r
ec
o
s
y
s
tem
.
2.
M
E
T
H
O
D
W
e
p
r
esen
t
a
w
o
r
k
ab
le
s
o
lu
tio
n
f
o
r
ef
f
icien
t
p
ictu
r
e
tr
a
n
s
m
is
s
io
n
in
au
to
n
o
m
o
u
s
v
eh
icles:
SR
C
NN
.
I
t
tack
les
th
e
d
if
f
icu
lties
ass
o
ciate
d
with
s
en
d
in
g
h
ig
h
-
r
eso
lu
tio
n
p
ictu
r
es
in
b
an
d
wid
t
h
-
c
o
n
s
tr
ain
ed
s
ettin
g
s
.
B
y
in
teg
r
atin
g
SR
C
NN,
lo
w
-
r
eso
lu
tio
n
im
ag
es
ca
n
b
e
tr
an
s
m
itted
q
u
ick
ly
an
d
e
n
h
an
ce
d
in
to
h
ig
h
-
r
eso
lu
tio
n
im
ag
es
n
ec
ess
ar
y
f
o
r
p
r
ec
is
e
n
av
ig
atio
n
[
1
3
]
,
[
1
4
]
,
m
ak
in
g
o
u
r
ap
p
r
o
ac
h
v
iab
le
f
o
r
r
ea
l
-
wo
r
l
d
im
p
lem
en
tatio
n
.
I
n
Fig
u
r
e
1
,
we
p
r
esen
t
t
h
e
p
r
o
ce
s
s
o
f
tr
an
s
m
itti
n
g
im
ag
es
in
o
u
r
r
es
ea
r
ch
s
ce
n
ar
io
.
T
h
e
p
r
o
ce
s
s
o
f
im
p
r
o
v
in
g
l
o
w
-
r
e
s
o
lu
tio
n
p
h
o
t
o
s
u
s
in
g
a
SR
C
NN
,
a
cr
u
cial
co
m
p
o
n
en
t
o
f
o
u
r
a
p
p
r
o
ac
h
,
is
d
ep
icted
i
n
th
e
d
ia
g
r
am
.
A
lo
w
-
r
eso
lu
tio
n
im
a
g
e
is
f
ir
s
t
tr
an
s
m
itted
b
y
a
tr
an
s
m
it
ter
.
T
h
is
im
ag
e
is
s
u
b
s
eq
u
en
tly
tr
an
s
m
itted
to
a
r
ec
eiv
er
,
wh
er
e
its
lo
w
r
eso
lu
tio
n
is
r
etain
ed
.
Fig
u
r
e
1
.
Su
p
er
r
eso
lu
tio
n
-
b
as
ed
im
ag
e
tr
an
s
m
is
s
io
n
s
ch
em
e
T
h
e
p
r
im
ar
y
f
o
c
u
s
o
f
th
e
p
r
o
c
ess
is
to
en
h
an
ce
th
is
lo
w
-
r
es
o
lu
tio
n
im
ag
e
u
s
in
g
SR
C
NN
t
o
o
b
tain
a
h
ig
h
-
r
eso
lu
tio
n
v
er
s
io
n
.
T
h
e
SR
C
NN
p
r
o
ce
s
s
b
eg
in
s
b
y
u
p
-
s
am
p
lin
g
th
e
l
o
w
-
r
eso
lu
tio
n
im
ag
e
th
r
o
u
g
h
b
icu
b
ic
in
ter
p
o
latio
n
.
T
h
is
s
te
p
in
cr
ea
s
es
th
e
im
ag
e's
s
ize
t
o
m
atch
th
e
d
esire
d
h
ig
h
-
r
eso
lu
tio
n
d
im
en
s
io
n
s
,
s
er
v
in
g
as
a
p
r
elim
in
ar
y
en
h
a
n
ce
m
en
t.
Fo
llo
win
g
th
is
,
th
e
u
p
-
s
am
p
led
im
a
g
e
u
n
d
er
g
o
es
to
p
atch
ex
tr
ac
tio
n
an
d
r
ep
r
esen
tatio
n
.
T
h
is
is
a
ch
iev
ed
u
s
in
g
th
e
f
ir
s
t
co
n
v
o
lu
tio
n
al
lay
er
with
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U)
ac
tiv
atio
n
,
wh
ich
ex
tr
ac
ts
s
m
all
r
eg
io
n
s
o
r
p
atc
h
es
f
r
o
m
th
e
im
ag
e
an
d
r
ep
r
esen
ts
th
em
in
a
h
ig
h
er
-
d
im
en
s
io
n
al
f
ea
tu
r
e
s
p
ac
e.
N
ex
t,
th
ese
p
atch
es
ar
e
p
r
o
ce
s
s
ed
th
r
o
u
g
h
a
s
ec
o
n
d
co
n
v
o
l
u
tio
n
al
lay
er
with
R
eL
U
ac
tiv
atio
n
,
p
er
f
o
r
m
in
g
a
n
o
n
-
lin
ea
r
m
ap
p
in
g
.
T
h
is
lay
er
p
la
y
s
an
ess
en
tial
r
o
le
b
e
ca
u
s
e
it
b
r
id
g
es
th
e
r
eso
lu
tio
n
g
a
p
b
y
f
ig
u
r
in
g
o
u
t
th
e
co
m
p
lex
in
ter
ac
tio
n
s
b
et
wee
n
h
ig
h
-
r
eso
lu
tio
n
an
d
lo
w
-
r
eso
lu
tio
n
p
atch
es.
Ultim
ately
,
a
th
ir
d
co
n
v
o
lu
tio
n
al
lay
er
r
ec
ei
v
es
th
e
o
u
tp
u
t
f
r
o
m
th
is
lay
er
an
d
u
s
es
th
e
p
r
o
ce
s
s
ed
f
ea
tu
r
es
to
r
ec
r
ea
te
th
e
h
ig
h
-
r
eso
lu
tio
n
i
m
ag
e.
An
e
n
o
r
m
o
u
s
ly
im
p
r
o
v
ed
h
ig
h
-
r
eso
lu
tio
n
im
ag
e
r
e
p
lace
s
th
e
in
itial
lo
w
-
r
eso
lu
tio
n
in
p
u
t.
A
m
o
d
if
ied
n
eu
r
al
n
etwo
r
k
c
alled
th
e
SR
C
NN
was
cr
ea
ted
to
en
h
an
ce
im
ag
e
r
eso
lu
tio
n
[
1
5
]
,
[
1
6
]
.
T
h
e
SR
C
NN
h
as
th
r
ee
m
ain
l
ay
er
s
,
as
T
ab
le
1
illu
s
tr
ates.
T
h
e
in
p
u
t
lay
e
r
wo
r
k
s
with
a
lo
w
-
r
eso
lu
tio
n
im
a
g
e
th
at
h
as
b
ee
n
d
o
wn
-
s
am
p
le
d
f
r
o
m
its
o
r
ig
in
al
h
ig
h
-
r
eso
l
u
tio
n
f
o
r
m
b
y
an
a
v
er
ag
e
o
f
2
,
3
,
o
r
4
.
T
h
r
ee
co
n
v
o
l
u
tio
n
al
lay
er
s
m
ak
e
u
p
th
e
SR
C
NN
m
o
d
el.
T
h
e
f
ir
s
t
l
ay
er
u
s
es
6
4
f
ilter
s
(
9
×9
)
an
d
R
eL
U
ac
tiv
atio
n
to
r
ec
o
v
er
o
v
e
r
lap
p
in
g
p
atch
es
f
r
o
m
th
e
lo
w
-
r
eso
lu
tio
n
p
ictu
r
e
an
d
c
o
n
v
er
ts
t
h
em
i
n
to
h
ig
h
-
d
im
en
s
io
n
al
f
ea
tu
r
e
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:
2
5
0
2
-
4
7
52
S
R
C
N
N
-
b
a
s
ed
ima
g
e
tr
a
n
s
mis
s
io
n
fo
r
a
u
to
n
o
mo
u
s
ve
h
icles in
limited
n
etw
o
r
k
… (
A
n
in
d
ya
A
fin
a
C
a
r
mely
a
)
905
v
ec
to
r
s
.
T
h
e
s
ec
o
n
d
lay
e
r
m
a
p
s
th
ese
v
ec
to
r
s
to
an
o
th
er
s
et
o
f
h
i
g
h
-
d
im
en
s
io
n
al
f
ea
tu
r
e
m
ap
s
u
s
in
g
3
2
f
ilter
s
(
1
×1
)
with
R
eL
U.
T
h
e
th
ir
d
la
y
er
r
ec
o
n
s
tr
u
cts
th
e
h
ig
h
-
r
eso
l
u
tio
n
im
ag
e
u
s
in
g
f
ilter
s
(
5
×
5
)
,
with
th
e
n
u
m
b
e
r
o
f
f
ilter
s
d
ep
en
d
in
g
o
n
th
e
im
ag
e
ty
p
e,
p
r
o
d
u
cin
g
a
lin
ea
r
o
u
tp
u
t.
T
h
e
m
o
d
el
is
tr
ain
ed
u
s
in
g
a
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
lo
s
s
f
u
n
ctio
n
in
o
r
d
er
to
m
i
n
im
ize
th
e
d
if
f
e
r
en
ce
b
etwe
en
th
e
r
ec
o
n
s
tr
u
ct
ed
h
ig
h
-
r
eso
lu
tio
n
im
ag
e
an
d
th
e
o
r
ig
in
al
h
ig
h
-
r
eso
lu
tio
n
im
ag
e
[
1
7
]
.
T
h
e
m
o
d
el
is
th
en
o
p
tim
ized
u
s
in
g
s
to
ch
asti
c
g
r
ad
ien
t
d
escen
t
(
SGD)
,
w
h
ich
m
ak
es
u
s
e
o
f
d
ata
au
g
m
en
tatio
n
m
et
h
o
d
s
lik
e
r
o
tatio
n
,
s
ca
lin
g
,
an
d
m
ir
r
o
r
in
g
to
b
o
o
s
t
r
o
b
u
s
tn
ess
.
T
h
e
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
lo
s
s
f
u
n
ctio
n
is
u
s
ed
b
y
SR
C
N
Ns
to
in
cr
ea
s
e
i
m
ag
e
r
eso
lu
tio
n
.
I
n
o
r
d
er
to
m
in
im
ize
th
e
d
is
cr
ep
an
cy
b
etwe
en
th
e
r
ea
l
h
ig
h
-
r
e
s
o
lu
tio
n
im
ag
es
an
d
th
e
h
ig
h
-
r
eso
lu
tio
n
im
ag
es
p
r
ed
icted
b
y
SR
C
NN,
th
is
f
u
n
ctio
n
is
ess
en
tial f
o
r
tr
ain
in
g
[
1
8
]
,
[
1
9
]
.
T
ab
le
1
.
T
h
e
n
etwo
r
k
ar
ch
itec
tu
r
e
o
f
SR
C
NN
La
y
e
r
P
u
r
p
o
se
La
y
e
r
t
y
p
e
F
i
l
t
e
r
si
z
e
N
u
mb
e
r
o
f
f
i
l
t
e
r
s
S
t
r
i
d
e
A
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
O
u
t
p
u
t
I
n
p
u
t
Lo
w
-
r
e
s
o
l
u
t
i
o
n
i
ma
g
e
-
-
-
-
-
Lo
w
-
r
e
s
o
l
u
t
i
o
n
i
ma
g
e
La
y
e
r
1
Ex
t
r
a
c
t
s
o
v
e
r
l
a
p
p
i
n
g
p
a
t
c
h
e
s
a
n
d
ma
p
s
t
o
h
i
g
h
-
d
i
m
e
n
s
i
o
n
a
l
f
e
a
t
u
r
e
v
e
c
t
o
r
s
C
o
n
v
o
l
u
t
i
o
n
a
l
l
a
y
e
r
9
×
9
64
1
R
e
LU
F
e
a
t
u
r
e
ma
p
s
r
e
p
r
e
s
e
n
t
i
n
g
e
x
t
r
a
c
t
e
d
p
a
t
c
h
e
s
La
y
e
r
2
M
a
p
s fe
a
t
u
r
e
v
e
c
t
o
r
s
t
o
a
n
o
t
h
e
r
s
e
t
o
f
h
i
g
h
-
d
i
m
e
n
s
i
o
n
a
l
f
e
a
t
u
r
e
v
e
c
t
o
r
s
C
o
n
v
o
l
u
t
i
o
n
a
l
l
a
y
e
r
1
×
1
32
1
R
e
LU
N
e
w
set
o
f
h
i
g
h
-
d
i
m
e
n
s
i
o
n
a
l
f
e
a
t
u
r
e
map
s
La
y
e
r
3
R
e
c
o
n
st
r
u
c
t
s t
h
e
h
i
g
h
-
r
e
so
l
u
t
i
o
n
i
m
a
g
e
f
r
o
m t
h
e
f
e
a
t
u
r
e
v
e
c
t
o
r
s
C
o
n
v
o
l
u
t
i
o
n
a
l
l
a
y
e
r
5
×
5
1
(
g
r
a
y
s
c
a
l
e
)
o
r
3
(
R
G
B
)
1
N
o
n
e
(
l
i
n
e
a
r
a
c
t
i
v
a
t
i
o
n
)
F
i
n
a
l
h
i
g
h
-
r
e
so
l
u
t
i
o
n
i
m
a
g
e
Ma
th
em
atica
lly
,
th
e
MSE
lo
s
s
f
u
n
ctio
n
is
d
ef
in
ed
as f
o
llo
ws
(
1
)
[
2
0
]
:
=
1
∑
(
ℎ
ℎ
(
)
−
Î
ℎ
ℎ
(
)
)
2
=
1
(
1
)
I
n
th
is
ca
s
e,
s
tan
d
s
f
o
r
th
e
to
tal
n
u
m
b
er
o
f
p
i
x
els
in
th
e
im
ag
e,
ℎ
ℎ
f
o
r
th
e
g
r
o
u
n
d
tr
u
th
h
ig
h
-
r
eso
lu
tio
n
im
ag
e,
an
d
Î
ℎ
ℎ
f
o
r
th
e
p
r
ed
icted
h
ig
h
-
r
eso
lu
tio
n
im
ag
e
g
en
e
r
ated
b
y
SR
C
NN.
T
h
e
lo
s
s
f
u
n
ctio
n
ca
lcu
lates
th
e
av
er
ag
e
s
q
u
ar
ed
d
if
f
er
en
ce
b
etwe
en
co
r
r
esp
o
n
d
in
g
p
ix
els
ac
r
o
s
s
th
e
en
tire
im
ag
e,
p
r
o
v
i
d
in
g
a
q
u
an
titativ
e
m
ea
s
u
r
e
o
f
h
o
w
well
th
e
p
r
e
d
icted
o
u
tp
u
t
m
atch
es
th
e
ac
tu
al
h
ig
h
-
r
eso
lu
tio
n
tar
g
et.
Usi
n
g
g
r
a
d
ien
t
d
escen
t
an
d
b
ac
k
p
r
o
p
a
g
atio
n
to
m
o
d
i
f
y
p
ar
am
eter
s
,
th
e
SR
C
NN
m
o
d
el
m
in
im
izes
th
e
MSE
l
o
s
s
d
u
r
in
g
tr
ain
in
g
,
lo
wer
in
g
p
r
e
d
ictio
n
er
r
o
r
an
d
en
h
an
cin
g
im
ag
e
q
u
ality
.
SR
C
N
N
ef
f
ec
tiv
ely
g
en
er
ates
h
ig
h
-
f
i
d
elity
s
u
p
er
-
r
eso
lu
tio
n
im
ag
es th
at
clo
s
ely
r
esem
b
le
th
e
h
ig
h
-
r
eso
lu
tio
n
g
r
o
u
n
d
-
tr
u
th
im
ag
es b
y
u
s
in
g
MSE
as th
e
p
r
im
ar
y
lo
s
s
f
u
n
ctio
n
.
T
h
is
s
h
o
ws
s
ig
n
if
ican
t
im
p
r
o
v
e
m
en
ts
o
v
er
t
r
ad
itio
n
al
in
ter
p
o
latio
n
m
eth
o
d
s
u
s
in
g
C
NN
s
an
d
d
ee
p
lear
n
in
g
.
T
h
is
s
u
p
er
-
r
eso
lu
tio
n
ap
p
licatio
n
will
b
e
em
p
lo
y
ed
to
ef
f
icie
n
tly
d
eliv
er
im
ag
es
in
au
to
n
o
m
o
u
s
elec
tr
ic
v
eh
icles
o
p
er
atin
g
in
c
o
n
f
in
e
d
en
v
ir
o
n
m
en
ts
.
T
h
e
r
esear
ch
p
r
o
ce
s
s
was
d
ev
elo
p
ed
in
th
r
ee
p
h
ases
,
as d
ep
icted
i
n
Fig
u
r
e
2
.
T
h
e
r
aw
d
ataset
was
ac
q
u
ir
ed
an
d
p
r
ep
ar
e
d
in
th
e
f
ir
s
t
p
h
ase.
As
s
h
o
wn
in
Fig
u
r
e
3
,
d
ata
co
llectio
n
f
o
r
th
is
s
tag
e
s
tar
ted
at
th
e
KST
Sam
au
n
Sam
ad
i
k
u
n
B
R
I
N
in
th
e
B
an
d
u
n
g
ar
ea
.
W
ith
a
4
:
3
asp
ec
t
r
atio
a
n
d
a
3
0
f
r
am
es
p
er
s
ec
o
n
d
f
r
am
e
r
ate,
th
e
I
n
s
ta3
6
0
ONE
R
S
T
win
E
d
itio
n
4
K
B
o
o
s
t
L
en
s
c
am
er
a
was
u
s
ed
to
r
ec
o
r
d
t
h
e
d
ata.
Fig
u
r
e
2
.
T
h
e
r
esear
ch
p
r
o
ce
s
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
2
,
Feb
r
u
a
r
y
20
2
5
:
9
0
3
-
9
1
2
906
Fig
u
r
e
3
.
T
h
e
lo
ca
tio
n
o
f
th
e
r
aw
d
ataset
co
llectio
n
T
h
e
s
ec
o
n
d
p
h
ase
f
o
cu
s
es
o
n
m
o
d
el
p
r
e
p
ar
atio
n
a
n
d
tr
ain
i
n
g
.
T
h
e
Ad
am
o
p
tim
is
er
was
u
s
ed
with
a
b
atch
s
ize
o
f
o
n
e
an
d
a
tr
ain
in
g
s
tep
o
f
2
0
0
,
0
0
0
.
T
h
e
d
e
v
elo
p
ed
SR
C
NN
m
o
d
el
h
ad
a
lay
e
r
co
m
p
o
s
itio
n
with
f
ilter
s
izes
o
f
9
×9
,
1
×1
(
o
r
3
×
3
; 5
×5
)
an
d
5
×
5
in
its
th
r
ee
co
n
v
o
lu
tio
n
al
lay
er
s
.
T
h
ese
lay
er
s
co
r
r
esp
o
n
d
to
th
e
p
atch
ex
tr
ac
tio
n
an
d
r
ep
r
esen
t
atio
n
lay
er
,
th
e
n
o
n
-
lin
ea
r
m
a
p
p
in
g
la
y
er
,
an
d
th
e
r
ec
o
n
s
tr
u
ctio
n
lay
er
.
Du
r
in
g
th
is
s
tag
e,
a
p
r
e
-
p
r
ep
ar
ed
tr
ain
in
g
an
d
v
alid
atio
n
d
ataset
was
u
s
ed
to
tr
ai
n
th
e
m
o
d
el.
T
h
is
p
h
ase
r
esu
lted
in
a
f
u
lly
tr
ai
n
ed
m
o
d
el,
wh
ich
wa
s
ass
ess
ed
u
s
in
g
f
o
u
r
im
p
o
r
ta
n
t
m
etr
ics
at
ea
ch
tr
ain
in
g
s
ta
g
e.
T
h
r
o
u
g
h
o
u
t
th
e
tr
ain
in
g
p
r
o
ce
d
u
r
e,
tr
ain
in
g
p
e
ak
s
ig
n
al
to
n
o
is
e
r
atio
(
PS
NR
)
,
v
alid
atio
n
PS
NR
,
tr
ain
in
g
lo
s
s
,
an
d
v
ali
d
atio
n
PS
NR
wer
e
tr
ac
k
ed
.
Af
ter
th
at,
th
e
m
o
d
el
was r
ea
d
y
f
o
r
test
in
g
in
th
e
s
u
b
s
eq
u
en
t stag
e.
Usi
n
g
th
e
p
r
e
v
io
u
s
ly
c
r
ea
ted
test
d
ataset,
th
e
tr
ain
ed
m
o
d
el's
p
er
f
o
r
m
a
n
ce
is
ass
ess
ed
in
th
e
last
p
h
ase.
B
ased
o
n
th
e
im
ag
es
in
th
e
test
in
g
d
ataset,
th
e
tes
tin
g
m
eth
o
d
p
r
o
d
u
ce
s
th
e
m
ea
n
PS
NR
v
alu
e,
a
q
u
alitativ
e
in
d
icato
r
o
f
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
.
T
o
as
s
ess
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
c
e
q
u
an
titativ
ely
,
a
d
em
o
n
s
tr
atio
n
is
ca
r
r
ied
o
u
t
to
h
ig
h
lig
h
t
th
e
ca
lib
er
o
f
th
e
s
u
p
er
-
r
eso
lu
tio
n
p
h
o
to
g
r
ap
h
s
p
r
o
d
u
ce
d
b
y
th
e
t
r
a
i
n
e
d
m
o
d
e
l
.
T
h
is
p
h
a
s
e
m
a
k
e
s
s
u
r
e
t
h
a
t
t
h
e
m
o
d
e
l'
s
e
f
f
ic
a
cy
i
s
t
h
o
r
o
u
g
h
l
y
a
s
s
e
s
s
e
d
b
e
f
o
r
e
i
t
is
p
u
t
i
n
t
o
u
s
e
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
will
d
is
cu
s
s
co
m
p
r
eh
en
s
iv
ely
th
e
r
esu
lts
g
iv
en
f
r
o
m
th
e
ex
p
er
im
e
n
t
b
ased
o
n
th
e
th
r
ee
p
h
ase
d
escr
ib
ed
in
m
eth
o
d
s
ec
tio
n
.
3
.
1
.
F
irst
ph
a
s
e
re
s
ults
T
h
e
d
ata
co
llectio
n
p
r
o
ce
s
s
in
v
o
lv
ed
ca
p
tu
r
in
g
v
id
eo
s
at
f
if
teen
d
is
tin
ct
lo
ca
tio
n
s
with
in
th
e
d
esig
n
ated
ar
ea
,
r
esu
ltin
g
in
f
if
teen
s
ep
ar
ate
v
id
eo
s
.
E
ac
h
v
id
eo
was
th
e
n
c
o
n
v
er
te
d
in
t
o
a
s
er
ies
o
f
d
is
cr
ete
f
r
am
es
u
s
in
g
Py
th
o
n
,
g
en
e
r
atin
g
a
s
u
b
s
tan
tial
n
u
m
b
er
o
f
f
r
am
es.
Fra
m
es
co
n
tain
in
g
co
m
p
ar
ab
le
d
ata
wer
e
elim
in
ated
to
g
u
ar
a
n
tee
a
d
i
v
er
s
if
ied
d
ataset.
Nex
t,
th
r
ee
s
u
b
s
ets
o
f
t
h
e
r
e
m
ain
in
g
f
r
am
es
wer
e
cr
ea
ted
:
tr
ain
in
g
,
v
alid
atio
n
,
an
d
test
in
g
s
ets
[
2
1
]
-
[
2
6
]
.
Fro
m
t
h
e
s
el
ec
ted
f
r
am
es,
3
2
2
im
ag
es
wer
e
ca
teg
o
r
ized
in
to
eig
h
t
s
ize
ca
teg
o
r
ies,
with
d
im
en
s
io
n
s
o
f
1
5
0
×
1
5
0
,
2
0
0
×2
0
0
,
2
5
0
×2
5
0
,
3
0
0
×
3
0
0
,
3
5
0
×3
5
0
,
4
0
0
×
4
0
0
,
4
5
0
×4
5
0
,
a
n
d
5
0
0
×5
0
0
p
i
x
els,
f
o
r
m
in
g
t
h
e
tr
ain
in
g
d
ataset
.
T
h
e
v
alid
atio
n
d
ataset
co
m
p
r
is
ed
1
4
1
im
ag
es,
ch
o
s
en
to
co
v
er
a
b
r
o
ad
e
r
r
a
n
g
e
o
f
s
izes
th
an
th
e
tr
ain
in
g
d
ataset
.
T
h
e
test
in
g
d
ataset
in
cl
u
d
ed
im
a
g
es
s
ca
led
in
th
r
ee
ca
teg
o
r
ies:
×2
,
×3
,
an
d
×4
.
E
ac
h
s
ca
le
h
a
d
a
c
o
r
r
esp
o
n
d
in
g
ca
te
g
o
r
y
f
o
r
d
a
ta
an
d
lab
els,
ea
ch
co
n
tain
in
g
5
0
im
ag
es
o
f
th
e
s
am
e
d
ataset
b
u
t
in
d
i
f
f
er
en
t
s
izes.
T
h
e
lab
els
f
ea
tu
r
ed
g
r
o
u
n
d
tr
u
th
im
ag
es
with
d
im
en
s
io
n
s
o
f
5
1
2
×5
1
2
p
ix
el
s
f
o
r
s
ca
les
×2
an
d
×4
an
d
5
1
0
×5
1
0
p
ix
els
f
o
r
s
ca
le
×3
.
T
h
e
d
ata
ca
teg
o
r
y
in
clu
d
ed
im
ag
es
r
esized
t
o
h
alf
,
a
th
i
r
d
,
o
r
a
q
u
ar
ter
o
f
th
e
g
r
o
u
n
d
tr
u
t
h
s
ize,
d
e
p
en
d
i
n
g
o
n
th
e
s
ca
le.
Fo
r
test
in
g
in
p
u
t,
th
e
im
ag
e
r
eso
lu
tio
n
s
wer
e
2
5
6
×2
5
6
p
i
x
els
f
o
r
s
ca
le
×2
,
1
7
0
×1
7
0
p
ix
els
f
o
r
s
ca
le
×3
,
an
d
1
2
8
×1
2
8
p
ix
els
f
o
r
s
ca
le
×4
.
Up
o
n
co
m
p
letio
n
o
f
th
is
p
h
ase,
th
e
tr
ain
in
g
,
v
alid
atio
n
,
an
d
test
in
g
d
atasets
wer
e
p
r
ep
ar
e
d
f
o
r
s
u
b
s
eq
u
e
n
t
p
h
ases
o
f
th
e
r
esear
c
h
.
3
.
2
.
Seco
nd
ph
a
s
e
re
s
ults
Fig
u
r
e
4
d
escr
ib
es
th
e
r
esu
lts
o
f
tr
ain
in
g
an
d
v
alid
atio
n
f
o
r
ar
ch
itectu
r
e
9
1
5
f
o
r
b
o
t
h
lo
s
s
an
d
PS
NR
.
T
h
e
g
r
ap
h
in
Fig
u
r
e
4
(
a)
s
h
o
ws
th
e
tr
ain
in
g
an
d
v
alid
atio
n
lo
s
s
f
o
r
a
m
o
d
el
with
th
e
ar
c
h
itectu
r
e
m
o
d
el
9
1
5
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:
2
5
0
2
-
4
7
52
S
R
C
N
N
-
b
a
s
ed
ima
g
e
tr
a
n
s
mis
s
io
n
fo
r
a
u
to
n
o
mo
u
s
ve
h
icles in
limited
n
etw
o
r
k
… (
A
n
in
d
ya
A
fin
a
C
a
r
mely
a
)
907
o
v
er
a
s
er
ies
o
f
tr
ain
in
g
s
te
p
s
r
an
g
in
g
f
r
o
m
0
to
2
0
0
,
0
0
0
.
B
o
th
th
e
tr
ain
in
g
an
d
v
al
id
atio
n
lo
s
s
es
s
tar
t
r
elativ
ely
h
ig
h
,
a
r
o
u
n
d
0
.
0
2
5
,
b
u
t
s
h
o
w
a
s
teep
d
ec
lin
e
in
t
h
e
ea
r
ly
s
tep
s
,
d
ec
r
ea
s
in
g
s
ig
n
i
f
ican
tly
b
y
a
r
o
u
n
d
5
0
,
0
0
0
s
tep
s
.
Af
ter
th
is
r
ap
id
d
ec
r
ea
s
e,
th
e
lo
s
s
es
s
tab
ili
s
e
an
d
r
em
ain
lo
w
a
n
d
clo
s
e
to
ea
ch
o
th
er
f
o
r
th
e
r
em
ain
in
g
tr
ai
n
in
g
s
tep
s
.
T
h
e
co
n
v
e
r
g
en
ce
o
f
b
o
th
t
h
e
tr
ai
n
in
g
an
d
v
alid
atio
n
lo
s
s
es
to
s
im
ilar
lo
w
v
alu
es
s
u
g
g
ests
th
at
th
e
m
o
d
el
is
p
er
f
o
r
m
in
g
ef
f
ec
tiv
ely
,
with
a
s
tr
o
n
g
g
en
er
alis
atio
n
ca
p
ab
ilit
y
,
as
in
d
icate
d
b
y
th
e
v
alid
atio
n
lo
s
s
r
em
ain
in
g
clo
s
e
to
th
e
tr
ain
i
n
g
lo
s
s
an
d
s
h
o
win
g
n
o
s
ig
n
s
o
f
o
v
er
f
itti
n
g
.
T
h
e
s
tr
o
n
g
est
asp
ec
t
o
f
th
is
m
o
d
el
is
its
ab
ilit
y
to
l
ea
r
n
q
u
ick
ly
an
d
m
ain
tain
lo
w
lo
s
s
v
alu
es,
d
em
o
n
s
tr
atin
g
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
.
Ho
wev
er
,
th
e
wea
k
n
ess
lies
in
th
e
h
i
g
h
i
n
itial
lo
s
s
,
s
u
g
g
esti
n
g
th
at
f
u
r
th
e
r
o
p
tim
is
atio
n
o
r
b
etter
in
itialis
atio
n
s
tr
ateg
ies
co
u
ld
p
o
ten
tially
im
p
r
o
v
e
t
h
e
ea
r
ly
p
er
f
o
r
m
an
ce
o
f
th
e
m
o
d
el.
Ov
er
all,
th
e
m
o
d
el
s
h
o
ws
ex
ce
llen
t
p
er
f
o
r
m
an
ce
an
d
g
e
n
er
alis
atio
n
with
th
e
g
iv
e
n
ar
ch
itectu
r
e.
T
h
e
g
r
ap
h
in
Fig
u
r
e
4
(
b
)
s
h
o
ws
th
e
P
SNR
f
o
r
b
o
th
th
e
tr
a
in
in
g
an
d
v
alid
atio
n
d
ata
s
ets
u
s
in
g
th
e
ar
ch
itectu
r
e
m
o
d
el
9
1
5
o
v
er
a
s
er
ies
o
f
tr
ain
in
g
s
tep
s
r
an
g
in
g
f
r
o
m
0
t
o
2
0
0
,
0
0
0
.
I
n
itially
,
b
o
th
th
e
tr
ain
in
g
an
d
v
alid
atio
n
PS
NR
v
alu
es
s
tar
t
r
elativ
ely
lo
w,
ar
o
u
n
d
1
0
to
2
0
d
B
.
As
tr
ain
in
g
p
r
o
g
r
ess
es,
th
er
e
is
a
s
h
ar
p
in
cr
ea
s
e
in
th
e
PS
NR
v
alu
es
with
in
th
e
f
ir
s
t
5
0
,
0
0
0
s
tep
s
,
with
th
e
tr
ain
in
g
PS
NR
r
is
in
g
r
ap
id
ly
to
a
r
o
u
n
d
3
5
an
d
th
e
v
alid
atio
n
PS
NR
r
ea
ch
in
g
ar
o
u
n
d
3
0
.
Af
ter
t
h
is
in
itial
s
u
r
g
e,
b
o
t
h
th
e
tr
ai
n
in
g
a
n
d
v
alid
atio
n
PS
NR
v
alu
es
s
tab
ilis
e
an
d
r
em
ain
at
th
ese
lev
els
f
o
r
th
e
r
em
ain
d
er
o
f
th
e
tr
ain
in
g
s
tep
s
.
T
h
e
m
o
d
el
s
h
o
ws
a
s
tr
o
n
g
in
itial
lear
n
in
g
p
h
ase,
as
ev
id
en
ce
d
b
y
th
e
r
ap
id
in
cr
ea
s
e
i
n
PS
NR
,
in
d
icatin
g
t
h
at
it
q
u
ick
ly
im
p
r
o
v
es
t
h
e
q
u
ality
o
f
th
e
r
ec
o
n
s
tr
u
cted
i
m
ag
es.
I
n
a
d
d
itio
n
,
th
e
s
tab
ilis
atio
n
o
f
b
o
t
h
PS
NR
v
alu
es
s
u
g
g
ests
co
n
s
is
ten
t
p
er
f
o
r
m
an
ce
o
v
er
ex
ten
d
ed
tr
ain
in
g
.
Ho
wev
er
,
t
h
e
m
o
d
el
s
t
ar
ts
with
r
elativ
ely
lo
w
PS
N
R
v
alu
es,
s
u
g
g
esti
n
g
p
o
o
r
in
itial
p
er
f
o
r
m
an
ce
th
at
co
u
ld
b
e
im
p
r
o
v
ed
with
b
etter
in
itialis
atio
n
o
r
p
r
e
-
tr
ain
in
g
s
tr
ateg
ies.
Ad
d
itio
n
ally
,
th
er
e
ap
p
ea
r
s
to
b
e
a
m
in
o
r
o
v
er
f
itti
n
g
,
with
th
e
m
o
d
el
d
o
in
g
b
etter
o
n
th
e
tr
ain
in
g
d
ata
th
an
th
e
v
alid
atio
n
d
ata,
as
ev
id
e
n
ce
d
b
y
t
h
e
s
u
b
s
tan
tial
d
if
f
er
en
ce
b
etwe
en
th
e
tr
ain
in
g
PS
NR
(
ab
o
u
t
3
5
)
an
d
th
e
v
alid
atio
n
PS
NR
(
ar
o
u
n
d
3
0
)
.
Ov
er
all,
th
e
m
o
d
el
s
h
o
ws
s
tr
o
n
g
p
er
f
o
r
m
an
ce
with
r
a
p
id
i
n
itial
im
p
r
o
v
e
m
en
ts
an
d
s
tab
le
q
u
ality
m
ain
ten
a
n
c
e
o
v
er
tim
e,
alth
o
u
g
h
th
e
r
e
is
r
o
o
m
f
o
r
im
p
r
o
v
em
e
n
t
in
in
itial
p
er
f
o
r
m
a
n
ce
an
d
g
en
er
alis
atio
n
to
u
n
s
ee
n
d
ata.
(
a)
(
b
)
Fig
u
r
e
4
.
T
h
e
r
esu
lt o
f
tr
ain
in
g
an
d
v
alid
atio
n
f
o
r
ar
c
h
itectu
r
e
9
1
5
(
a)
lo
s
s
(
b
)
PS
NR
Fig
u
r
e
5
d
escr
ib
es
th
e
r
esu
lts
o
f
tr
ain
in
g
an
d
v
alid
atio
n
f
o
r
ar
ch
itectu
r
e
9
3
5
f
o
r
b
o
t
h
lo
s
s
an
d
PS
NR
.
T
h
e
g
r
ap
h
in
Fig
u
r
e
5
(
a)
s
h
o
ws
th
e
tr
ain
in
g
an
d
v
alid
atio
n
lo
s
s
f
o
r
a
m
o
d
el
with
th
e
ar
c
h
itectu
r
e
m
o
d
el
9
3
5
o
v
er
a
s
er
ies
o
f
tr
ain
in
g
s
tep
s
r
an
g
in
g
f
r
o
m
0
to
2
0
0
,
0
0
0
.
I
n
i
tially
,
b
o
th
th
e
t
r
ain
in
g
an
d
v
a
lid
atio
n
lo
s
s
es
s
tar
t
h
ig
h
,
ar
o
u
n
d
0
.
0
9
.
T
h
e
r
e
is
a
s
h
ar
p
d
r
o
p
in
lo
s
s
with
in
th
e
f
ir
s
t
5
0
,
0
0
0
s
tep
s
,
wh
er
e
b
o
th
tr
ain
in
g
an
d
v
alid
atio
n
lo
s
s
es
d
r
o
p
t
o
ar
o
u
n
d
0
.
0
1
.
Af
te
r
th
is
s
h
ar
p
d
e
clin
e,
th
e
lo
s
s
es
s
tab
ilize
an
d
r
em
ain
co
n
s
is
ten
tly
lo
w
f
o
r
th
e
r
em
ain
d
er
o
f
th
e
tr
ain
in
g
s
tep
s
.
T
h
e
m
o
d
el
s
h
o
ws
s
tr
en
g
th
s
i
n
r
a
p
id
co
n
v
er
g
en
ce
,
ef
f
ec
tiv
ely
r
ed
u
cin
g
lo
s
s
es
with
in
th
e
f
ir
s
t
5
0
,
0
0
0
s
tep
s
,
an
d
m
ain
tain
in
g
lo
w
a
n
d
s
tab
le
lo
s
s
v
al
u
es
th
r
o
u
g
h
o
u
t
th
e
tr
ain
in
g
p
e
r
io
d
,
in
d
icatin
g
s
tr
o
n
g
p
e
r
f
o
r
m
an
ce
a
n
d
n
o
d
e
g
r
a
d
atio
n
o
v
er
tim
e.
Ho
wev
er
,
th
e
m
o
d
el
s
tar
ts
with
r
elativ
ely
h
ig
h
lo
s
s
v
alu
es,
s
u
g
g
esti
n
g
p
o
o
r
in
itial
p
er
f
o
r
m
an
ce
th
at
co
u
ld
b
e
im
p
r
o
v
e
d
with
b
etter
in
itializatio
n
tech
n
iq
u
es.
I
n
a
d
d
itio
n
,
th
e
alm
o
s
t
id
en
tical
n
atu
r
e
o
f
t
h
e
tr
ain
i
n
g
a
n
d
v
al
id
atio
n
lo
s
s
es
co
u
ld
in
d
icate
th
at
th
e
v
alid
atio
n
d
ata
is
n
o
t
s
u
f
f
icien
tly
ch
allen
g
in
g
o
r
d
i
v
er
s
e
co
m
p
ar
e
d
to
th
e
tr
ain
in
g
d
ata,
p
o
ten
tially
m
ask
in
g
o
v
er
f
itti
n
g
.
Ov
e
r
all,
th
e
m
o
d
el
s
h
o
ws
r
o
b
u
s
t
p
er
f
o
r
m
a
n
ce
b
u
t
co
u
ld
b
en
ef
it
f
r
o
m
im
p
r
o
v
em
e
n
ts
in
in
itial p
er
f
o
r
m
an
ce
an
d
v
alid
atio
n
d
ata
d
iv
er
s
ity
.
T
h
e
PS
NR
f
o
r
t
h
e
tr
ai
n
in
g
a
n
d
v
alid
atio
n
s
ets
o
f
th
e
9
3
5
-
m
o
d
el
ar
ch
itectu
r
e
is
p
r
esen
ted
o
n
th
e
g
r
ap
h
in
Fig
u
r
e
5
(
b
)
.
T
h
e
s
tep
s
,
wh
ich
s
p
a
n
f
r
o
m
0
to
2
0
0
,
0
0
0
,
a
r
e
r
ep
r
esen
ted
b
y
th
e
x
-
a
x
is
,
wh
ile
th
e
PS
NR
v
alu
es
ar
e
d
is
p
lay
ed
o
n
th
e
y
-
ax
is
.
T
h
e
tr
ain
in
g
PS
NR
i
s
s
h
o
wn
b
y
th
e
s
o
lid
b
lu
e
lin
e,
wh
ile
th
e
v
alid
atio
n
PS
NR
is
s
h
o
wn
b
y
th
e
d
ash
e
d
o
r
an
g
e
lin
e
.
T
h
e
m
o
d
el
s
h
o
ws
a
r
ap
id
im
p
r
o
v
em
e
n
t
in
PS
NR
w
ith
in
th
e
f
ir
s
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
2
,
Feb
r
u
a
r
y
20
2
5
:
9
0
3
-
9
1
2
908
5
0
,
0
0
0
s
tep
s
,
in
d
icatin
g
ef
f
e
ctiv
e
in
itial
lear
n
in
g
.
T
h
e
tr
a
in
in
g
PS
NR
s
tab
il
izes
ar
o
u
n
d
3
5
d
B
,
wh
ich
is
r
elativ
ely
h
ig
h
an
d
in
d
icate
s
t
h
at
th
e
m
o
d
el
f
its
th
e
tr
ain
in
g
d
ata
well.
T
h
e
s
ig
n
i
f
ican
t
d
if
f
er
en
ce
b
etwe
en
th
e
PS
NR
s
f
o
r
tr
ain
in
g
a
n
d
v
alid
atio
n
p
o
in
ts
to
t
h
e
p
o
s
s
ib
ilit
y
o
f
o
v
e
r
f
itti
n
g
,
i
n
wh
ich
t
h
e
m
o
d
el
p
er
f
o
r
m
s
well
o
n
tr
ain
in
g
d
ata
b
u
t p
o
o
r
ly
o
n
v
alid
atio
n
d
ata.
Ov
er
all,
wh
ile
th
e
m
o
d
el
lear
n
s
th
e
tr
ain
in
g
d
ata
ef
f
ec
tiv
ely
,
its
g
en
er
aliza
tio
n
ab
ilit
y
n
ee
d
s
i
m
p
r
o
v
e
m
en
t,
p
o
ten
tially
r
eq
u
i
r
in
g
tech
n
iq
u
es
s
u
c
h
as
r
eg
u
la
r
izatio
n
,
d
r
o
p
o
u
t
o
r
m
o
r
e
d
iv
e
r
s
e
tr
ain
in
g
d
ata
to
i
m
p
r
o
v
e
p
er
f
o
r
m
a
n
ce
o
n
th
e
v
a
lid
atio
n
s
et.
(
a)
(
b
)
Fig
u
r
e
5
.
T
h
e
r
esu
lt o
f
tr
ain
in
g
an
d
v
alid
atio
n
f
o
r
ar
c
h
itectu
r
e
9
3
5
(
a)
lo
s
s
(
b
)
PS
NR
Fig
u
r
e
6
d
escr
ib
es
th
e
r
esu
lts
o
f
tr
ain
in
g
an
d
v
alid
atio
n
f
o
r
ar
ch
itectu
r
e
9
5
5
f
o
r
b
o
t
h
lo
s
s
an
d
PS
NR
.
T
h
e
g
r
a
p
h
i
n
Fig
u
r
e
6
(
a)
s
h
o
ws
th
e
lo
s
s
o
v
er
tr
ain
i
n
g
s
tep
s
f
o
r
b
o
th
th
e
tr
ai
n
in
g
a
n
d
v
alid
atio
n
s
ets
o
f
m
o
d
el
ar
ch
itectu
r
e
9
5
5
,
with
th
e
x
-
a
x
is
r
ep
r
esen
tin
g
s
tep
s
f
r
o
m
0
to
2
0
0
,
0
0
0
an
d
th
e
y
-
ax
is
s
h
o
win
g
th
e
lo
s
s
v
alu
es.
B
o
th
th
e
v
alid
atio
n
lo
s
s
(
d
ash
ed
o
r
an
g
e
lin
e
)
an
d
th
e
tr
ain
i
n
g
lo
s
s
(
s
o
lid
b
lu
e
lin
e)
b
eg
in
at
r
o
u
g
h
ly
0
.
1
an
d
d
r
o
p
o
f
f
q
u
ick
ly
t
h
r
o
u
g
h
o
u
t
t
h
e
f
ir
s
t
1
0
,
0
0
0
s
tep
s
b
ef
o
r
e
s
tab
ilizin
g
at
clo
s
e
to
ze
r
o
f
o
r
t
h
e
r
em
ain
in
g
s
tep
s
.
T
h
e
m
o
d
el
d
em
o
n
s
tr
ates
ef
f
icien
t
lear
n
in
g
with
r
ap
i
d
lo
s
s
r
e
d
u
ctio
n
a
n
d
ac
h
iev
es
a
lo
w
f
i
n
al
lo
s
s
,
s
u
g
g
esti
n
g
a
g
o
o
d
f
it
to
th
e
d
ata
with
o
u
t
s
ig
n
if
ican
t
o
v
er
f
itti
n
g
.
Ho
wev
er
,
th
e
n
ea
r
-
ze
r
o
lo
s
s
r
aises
co
n
ce
r
n
s
ab
o
u
t
p
o
ten
tial
o
v
e
r
f
itti
n
g
,
as
th
e
m
o
d
el
m
ay
b
e
h
ig
h
ly
tu
n
e
d
to
th
e
tr
ain
in
g
d
ata.
Fu
r
th
er
m
o
r
e,
th
e
f
ac
t
t
h
at
th
e
v
alid
atio
n
lo
s
s
clo
s
ely
f
o
llo
ws
th
e
tr
ain
in
g
lo
s
s
w
ith
o
u
t
m
u
ch
v
ar
iatio
n
s
u
g
g
ests
a
lack
o
f
d
iv
er
s
ity
in
th
e
v
alid
atio
n
s
et
o
r
p
o
ten
tial
p
r
o
b
lem
s
with
th
e
v
alid
atio
n
p
r
o
ce
s
s
.
T
h
er
ef
o
r
e,
wh
ile
th
e
m
o
d
el
s
h
o
ws
s
tr
o
n
g
in
itial p
er
f
o
r
m
a
n
ce
,
f
u
r
th
er
v
a
lid
atio
n
is
r
eq
u
ir
ed
to
en
s
u
r
e
r
o
b
u
s
t g
en
er
aliza
tio
n
.
T
h
e
g
r
ap
h
in
Fig
u
r
e
6
(
b
)
s
h
o
ws
th
e
P
SNR
o
v
er
tr
ain
in
g
s
tep
s
f
o
r
b
o
th
th
e
tr
ain
in
g
an
d
v
alid
atio
n
s
ets
o
f
m
o
d
el
ar
ch
itectu
r
e
9
5
5
,
with
th
e
x
-
a
x
is
r
ep
r
esen
tin
g
s
tep
s
f
r
o
m
0
t
o
2
0
0
,
0
0
0
an
d
th
e
y
-
ax
is
s
h
o
win
g
PS
NR
v
alu
es.
T
h
e
s
o
lid
b
lu
e
lin
e
(
tr
ain
in
g
PS
NR
)
an
d
th
e
d
ash
ed
o
r
a
n
g
e
lin
e
(
v
ali
d
atio
n
PS
NR
)
b
o
th
in
cr
ea
s
e
r
ap
id
ly
with
i
n
th
e
f
i
r
s
t
5
0
,
0
0
0
s
tep
s
.
T
h
e
tr
ain
i
n
g
P
SNR
s
tab
il
izes
at
ar
o
u
n
d
3
5
,
wh
ile
th
e
v
alid
atio
n
PS
NR
lev
els
o
f
f
at
ar
o
u
n
d
3
0
.
T
h
e
m
o
d
el
s
h
o
ws
ef
f
icien
t
in
itial
lear
n
in
g
a
n
d
ac
h
iev
es
h
ig
h
PS
NR
v
alu
es
f
o
r
b
o
th
tr
ain
in
g
an
d
v
alid
atio
n
s
ets,
in
d
icatin
g
g
o
o
d
p
er
f
o
r
m
a
n
ce
.
T
h
e
m
o
d
el
d
o
es
b
etter
o
n
th
e
tr
ain
in
g
d
ata
th
an
th
e
v
alid
atio
n
d
ata,
th
o
u
g
h
,
an
d
a
n
o
ticea
b
le
d
if
f
er
en
c
e
in
th
e
PS
NR
s
b
etwe
en
th
e
tr
ain
in
g
an
d
v
alid
atio
n
s
ets
th
e
p
o
s
s
ib
ilit
y
o
f
o
v
er
f
itti
n
g
.
Fu
r
th
e
r
m
o
r
e
,
th
e
v
alid
atio
n
PS
NR
p
latea
u
s
an
d
f
alls
s
h
o
r
t
o
f
th
e
tr
ai
n
in
g
PS
NR
,
s
u
g
g
esti
n
g
th
at
th
e
m
o
d
el'
s
ab
ilit
y
to
g
en
er
alize
u
p
o
n
n
ew
d
ata
n
ee
d
s
im
p
r
o
v
em
en
t.
(
a)
(
b
)
Fig
u
r
e
6
.
T
h
e
r
esu
lt o
f
tr
ain
in
g
an
d
v
alid
atio
n
f
o
r
ar
c
h
itectu
r
e
9
3
5
(
a)
lo
s
s
(
b
)
PS
NR
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:
2
5
0
2
-
4
7
52
S
R
C
N
N
-
b
a
s
ed
ima
g
e
tr
a
n
s
mis
s
io
n
fo
r
a
u
to
n
o
mo
u
s
ve
h
icles in
limited
n
etw
o
r
k
… (
A
n
in
d
ya
A
fin
a
C
a
r
mely
a
)
909
3
.
3
.
T
hird
ph
a
s
e
re
s
ult
s
T
ab
le
2
s
h
o
ws
a
co
m
p
ar
ativ
e
an
aly
s
is
o
f
th
r
ee
SR
C
NN
m
o
d
els,
9
1
5
,
9
3
5
,
a
n
d
9
5
5
,
ac
r
o
s
s
th
r
ee
s
ca
le
m
etr
ics:
×
2
,
×
3
an
d
×
4
.
E
ac
h
m
etr
ic
co
n
tain
s
th
e
m
ax
i
m
u
m
,
m
in
im
u
m
an
d
av
er
a
g
e
v
alu
es
o
f
PS
NR
f
o
r
th
ese
m
o
d
els.
T
a
b
le
2
d
em
o
n
s
tr
ate
SR
C
NN
m
o
d
el
9
1
5
n
o
t
ab
le
s
tr
en
g
th
s
,
p
ar
ticu
lar
ly
wi
th
a
h
ig
h
m
ax
im
u
m
v
alu
e
o
f
3
5
.
7
0
5
f
o
r
th
e
×
2
s
ca
le,
in
d
icatin
g
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
in
th
is
s
p
ec
if
ic
asp
ec
t.
T
h
e
m
o
d
el
also
m
ain
tain
s
r
elativ
ely
h
ig
h
a
v
er
ag
e
v
alu
es
ac
r
o
s
s
all
s
ca
les
(
×
2
:
3
3
.
5
4
9
,
×
3
:
3
0
.
9
1
3
,
×
4
:
2
9
.
5
9
4
)
,
d
em
o
n
s
tr
atin
g
co
n
s
is
ten
t
p
er
f
o
r
m
an
ce
.
Ho
we
v
er
,
t
h
e
m
o
d
el
s
h
o
ws
wea
k
n
e
s
s
es
with
lo
wer
m
in
im
u
m
v
alu
es
f
o
r
×
3
(
2
7
.
7
2
1
)
an
d
×
4
(
2
6
.
3
7
9
)
,
in
d
icatin
g
v
a
r
iab
ilit
y
an
d
in
co
n
s
is
ten
cy
in
p
er
f
o
r
m
an
ce
f
o
r
th
ese
s
ca
les.
T
h
e
SR
C
NN
m
o
d
el
9
3
5
a
c
o
m
m
en
d
ab
le
m
ax
im
u
m
v
alu
e
o
f
3
5
.
5
5
0
f
o
r
t
h
e
×
2
s
ca
les,
s
lig
h
tly
lo
wer
th
an
m
o
d
el
9
1
5
b
u
t
s
till
s
tr
o
n
g
.
I
t
r
em
ai
n
s
co
n
s
is
ten
t
with
r
esp
ec
tab
le
av
er
ag
es
(
×
2
:
3
3
.
4
8
3
,
×
3
:
3
0
.
9
2
6
,
×
4
:
2
9
.
5
8
9
)
.
Ne
v
er
th
eless
,
th
e
m
o
d
el'
s
m
in
im
u
m
v
alu
es
f
o
r
×
2
(
2
9
.
9
4
1
)
a
n
d
×
4
(
2
6
.
3
6
9
)
ar
e
th
e
lo
west
o
f
th
e
th
r
e
e
s
ca
les,
in
d
icatin
g
p
o
ten
tial
f
o
r
im
p
r
o
v
em
e
n
t
in
m
ain
tain
in
g
a
h
ig
h
e
r
b
aselin
e
p
er
f
o
r
m
an
ce
.
T
h
e
SR
C
NN
m
o
d
el
9
5
5
s
tan
d
s
o
u
t
with
th
e
h
ig
h
est
m
ax
im
u
m
v
alu
e
f
o
r
th
e
×
2
m
etr
ic
(
3
5
.
7
6
1
)
,
th
e
b
est
o
f
all
th
r
ee
m
o
d
els.
I
t
also
h
as
th
e
h
ig
h
est
m
in
im
u
m
v
alu
e
f
o
r
t
h
e
×
2
s
ca
les
(
3
0
.
0
9
2
)
,
d
em
o
n
s
tr
atin
g
s
tr
o
n
g
b
asic
p
er
f
o
r
m
an
ce
.
T
h
e
av
er
ag
e
v
alu
es
f
o
r
×
2
(
3
3
.
6
5
5
)
a
n
d
×
4
(
2
9
.
5
9
6
)
ar
e
th
e
h
ig
h
est
o
f
th
e
th
r
ee
m
o
d
els,
in
d
ica
tin
g
o
v
e
r
all
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
.
Ho
wev
er
,
th
e
×
3
m
etr
ic
s
h
o
ws
a
s
lig
h
tly
lo
w
er
m
ax
im
u
m
v
alu
e
(
3
2
.
8
9
5
)
c
o
m
p
ar
ed
to
th
e
o
th
e
r
m
o
d
els,
wh
ich
co
u
ld
b
e
an
ar
e
a
f
o
r
im
p
r
o
v
e
m
en
t.
T
ab
le
2
.
T
h
e
PS
NR
o
f
test
in
s
ca
le
×2
,
×3
,
×
4
f
o
r
all
ar
ch
itec
tu
r
e
m
o
d
el
Th
e
P
S
N
R
o
f
t
e
st
i
n
s
c
a
l
e
×
2
,
×
3
,
×
4
f
o
r
a
r
c
h
i
t
e
c
t
u
r
e
mo
d
e
l
9
1
5
×2
×3
×4
M
a
x
3
5
.
7
0
5
3
2
.
9
1
8
3
1
.
6
0
9
M
i
n
2
9
.
9
5
9
2
7
.
7
2
1
2
6
.
3
7
9
A
v
e
r
a
g
e
3
3
.
5
4
9
3
0
.
9
1
3
2
9
.
5
9
4
Th
e
P
S
N
R
o
f
t
e
st
i
n
s
c
a
l
e
×
2
,
×
3
,
×
4
f
o
r
a
r
c
h
i
t
e
c
t
u
r
e
mo
d
e
l
9
3
5
×2
×3
×4
M
a
x
3
5
.
5
5
0
3
2
.
9
1
6
3
1
.
5
8
4
M
i
n
2
9
.
9
4
1
2
7
.
7
4
1
2
6
.
3
6
9
A
v
e
r
a
g
e
3
3
.
4
8
3
3
0
.
9
2
6
2
9
.
5
8
9
Th
e
P
S
N
R
o
f
t
e
st
i
n
s
c
a
l
e
×
2
,
×
3
,
×
4
f
o
r
a
r
c
h
i
t
e
c
t
u
r
e
mo
d
e
l
9
5
5
×2
×3
×4
M
a
x
3
5
.
7
6
1
3
2
.
8
9
5
3
1
.
5
7
5
M
i
n
3
0
.
0
9
2
2
7
.
7
2
3
2
6
.
3
7
4
A
v
e
r
a
g
e
3
3
.
6
5
5
3
0
.
9
1
6
2
9
.
5
9
6
T
h
e
m
o
d
el
9
5
5
o
u
tp
e
r
f
o
r
m
s
th
e
o
th
er
two
SR
C
NN
m
o
d
els
in
ter
m
s
o
f
p
e
r
f
o
r
m
an
ce
co
n
s
is
ten
cy
o
n
all
s
ca
les.
Mo
d
el
9
3
5
h
as
th
e
lo
west
m
in
im
u
m
v
alu
es
b
u
t
h
as
a
co
n
s
is
ten
t
av
er
ag
e
p
er
f
o
r
m
an
ce
,
wh
er
ea
s
m
o
d
el
9
1
5
co
m
es
in
s
ec
o
n
d
with
g
r
ea
t
r
esu
lts
b
u
t
s
ig
n
if
ica
n
t
v
ar
iab
ilit
y
.
T
o
im
p
r
o
v
e
th
eir
r
eliab
ilit
y
,
m
o
d
els
9
1
5
a
n
d
9
3
5
co
u
ld
f
o
cu
s
o
n
im
p
r
o
v
i
n
g
t
h
eir
m
i
n
im
u
m
p
er
f
o
r
m
an
ce
v
alu
es.
O
v
er
all,
m
o
d
el
9
5
5
em
er
g
es
as
th
e
m
o
s
t r
eliab
le
an
d
ef
f
ec
tiv
e
o
f
th
e
th
r
ee
,
s
h
o
win
g
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
a
n
d
co
n
s
is
ten
cy
.
E
x
am
p
les
o
f
in
p
u
t
an
d
o
u
tp
u
t
im
ag
es
u
tili
ze
d
d
u
r
in
g
th
e
test
in
g
p
r
o
ce
s
s
ar
e
s
h
o
wn
in
Fig
u
r
e
7
.
T
h
e
in
p
u
t
im
a
g
e
f
o
r
ev
er
y
m
o
d
el
in
p
u
t
th
at
h
as
th
e
h
ig
h
est
P
SNR
v
alu
e
is
d
is
p
lay
ed
in
F
ig
u
r
e
7
(
a)
.
I
n
th
e
m
ea
n
tim
e,
th
e
9
5
5
-
ar
c
h
itectu
r
e
m
o
d
el'
s
o
u
tp
u
t
im
ag
e,
wh
ic
h
was
f
o
u
n
d
to
b
e
th
e
to
p
p
er
f
o
r
m
an
ce
at
a
s
ca
le
f
ac
to
r
o
f
2
,
is
d
is
p
lay
ed
in
Fig
u
r
e
7
(
b
)
.
Owin
g
to
t
h
e
m
an
u
s
cr
ip
t'
s
p
ag
e
lim
it
r
estrictio
n
s
,
t
h
e
r
eso
lu
tio
n
o
f
th
e
in
p
u
t
an
d
o
u
t
p
u
t
p
h
o
t
o
g
r
a
p
h
s
h
as
b
ee
n
r
ed
u
ce
d
b
y
f
o
u
r
tim
es.
T
h
is
o
u
tp
u
t
im
ag
e
s
h
o
ws
h
o
w
th
e
m
o
d
el
ca
n
r
ec
o
n
s
tr
u
ct
o
r
im
p
r
o
v
e
im
ag
e
s
at
twice
th
e
o
r
ig
in
al
s
ize
wi
th
o
u
t
s
ac
r
if
icin
g
q
u
ality
.
T
h
e
ca
p
ac
ity
o
f
th
e
9
5
5
ar
ch
itectu
r
es
to
p
r
o
d
u
ce
c
r
is
p
,
d
etailed
im
ag
es
h
i
g
h
lig
h
ts
th
e
r
em
ar
k
a
b
le
p
er
f
o
r
m
a
n
ce
o
f
th
e
s
y
s
tem
an
d
p
o
in
ts
o
u
t th
e
e
f
f
ec
tiv
en
ess
o
f
th
e
alg
o
r
ith
m
s
an
d
d
esig
n
i
n
h
an
d
lin
g
im
a
g
e
u
p
s
ca
lin
g
j
o
b
s
.
(
a)
(
b
)
Fig
u
r
e
7
.
T
esti
n
g
p
r
o
ce
s
s
(
a)
a
s
am
p
le
in
p
u
t im
ag
e
an
d
(
b
)
a
s
am
p
le
o
u
tp
u
t
g
iv
en
by
s
u
p
e
r
r
eso
lu
tio
n
(
SR
C
N
N
-
955)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
2
,
Feb
r
u
a
r
y
20
2
5
:
9
0
3
-
9
1
2
910
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
u
s
es
SR
C
NN
to
in
cr
ea
s
e
th
e
p
ictu
r
e
tr
an
s
m
is
s
io
n
ef
f
icien
cy
in
au
to
n
o
m
o
u
s
ca
r
s
n
av
ig
atin
g
lim
ited
n
etwo
r
k
co
n
d
itio
n
s
.
T
h
r
ee
d
if
f
e
r
en
t SR
C
NN
m
o
d
el
co
n
f
ig
u
r
atio
n
s
wer
e
cr
ea
ted
in
o
r
d
er
to
m
ee
t
th
e
u
n
i
q
u
e
n
ee
d
s
o
f
a
u
to
n
o
m
o
u
s
elec
tr
ic
v
eh
icles
t
h
at
o
p
er
ate
i
n
r
estricte
d
s
p
ac
es,
s
u
ch
B
an
d
u
n
g
,
I
n
d
o
n
esia'
s
KST
Sam
au
n
Sam
ad
ik
u
n
.
T
h
e
v
alid
ity
o
f
th
is
r
e
s
ea
r
ch
is
s
u
p
p
o
r
ted
b
y
th
e
u
s
e
o
f
a
lar
g
e
d
ataset
co
n
s
is
tin
g
o
f
7
6
5
p
h
o
to
s
,
w
h
ich
g
u
a
r
an
teed
th
at
th
e
m
o
d
els
wer
e
ass
es
s
ed
u
n
d
er
d
i
f
f
er
en
t
co
n
d
itio
n
s
to
ac
h
iev
e
o
p
tim
al
p
er
f
o
r
m
an
ce
.
T
h
e
in
p
u
t
im
ag
e
r
eso
lu
tio
n
in
th
is
ex
p
er
im
en
t
v
ar
ied
f
r
o
m
1
2
8
×
1
2
8
p
ix
els
to
2
5
6
×
2
5
6
p
ix
els.
I
n
co
n
tr
ast,
th
e
s
ca
le
f
ac
to
r
u
s
ed
af
f
ec
ted
th
e
o
u
tp
u
t
r
eso
lu
tio
n
,
wh
ich
r
an
g
ed
f
r
o
m
2
5
6
×
2
5
6
p
ix
els
to
5
1
2
×
5
1
2
p
ix
els.
E
v
en
with
th
e
s
m
allest
in
p
u
t
im
ag
e
s
ize
o
f
1
2
8
×
1
2
8
p
ix
els,
th
e
SR
C
NN
s
etu
p
m
an
ag
ed
to
ac
h
iev
e
a
PS
NR
o
f
2
9
.
5
9
6
b
y
u
p
s
izin
g
th
e
im
ag
e
to
5
1
2
×
5
1
2
p
i
x
els.
T
h
e
s
am
e
m
o
d
el
co
u
ld
u
p
s
ca
le
an
im
ag
e
with
a
n
in
p
u
t
r
eso
lu
tio
n
o
f
1
7
0
×
1
7
0
p
ix
els
to
5
1
0
×
5
1
0
p
i
x
els
with
a
PS
NR
o
f
3
0
.
9
2
6
.
T
h
e
f
i
n
al
p
h
o
to
s
lo
o
k
e
d
f
an
tast
ic.
I
t
will
b
e
m
o
r
e
ef
f
ec
tiv
e
to
tr
an
s
f
er
th
ese
im
ag
es
o
v
er
th
e
I
E
E
E
8
0
2
.
1
1
ac
W
i
-
Fi
n
etwo
r
k
,
b
u
t
th
er
e
will
b
e
n
o
ticea
b
le
d
ela
y
s
wh
e
n
u
s
in
g
th
e
L
o
R
a
n
etwo
r
k
a
1
2
8
×1
2
8
im
a
g
e
will
tak
e
5
.
6
s
e
co
n
d
s
to
b
r
o
ad
ca
s
t.
T
h
is
em
p
h
asizes
th
e
n
ec
ess
ity
f
o
r
ad
d
itio
n
al
s
tu
d
y
t
o
en
h
an
ce
SR
C
N
N
'
s
f
u
n
ctio
n
ality
,
es
p
ec
ially
with
r
eg
a
r
d
to
tr
an
s
m
is
s
io
n
ef
f
icien
cy
ac
r
o
s
s
v
ar
io
u
s
n
etwo
r
k
s
.
ACK
NO
WL
E
DG
E
M
E
NT
S
Sp
ec
ial
r
ec
o
g
n
itio
n
is
g
iv
en
to
th
e
I
n
d
o
n
esia
E
n
d
o
wm
e
n
t
F
u
n
d
f
o
r
E
d
u
ca
tio
n
(
Lemb
a
g
a
P
en
g
elo
la
Da
n
a
P
en
d
id
ika
n
)
,
wh
ich
p
r
o
v
id
ed
f
u
n
d
in
g
f
o
r
th
is
s
tu
d
y
as
p
ar
t
o
f
th
e
Natio
n
al
R
esear
ch
an
d
I
n
n
o
v
atio
n
Ag
en
cy
'
s
R
is
et
d
a
n
I
n
o
va
s
i u
n
tu
k
I
n
d
o
n
esia
Ma
ju
(
R
I
I
M)
p
r
o
g
r
am
(
B
R
I
N)
.
RE
F
E
R
E
NC
E
S
[
1
]
Y
.
G
u
o
,
Y
.
Li
u
,
A
.
O
e
r
l
e
m
a
n
s
,
S
.
La
o
,
S
.
W
u
,
a
n
d
M
.
S
.
L
e
w
,
“
D
e
e
p
l
e
a
r
n
i
n
g
f
o
r
v
i
s
u
a
l
u
n
d
e
r
st
a
n
d
i
n
g
:
a
r
e
v
i
e
w
,
”
N
e
u
ro
c
o
m
p
u
t
i
n
g
,
v
o
l
.
1
8
7
,
p
p
.
2
7
–
4
8
,
A
p
r
.
2
0
1
6
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.
2
0
1
5
.
0
9
.
1
1
6
.
[
2
]
F
.
F
r
a
u
n
d
o
r
f
e
r
a
n
d
D
.
S
c
a
r
a
m
u
z
z
a
,
“
V
i
s
u
a
l
o
d
o
m
e
t
r
y
:
p
a
r
t
I
I
:
mat
c
h
i
n
g
,
r
o
b
u
s
t
n
e
ss,
o
p
t
i
m
i
z
a
t
i
o
n
,
a
n
d
a
p
p
l
i
c
a
t
i
o
n
s,
”
I
E
EE
Ro
b
o
t
i
c
s
a
n
d
Au
t
o
m
a
t
i
o
n
M
a
g
a
z
i
n
e
,
v
o
l
.
1
9
,
n
o
.
2
,
p
p
.
7
8
–
90
,
Ju
n
.
2
0
1
2
,
d
o
i
:
1
0
.
1
1
0
9
/
M
R
A
.
2
0
1
2
.
2
1
8
2
8
1
0
.
[
3
]
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
.
[
4
]
R
.
H
a
r
t
l
e
y
a
n
d
A
.
Zi
ss
e
r
ma
n
,
M
u
l
t
i
p
l
e
v
i
e
w
g
e
o
m
e
t
r
y
i
n
c
o
m
p
u
t
e
r
v
i
s
i
o
n
.
C
a
m
b
r
i
d
g
e
U
n
i
v
e
r
s
i
t
y
P
r
e
ss,
2
0
0
4
,
d
o
i
:
1
0
.
1
0
1
7
/
C
B
O
9
7
8
0
5
1
1
8
1
1
6
8
5
.
[
5
]
R
.
S
z
e
l
i
s
k
i
,
C
o
m
p
u
t
e
r
v
i
s
i
o
n
.
C
h
a
m:
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
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
0
-
3
4
3
7
2
-
9.
[
6
]
M.
G
a
s
t
,
8
0
2
.
1
1
a
c
:
A
S
u
rv
i
v
a
l
G
u
i
d
e
.
O
’
R
e
i
l
l
y
M
e
d
i
a
,
I
n
c
,
2
0
1
3
.
[
7
]
N
.
S
.
R
a
v
i
n
d
r
a
n
a
t
h
,
I
.
S
i
n
g
h
,
A
.
P
r
a
sad
,
a
n
d
V
.
S
.
R
a
o
,
“
P
e
r
f
o
r
m
a
n
c
e
e
v
a
l
u
a
t
i
o
n
o
f
I
EEE
8
0
2
.
1
1
a
c
a
n
d
8
0
2
.
1
1
n
u
s
i
n
g
N
S
3
,
”
I
n
d
i
a
n
J
o
u
r
n
a
l
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
9
,
n
o
.
2
6
,
Ju
l
.
2
0
1
6
,
d
o
i
:
1
0
.
1
7
4
8
5
/
i
j
st
/
2
0
1
6
/
v
9
i
2
6
/
9
3
5
6
5
.
[
8
]
V
.
D
.
P
h
a
m,
V
.
V
i
s
h
n
e
v
s
k
y
,
D
.
C
.
N
g
u
y
e
n
,
a
n
d
R
.
K
i
r
i
c
h
e
k
,
“
Lo
R
a
mes
h
n
e
t
w
o
r
k
f
o
r
i
ma
g
e
t
r
a
n
sm
i
ssi
o
n
:
a
n
e
x
p
e
r
i
me
n
t
a
l
s
t
u
d
y
,
”
I
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
N
e
x
t
G
e
n
e
ra
t
i
o
n
Wi
re
d
/
Wi
rel
e
ss
N
e
t
w
o
rk
i
n
g
,
2
0
2
3
,
p
p
.
6
0
6
–
6
1
7
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
3
0
2
5
8
-
9
_
5
4
.
[
9
]
B
.
S
.
C
h
a
u
d
h
a
r
i
a
n
d
M
.
Ze
n
n
a
r
o
,
L
PW
A
N
t
e
c
h
n
o
l
o
g
i
e
s
f
o
r
I
o
T
a
n
d
M2
M
a
p
p
l
i
c
a
t
i
o
n
s
.
E
l
se
v
i
e
r
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
C
2
0
1
8
-
0
-
0
4
7
8
7
-
8.
[
1
0
]
W
.
G
u
i
b
e
n
e
,
J.
N
o
w
a
c
k
,
N
.
C
h
a
l
i
k
i
a
s
,
K
.
F
i
t
z
g
i
b
b
o
n
,
M
.
K
e
l
l
y
,
a
n
d
D
.
P
r
e
n
d
e
r
g
a
st
,
“
Ev
a
l
u
a
t
i
o
n
o
f
LPW
A
N
t
e
c
h
n
o
l
o
g
i
e
s
f
o
r
smar
t
c
i
t
i
e
s:
r
i
v
e
r
mo
n
i
t
o
r
i
n
g
u
se
-
c
a
s
e
,
”
i
n
2
0
1
7
I
EE
E
Wi
r
e
l
e
ss
C
o
m
m
u
n
i
c
a
t
i
o
n
s
a
n
d
N
e
t
w
o
r
k
i
n
g
C
o
n
f
e
ren
c
e
W
o
rks
h
o
p
s
(
WC
N
C
W)
,
I
E
EE,
M
a
r
.
2
0
1
7
,
p
p
.
1
–
5,
d
o
i
:
1
0
.
1
1
0
9
/
W
C
N
C
W
.
2
0
1
7
.
7
9
1
9
0
8
9
.
[
1
1
]
D
.
M
a
g
r
i
n
,
M
.
C
e
n
t
e
n
a
r
o
,
a
n
d
L.
V
a
n
g
e
l
i
s
t
a
,
“
P
e
r
f
o
r
m
a
n
c
e
e
v
a
l
u
a
t
i
o
n
o
f
L
o
R
a
n
e
t
w
o
r
k
s
i
n
a
smar
t
c
i
t
y
sce
n
a
r
i
o
,
”
i
n
2
0
1
7
I
EE
E
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
m
u
n
i
c
a
t
i
o
n
s (I
C
C
)
,
I
EEE,
M
a
y
2
0
1
7
,
p
p
.
1
–
7,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
.
2
0
1
7
.
7
9
9
6
3
8
4
.
[
1
2
]
C.
-
C
.
W
e
i
,
S
.
-
T.
C
h
e
n
,
a
n
d
P
.
-
Y
.
S
u
,
“
I
mag
e
t
r
a
n
sm
i
ssi
o
n
u
s
i
n
g
L
o
R
a
t
e
c
h
n
o
l
o
g
y
w
i
t
h
v
a
r
i
o
u
s
s
p
r
e
a
d
i
n
g
f
a
c
t
o
r
s
,
”
i
n
2
0
1
9
2
n
d
Wo
r
l
d
S
y
m
p
o
si
u
m
o
n
C
o
m
m
u
n
i
c
a
t
i
o
n
En
g
i
n
e
e
ri
n
g
(
W
S
C
E)
,
I
EEE,
D
e
c
.
2
0
1
9
,
p
p
.
4
8
–
5
2
,
d
o
i
:
1
0
.
1
1
0
9
/
W
S
C
E4
9
0
0
0
.
2
0
1
9
.
9
0
4
1
0
4
4
.
[
1
3
]
C
.
D
o
n
g
,
C
.
C
.
L
o
y
,
a
n
d
X
.
Ta
n
g
,
“
A
c
c
e
l
e
r
a
t
i
n
g
t
h
e
s
u
p
e
r
-
r
e
s
o
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
,
”
I
n
C
o
m
p
u
t
e
r
Vi
si
o
n
–
E
C
C
V
2
0
1
6
:
1
4
t
h
E
u
ro
p
e
a
n
C
o
n
f
e
re
n
c
e
,
Am
st
e
r
d
a
m
,
T
h
e
N
e
t
h
e
rl
a
n
d
s
,
2
0
1
6
,
p
p
.
3
9
1
–
4
0
7
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
3
1
9
-
4
6
4
7
5
-
6
_
2
5
.
[
1
4
]
C
.
D
o
n
g
,
C
.
C
.
L
o
y
,
K
.
H
e
,
a
n
d
X
.
Ta
n
g
,
“
Le
a
r
n
i
n
g
a
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
f
o
r
i
m
a
g
e
su
p
e
r
-
r
e
so
l
u
t
i
o
n
,
”
I
n
C
o
m
p
u
t
e
r
V
i
si
o
n
–
EC
C
V
2
0
1
6
:
1
4
t
h
E
u
ro
p
e
a
n
C
o
n
f
e
r
e
n
c
e
,
Am
s
t
e
rd
a
m
,
T
h
e
N
e
t
h
e
rl
a
n
d
s
,
2
0
1
4
,
p
p
.
1
8
4
–
1
9
9
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
3
1
9
-
1
0
5
9
3
-
2
_
1
3
.
[
1
5
]
K
.
U
m
e
h
a
r
a
,
J
.
O
t
a
,
a
n
d
T.
I
sh
i
d
a
,
“
A
p
p
l
i
c
a
t
i
o
n
o
f
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
f
o
r
e
n
h
a
n
c
i
n
g
i
ma
g
e
r
e
s
o
l
u
t
i
o
n
i
n
c
h
e
s
t
C
T,
”
J
o
u
r
n
a
l
D
i
g
i
t
I
m
a
g
i
n
g
,
v
o
l
.
3
1
,
n
o
.
4
,
p
p
.
4
4
1
–
4
5
0
,
A
u
g
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
0
2
7
8
-
0
1
7
-
0
0
3
3
-
z.
[
1
6
]
X
.
Ji
,
Y
.
L
u
,
a
n
d
L.
G
u
o
,
“
I
mag
e
s
u
p
e
r
-
r
e
s
o
l
u
t
i
o
n
w
i
t
h
d
e
e
p
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
EEE
F
i
rst
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
D
a
t
a
S
c
i
e
n
c
e
i
n
C
y
b
e
r
sp
a
c
e
(
D
S
C
)
,
I
EEE,
J
u
n
.
2
0
1
6
,
p
p
.
6
2
6
–
6
3
0
.
d
o
i
:
1
0
.
1
1
0
9
/
D
S
C
.
2
0
1
6
.
1
0
4
.
[
1
7
]
M
.
Z
h
a
o
,
H
.
P
e
n
g
,
L
.
L
i
,
a
n
d
Y
.
R
e
n
,
“
G
r
a
p
h
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
a
n
d
i
n
f
o
r
mer
f
o
r
mu
l
t
i
v
a
r
i
a
t
e
t
i
me
s
e
r
i
e
s
a
n
o
mal
y
d
e
t
e
c
t
i
o
n
,
”
S
e
n
so
rs
,
v
o
l
.
2
4
,
n
o
.
5
,
p
.
1
5
2
2
,
F
e
b
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
s2
4
0
5
1
5
2
2
.
[
1
8
]
R
.
S
h
a
n
mu
g
a
ma
n
i
,
D
e
e
p
l
e
a
rn
i
n
g
f
o
r
c
o
m
p
u
t
e
r
v
i
s
i
o
n
:
e
x
p
e
rt
t
e
c
h
n
i
q
u
e
s
t
o
t
ra
i
n
a
d
v
a
n
c
e
d
n
e
u
r
a
l
n
e
t
w
o
r
k
s
u
si
n
g
T
e
n
s
o
rFl
o
w
a
n
d
K
e
r
a
s
,
P
a
c
k
t
P
u
b
l
i
s
h
i
n
g
L
t
d
,
2
0
1
8
.
[
1
9
]
B
.
Li
m
,
S
.
S
o
n
,
H
.
K
i
m,
S
.
N
a
h
,
a
n
d
K
.
M
.
Le
e
,
“
E
n
h
a
n
c
e
d
d
e
e
p
r
e
si
d
u
a
l
n
e
t
w
o
r
k
s
f
o
r
si
n
g
l
e
i
ma
g
e
s
u
p
e
r
-
r
e
s
o
l
u
t
i
o
n
,
”
i
n
2
0
1
7
I
EEE
C
o
n
f
e
re
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
Re
c
o
g
n
i
t
i
o
n
W
o
rks
h
o
p
s
(
C
VP
RW
)
,
I
EEE,
Ju
l
.
2
0
1
7
,
p
p
.
1
1
3
2
–
1
1
4
0
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
W
.
2
0
1
7
.
1
5
1
.
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:
2
5
0
2
-
4
7
52
S
R
C
N
N
-
b
a
s
ed
ima
g
e
tr
a
n
s
mis
s
io
n
fo
r
a
u
to
n
o
mo
u
s
ve
h
icles in
limited
n
etw
o
r
k
… (
A
n
in
d
ya
A
fin
a
C
a
r
mely
a
)
911
[
2
0
]
A
.
S
.
S
a
t
y
a
w
a
n
,
S
.
U
.
P
r
i
n
i
,
S
.
A
.
R
.
A
b
u
-
B
a
k
a
r
,
a
n
d
Y
.
N
.
W
i
j
a
y
a
n
t
o
,
“
A
n
e
l
a
st
i
c
f
r
a
m
e
r
a
t
e
u
p
-
c
o
n
v
e
r
si
o
n
f
o
r
se
q
u
e
n
t
i
a
l
o
mn
i
d
i
r
e
c
t
i
o
n
a
l
i
m
a
g
e
s,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Ad
v
a
n
c
e
d
S
c
i
e
n
c
e
,
E
n
g
i
n
e
e
ri
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
,
v
o
l
.
1
2
,
n
o
.
1
,
p
.
1
5
8
,
Ja
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
8
5
1
7
/
i
j
a
sei
t
.
1
2
.
1
.
1
2
9
6
3
.
[
2
1
]
J.
H
e
a
t
o
n
,
“
I
a
n
g
o
o
d
f
e
l
l
o
w
,
y
o
sh
u
a
b
e
n
g
i
o
,
a
n
d
a
a
r
o
n
c
o
u
r
v
i
l
l
e
:
d
e
e
p
l
e
a
r
n
i
n
g
,
”
G
e
n
e
t
Pr
o
g
r
a
m
Ev
o
l
v
a
b
l
e
Ma
c
h
,
v
o
l
.
1
9
,
n
o
.
1
–
2
,
p
p
.
3
0
5
–
3
0
7
,
J
u
n
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
7
1
0
-
0
1
7
-
9
3
1
4
-
z.
[
2
2
]
C
.
M
.
B
i
sh
o
p
,
Pa
t
t
e
rn
re
c
o
g
n
i
t
i
o
n
a
n
d
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
.
S
p
r
i
n
g
e
r
N
e
w
Y
o
r
k
,
N
Y
,
2
0
0
6
.
[
2
3
]
H
.
Li
,
J.
Li
,
X
.
G
u
a
n
,
B
.
L
i
a
n
g
,
Y
.
La
i
,
a
n
d
X
.
Lu
o
,
“
R
e
se
a
r
c
h
o
n
o
v
e
r
f
i
t
t
i
n
g
o
f
d
e
e
p
l
e
a
r
n
i
n
g
,
”
i
n
2
0
1
9
1
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
S
e
c
u
ri
t
y
(
C
I
S
)
,
I
EEE,
D
e
c
.
2
0
1
9
,
p
p
.
7
8
–
8
1
,
d
o
i
:
1
0
.
1
1
0
9
/
C
I
S
.
2
0
1
9
.
0
0
0
2
5
.
[
2
4
]
A
.
G
é
r
o
n
,
H
a
n
d
s
-
o
n
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
w
i
t
h
sci
k
i
t
-
l
e
a
r
n
,
k
e
r
a
s
,
a
n
d
T
e
n
s
o
r
Fl
o
w
,
2
n
d
e
d
.
O
’
R
e
i
l
l
y
M
e
d
i
a
,
I
n
c
,
2
0
1
9
.
[
2
5
]
K
.
H
e
,
X
.
Z
h
a
n
g
,
S
.
R
e
n
,
a
n
d
J
.
S
u
n
,
“
D
e
e
p
r
e
si
d
u
a
l
l
e
a
r
n
i
n
g
f
o
r
i
ma
g
e
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
2
0
1
6
I
EEE
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
,
2
0
1
6
.
[
2
6
]
K
.
P
.
M
u
r
p
h
y
,
M
a
c
h
i
n
e
l
e
a
rn
i
n
g
:
a
p
ro
b
a
b
i
l
i
s
t
i
c
p
e
rs
p
e
c
t
i
v
e
,
T
h
e
M
I
T
P
r
e
ss,
2
0
1
2
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Ani
n
d
y
a
Afin
a
Ca
r
m
e
l
y
a
wa
s
b
o
rn
in
Ba
n
d
u
n
g
,
I
n
d
o
n
e
sia
,
i
n
2
0
0
3
.
S
h
e
is
n
o
w a
fin
a
l
-
y
e
a
r
b
a
c
h
e
lo
r'
s
stu
d
e
n
t
i
n
th
e
Tele
c
o
m
m
u
n
ica
ti
o
n
S
y
ste
m
De
p
a
rtme
n
t
a
t
Un
iv
e
rsitas
P
e
n
d
i
d
ik
a
n
In
d
o
n
e
sia
.
F
o
r
th
e
p
a
st
fo
u
r
m
o
n
t
h
s,
sh
e
h
a
s
sta
rted
a
n
a
p
p
re
n
ti
c
e
sh
ip
in
BRIN'
s
re
se
a
rc
h
a
n
d
d
e
v
e
lo
p
m
e
n
t
o
f
o
b
j
e
c
t
d
e
tec
ti
o
n
fo
r
a
u
t
o
n
o
m
o
u
s
v
e
h
icle
s.
His
c
u
rre
n
t
re
se
a
rc
h
in
tere
sts
a
re
m
a
c
h
in
e
lea
rn
in
g
,
a
rti
ficia
l
in
telli
g
e
n
c
e
,
a
n
d
s
u
p
e
r
re
so
lu
ti
o
n
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
n
i
n
d
y
a
fin
a
@
u
p
i.
e
d
u
.
Ar
ief
S
u
r
y
a
d
i
S
a
t
y
a
w
a
n
re
c
e
iv
e
d
h
is
m
a
ste
r'
s
d
e
g
re
e
i
n
El
e
c
tri
c
a
l
En
g
in
e
e
ri
n
g
fro
m
In
stit
u
t
Tek
n
o
l
o
g
i
Ba
n
d
u
n
g
in
I
n
d
o
n
e
sia
i
n
2
0
0
7
.
In
2
0
1
9
,
h
e
re
c
e
iv
e
d
h
is
C
o
m
p
u
ter
a
n
d
Co
m
m
u
n
ica
ti
o
n
E
n
g
in
e
e
rin
g
D
o
c
to
ra
te
fro
m
Was
e
d
a
Un
i
v
e
rsity
,
Ja
p
a
n
.
S
i
n
c
e
1
9
9
7
,
h
e
h
a
s
wo
rk
e
d
f
o
r
th
e
I
n
d
o
n
e
sia
n
I
n
stit
u
te
o
f
S
c
ien
c
e
s,
wh
o
se
n
a
m
e
wa
s
c
h
a
n
g
e
d
to
th
e
Na
ti
o
n
a
l
Re
se
a
rc
h
a
n
d
In
n
o
v
a
ti
o
n
Ag
e
n
c
y
,
In
d
o
n
e
sia
,
in
2
0
2
1
.
His
re
se
a
rc
h
in
tere
sts
a
re
a
p
p
li
c
a
ti
o
n
s o
f
m
a
c
h
in
e
lea
rn
in
g
,
a
rti
ficia
l
i
n
telli
g
e
n
c
e
,
LIDAR
-
b
a
se
d
o
b
j
e
c
t
d
e
tec
ti
o
n
,
a
n
d
v
isu
a
l
c
o
m
m
u
n
ica
ti
o
n
b
a
se
d
o
n
m
a
c
h
i
n
e
lea
rn
in
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
rie0
2
1
@b
r
in
.
g
o
.
id
.
G
a
lu
r
a
Mu
h
a
m
m
a
d
S
u
r
a
n
e
g
a
r
a
h
o
ld
s
a
M
a
ste
r'
s
d
e
g
re
e
in
El
e
c
tri
c
a
l
En
g
i
n
e
e
rin
g
sp
e
c
ializin
g
i
n
Tele
m
a
ti
c
s
(Tele
c
o
m
m
u
n
ica
ti
o
n
s
a
n
d
I
n
fo
rm
a
ti
c
s)
a
n
d
Tele
c
o
m
m
u
n
ica
ti
o
n
Ne
two
r
k
s
fr
o
m
In
stit
u
t
Tek
n
o
l
o
g
i
Ba
n
d
u
n
g
in
In
d
o
n
e
sia
.
Cu
rre
n
tl
y
,
h
e
se
rv
e
s
a
s
th
e
He
a
d
o
f
th
e
S
tu
d
y
P
ro
g
ra
m
a
t
Un
i
v
e
rsitas
P
e
n
d
id
i
k
a
n
In
d
o
n
e
sia
.
His
a
c
a
d
e
m
ic
a
n
d
re
se
a
rc
h
e
n
d
e
a
v
o
rs
a
re
c
e
n
tere
d
a
ro
u
n
d
th
e
field
s
o
f
te
lec
o
m
m
u
n
ica
ti
o
n
a
n
d
s
o
ft
c
o
m
p
u
ti
n
g
,
with
a
p
a
rti
c
u
lar
f
o
c
u
s
o
n
n
e
two
r
k
i
n
g
tec
h
n
o
lo
g
ies
a
n
d
a
rti
ficia
l
in
telli
g
e
n
c
e
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
g
a
lu
ra
m
s@
u
p
i.
e
d
u
.
Mo
k
h
a
m
a
m
a
d
Mi
r
z
a
Et
n
isa
H
a
q
iq
i
e
a
rn
e
d
a
Ba
c
h
e
lo
r
o
f
S
c
ien
c
e
in
El
e
c
tri
c
a
l
En
g
i
n
e
e
rin
g
fr
o
m
Un
iv
e
rsitas
G
a
ru
t
in
2
0
1
7
a
n
d
a
M
a
ste
r
o
f
S
c
ien
c
e
in
En
g
i
n
e
e
rin
g
fr
o
m
Un
iv
e
rsitas
In
d
o
n
e
sia
in
2
0
2
2
.
He
is
a
lec
tu
re
r
in
t
h
e
e
n
g
in
e
e
rin
g
d
e
p
a
rtme
n
t
o
f
Un
i
v
e
rsitas
G
a
ru
t.
His
re
se
a
r
c
h
in
tere
sts
a
re
o
n
su
p
e
r
re
so
lu
ti
o
n
,
a
rti
ficia
l
in
telli
g
e
n
c
e
,
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
o
k
h
a
m
m
a
d
.
m
irza
2
1
@u
i.
a
c
.
id
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
37
,
No
.
2
,
Feb
r
u
a
r
y
20
2
5
:
9
0
3
-
9
1
2
912
H
e
lfy
S
u
sil
a
wa
ti
o
b
tain
e
d
a
m
a
ste
r'
s
d
e
g
re
e
in
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
fr
o
m
In
sti
tu
t
Tek
n
o
lo
g
i
Ba
n
d
u
n
g
i
n
Ba
n
d
u
n
g
,
I
n
d
o
n
e
sia
in
2
0
1
4
a
fter
e
a
rn
in
g
a
b
a
c
h
e
lo
r'
s
d
e
g
re
e
in
p
h
y
sic
a
l
e
d
u
c
a
ti
o
n
fr
o
m
UIN
S
u
n
a
n
G
u
n
u
n
g
Dja
ti
Ba
n
d
u
n
g
,
Ba
n
d
u
n
g
,
I
n
d
o
n
e
sia
,
i
n
2
0
1
1
.
S
in
c
e
2
0
1
2
,
sh
e
h
a
s
b
e
e
n
e
m
p
lo
y
e
d
a
s
a
lec
tu
re
r
a
t
U
n
iv
e
rsitas
G
a
ru
t
in
In
d
o
n
e
sia
.
Art
ifi
c
ial
in
telli
g
e
n
c
e
,
ima
g
e
p
r
o
c
e
ss
in
g
,
a
n
d
c
o
n
tr
o
l
sy
ste
m
s
a
re
a
m
o
n
g
h
e
r
a
re
a
s
o
f
in
tere
st
in
st
u
d
y
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
h
e
l
fy
.
su
silaw
a
ti
@u
n
ig
a
.
a
c
.
i
d
.
Niz
a
r
Ala
m
H
a
m
d
a
n
i
re
c
e
iv
e
d
a
d
o
c
to
ra
te
in
M
a
n
a
g
e
m
e
n
t
a
n
d
Ed
u
c
a
ti
o
n
a
t
t
h
e
In
d
o
n
e
sia
n
Un
iv
e
rsity
o
f
E
d
u
c
a
ti
o
n
,
c
o
m
p
letin
g
a
m
a
ste
r'
s P
ro
g
ra
m
a
t
th
e
Ba
n
d
u
n
g
In
sti
tu
te
o
f
Tec
h
n
o
l
o
g
y
a
n
d
Telk
o
m
Un
i
v
e
rsity
.
He
h
a
s
p
u
b
li
sh
e
d
re
se
a
rc
h
in
m
a
n
a
g
e
m
e
n
t,
i
n
fo
rm
a
ti
o
n
tec
h
n
o
l
o
g
y
,
a
n
d
e
d
u
c
a
ti
o
n
a
l
tec
h
n
o
lo
g
y
.
He
h
a
s
v
a
rio
u
s
e
x
p
e
rien
c
e
s
a
s
a
c
o
n
su
lt
a
n
t
i
n
e
c
o
n
o
m
ics
a
n
d
in
f
o
rm
a
ti
o
n
tec
h
n
o
lo
g
y
a
n
d
a
s
a
n
e
n
trep
re
n
e
u
r
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
n
iza
r_
h
a
m
d
a
n
i@
u
n
i
g
a
.
a
c
.
id
.
Pu
p
u
t
D
a
n
i
Pra
se
ty
o
Adi
g
ra
d
u
a
ted
wit
h
h
o
n
o
rs
fr
o
m
S
TM
IK
AK
AK
OM
Yo
g
y
a
k
a
rta
i
n
2
0
0
8
wit
h
a
b
a
c
h
e
lo
r'
s
d
e
g
re
e
in
i
n
fo
rm
a
ti
c
e
n
g
i
n
e
e
rin
g
,
U
n
iv
e
rsitas
Ha
sa
n
n
u
d
i
n
in
M
a
k
a
ss
a
r,
In
d
o
n
e
sia
in
2
0
1
1
with
a
m
a
ste
r'
s
d
e
g
re
e
in
e
n
g
in
e
e
rin
g
,
a
n
d
Ka
n
a
z
a
wa
Un
iv
e
rsity
in
Ja
p
a
n
i
n
2
0
2
0
wit
h
a
d
o
c
to
ra
te.
His
p
ri
m
a
ry
a
re
a
s
o
f
stu
d
y
i
n
c
lu
d
e
lo
w
p
o
we
r
wid
e
a
re
a
(LP
WA)
a
n
d
l
o
w
p
o
we
r
wid
e
a
re
a
n
e
two
rk
s
(L
P
WAN)
e
m
p
lo
y
in
g
v
a
rio
u
s
RF
tec
h
n
o
l
o
g
ies
a
n
d
lo
n
g
ra
n
g
e
(Lo
Ra
)
ra
d
io
fre
q
u
e
n
c
y
a
p
p
li
c
a
ti
o
n
s.
S
in
c
e
2
0
2
2
,
h
e
h
a
s
wo
r
k
e
d
in
th
e
Tele
c
o
m
m
u
n
ica
ti
o
n
s
Re
se
a
rc
h
Ce
n
ter
fo
r
t
h
e
Na
ti
o
n
a
l
Re
se
a
rc
h
a
n
d
In
n
o
v
a
ti
o
n
A
g
e
n
c
y
(BRIN).
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
p
u
p
u
0
0
8
@b
ri
n
.
g
o
.
i
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.