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.
42
,
No
.
3
,
J
u
n
e
2
0
2
6
,
p
p
.
8
5
6
~8
6
4
I
SS
N:
2
5
0
2
-
4
7
5
2
,
DOI
: 1
0
.
1
1
5
9
1
/ijeecs.v
42
.i
3
.
pp
856
-
8
6
4
856
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs.ia
esco
r
e.
co
m
A t
ra
nsf
er learni
ng
appro
a
ch f
o
r r
ea
l
-
time
detec
tion
and
cla
ss
ificatio
n of I
ndo
nesia
n
c
o
ins
Nur
H
a
dis
uk
m
a
na
,
R.
B
.
Wa
hy
u,
Ste
wa
rt
Q
iu
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
m
a
t
i
c
s,
F
a
c
u
l
t
y
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
,
P
r
e
si
d
e
n
t
U
n
i
v
e
r
si
t
y
,
C
i
k
a
r
a
n
g
,
I
n
d
o
n
e
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
2
8
,
2
0
2
5
R
ev
is
ed
Ma
r
2
3
,
2
0
2
6
Acc
ep
ted
Ma
y
2
6
,
2
0
2
6
Au
to
m
a
ted
c
u
rre
n
c
y
re
c
o
g
n
it
i
o
n
p
lay
s
a
n
imp
o
r
tan
t
ro
le
in
b
a
n
k
i
n
g
a
u
to
m
a
ti
o
n
,
re
tail
sy
ste
m
s,
a
n
d
a
ss
isti
v
e
tec
h
n
o
l
o
g
ies
.
W
h
il
e
b
a
n
k
n
o
te
re
c
o
g
n
it
i
o
n
h
a
s
b
e
e
n
e
x
ten
si
v
e
ly
stu
d
ied
,
c
o
in
re
c
o
g
n
it
io
n
re
m
a
in
s
c
h
a
ll
e
n
g
i
n
g
d
u
e
t
o
sm
a
ll
o
b
jec
t
siz
e
,
m
e
talli
c
re
flec
tan
c
e
,
v
is
u
a
l
sim
il
a
rit
y
a
c
ro
ss
d
e
n
o
m
in
a
ti
o
n
s,
a
n
d
c
ircu
l
a
ti
o
n
-
i
n
d
u
c
e
d
we
a
r
.
Th
is
stu
d
y
p
ro
p
o
se
s
a
re
a
l
-
ti
m
e
sy
ste
m
fo
r
d
e
tec
ti
n
g
a
n
d
c
l
a
ss
ify
in
g
In
d
o
n
e
sia
n
c
o
i
n
s
u
sin
g
a
tran
sfe
r
lea
rn
in
g
–
b
a
se
d
d
e
e
p
le
a
rn
in
g
a
p
p
ro
a
c
h
.
A
c
u
ra
ted
d
a
tas
e
t
wa
s
d
e
v
e
lo
p
e
d
to
a
d
d
re
ss
th
e
lac
k
o
f
p
u
b
li
c
ly
a
v
a
il
a
b
le
trai
n
in
g
d
a
t
a
fo
r
th
is
d
o
m
a
in
.
Th
e
m
o
d
e
l
wa
s
in
it
ialize
d
with
p
re
trai
n
e
d
we
ig
h
ts
a
n
d
fi
n
e
-
tu
n
e
d
to
a
d
a
p
t
t
o
t
h
e
s
p
e
c
ifi
c
c
o
in
c
las
sifi
c
a
ti
o
n
tas
k
.
E
x
p
e
rime
n
tal
e
v
a
lu
a
t
io
n
o
n
a
n
u
n
se
e
n
tes
t
se
t
d
e
m
o
n
stra
tes
h
i
g
h
d
e
tec
ti
o
n
a
c
c
u
ra
c
y
wh
i
le
m
a
in
tain
in
g
re
a
l
-
ti
m
e
in
fe
re
n
c
e
p
e
rfo
rm
a
n
c
e
.
Q
u
a
li
tativ
e
a
n
a
ly
sis
u
n
d
e
r
c
h
a
ll
e
n
g
in
g
c
o
n
d
i
ti
o
n
s
—
in
c
l
u
d
i
n
g
g
lare
,
l
o
w
il
l
u
m
i
n
a
ti
o
n
,
o
c
c
lu
sio
n
,
a
n
d
c
o
i
n
we
a
r
—
re
v
e
a
ls
o
p
e
ra
ti
o
n
a
l
li
m
it
a
ti
o
n
s
a
n
d
d
e
fin
e
s
ro
b
u
st
n
e
ss
b
o
u
n
d
a
ries
.
Th
e
fin
d
i
n
g
s c
o
n
firm t
h
a
t
fro
z
e
n
-
b
a
c
k
b
o
n
e
t
ra
n
sfe
r
lea
rn
in
g
p
r
o
v
i
d
e
s a
n
e
ffe
c
ti
v
e
a
n
d
c
o
m
p
u
tatio
n
a
ll
y
e
fficie
n
t
st
ra
teg
y
fo
r
a
d
a
p
ti
n
g
sta
te
-
of
-
t
h
e
-
a
rt
o
b
jec
t
d
e
tec
to
rs t
o
lo
w
-
re
so
u
rc
e
,
d
o
m
a
in
-
sp
e
c
ifi
c
c
u
rre
n
c
y
re
c
o
g
n
it
i
o
n
tas
k
s.
K
ey
w
o
r
d
s
:
C
o
in
r
ec
o
g
n
itio
n
C
o
m
p
u
ter
s
cien
ce
I
n
d
o
n
esian
cu
r
r
e
n
cy
Ob
ject
d
etec
tio
n
T
r
an
s
f
er
lear
n
i
n
g
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
:
Nu
r
Had
is
u
k
m
an
a
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atics,
Facu
lty
o
f
C
o
m
p
u
ter
Scien
ce
,
Pr
esid
en
t U
n
iv
er
s
ity
J
ab
ab
ek
a
E
d
u
ca
tio
n
Par
k
,
C
ik
ar
an
g
,
B
ek
asi,
J
awa
B
ar
at,
I
n
d
o
n
esia
E
m
ail: n
u
r
h
a
d
is
u
k
m
an
a@
p
r
es
id
en
t.a
c.
id
1.
I
NT
RO
D
UCT
I
O
N
Au
to
m
ated
s
y
s
tem
s
ar
e
b
ec
o
m
in
g
m
o
r
e
p
r
ev
ale
n
t,
d
r
iv
i
n
g
th
e
r
ap
id
ad
v
an
ce
m
e
n
t
o
f
o
b
je
ct
d
etec
tio
n
tech
n
o
lo
g
y
f
o
r
ap
p
licatio
n
s
s
u
ch
as
b
an
k
n
o
te
d
etec
tio
n
in
f
in
an
cial
au
t
o
m
atio
n
an
d
r
etai
l
s
elf
-
ch
ec
k
o
u
t
[
1
]
.
Alth
o
u
g
h
s
ig
n
if
ican
t
p
r
o
g
r
es
s
h
as
b
ee
n
m
ad
e
in
b
an
k
n
o
te
r
ec
o
g
n
itio
n
,
c
o
in
r
ec
o
g
n
itio
n
p
r
esen
ts
u
n
iq
u
e
ch
allen
g
es d
u
e
to
s
m
all
o
b
ject
s
ize,
v
ar
y
in
g
lig
h
t c
o
n
d
itio
n
s
,
g
lar
e
f
r
o
m
m
etalli
c
s
u
r
f
ac
es,
an
d
co
in
wea
r
f
r
o
m
cir
cu
latio
n
[
2
]
,
[
3
]
.
T
h
is
s
tu
d
y
f
o
cu
s
es
o
n
d
etec
ti
n
g
an
d
class
if
y
in
g
f
o
u
r
I
n
d
o
n
esian
co
in
s
f
r
o
m
th
e
2
0
1
6
s
er
ies:
I
DR
1
0
0
,
2
0
0
,
5
0
0
,
an
d
1
0
0
0
.
T
h
ese
co
in
s
p
r
esen
t sig
n
if
ican
t c
h
a
llen
g
es d
u
e
to
v
is
u
al
am
b
ig
u
it
y
—
s
im
ilar
m
etalli
c
f
in
is
h
es,
m
ar
g
in
al
s
ize
d
if
f
er
en
ce
s
,
an
d
th
e
ab
s
en
ce
o
f
p
u
b
licly
av
ailab
le
r
ef
er
e
n
ce
d
at
asets
.
T
h
e
r
esear
ch
g
ap
s
ar
e:
(
i
)
ab
s
en
ce
o
f
a
cu
r
ated
an
n
o
tated
d
ataset
o
f
I
n
d
o
n
esian
co
in
s
,
an
d
(
ii
)
lack
o
f
ef
f
icien
t
r
ea
l
-
tim
e
d
etec
tio
n
m
o
d
els ca
p
ab
le
o
f
h
an
d
lin
g
v
is
u
al
am
b
ig
u
ities
with
lim
ited
d
ata.
Dee
p
lear
n
in
g
s
o
lu
tio
n
s
,
s
p
e
cif
ically
C
o
n
v
o
lu
tio
n
al
Ne
u
r
al
Netwo
r
k
s
(
C
NNs),
d
eliv
er
ef
f
icien
t
o
b
ject
d
etec
tio
n
[
4
]
.
Sin
g
le
-
s
tag
e
d
etec
to
r
s
d
ef
in
e
c
o
in
d
e
tectio
n
as
a
s
in
g
le
r
eg
r
ess
io
n
p
r
o
b
lem
[
5
]
,
[
6
]
.
YOL
O
d
e
liv
er
s
s
tr
o
n
g
p
er
f
o
r
m
an
ce
,
o
f
f
er
in
g
e
x
ce
llen
t
b
ala
n
ce
b
etwe
en
s
p
ee
d
an
d
ac
cu
r
a
cy
in
r
ea
l
-
tim
e
[
7
]
,
[
8
]
.
YOL
Ov
8
,
th
e
m
o
s
t
ad
v
an
ce
d
iter
atio
n
,
p
r
o
v
id
es
en
h
an
ce
d
n
etwo
r
k
a
r
ch
itectu
r
e
f
o
r
s
tate
-
of
-
th
e
-
ar
t
p
er
f
o
r
m
an
ce
[
9
]
,
[
1
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
A
tr
a
n
s
fer lea
r
n
in
g
a
p
p
r
o
a
c
h
f
o
r
r
ea
l
-
time
d
etec
tio
n
a
n
d
cla
s
s
ifica
tio
n
…
(
N
u
r
Ha
d
is
u
kma
n
a
)
857
T
r
ain
in
g
d
ee
p
lear
n
in
g
m
o
d
el
s
f
r
o
m
s
cr
atch
r
eq
u
i
r
es
en
o
r
m
o
u
s
an
n
o
tated
d
ata,
ty
p
ically
u
n
av
ailab
le
f
o
r
d
o
m
ain
-
s
p
ec
if
ic
task
s
.
T
h
e
r
ef
o
r
e,
we
em
p
lo
y
tr
an
s
f
er
lear
n
in
g
,
r
eu
s
in
g
a
m
o
d
el
d
ev
elo
p
ed
f
o
r
o
n
e
task
as
th
e
s
tar
tin
g
p
o
in
t
f
o
r
a
r
elate
d
task
.
T
r
a
n
s
f
er
lear
n
in
g
lev
er
a
g
es
p
r
e
-
tr
ai
n
ed
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
,
en
a
b
lin
g
b
etter
k
n
o
wled
g
e
ac
q
u
is
itio
n
with
less
d
ata
[
11
]
,
[
1
2
]
.
T
h
is
s
tu
d
y
e
m
p
lo
y
s
t
r
an
s
f
er
lear
n
in
g
with
f
r
o
ze
n
b
ac
k
b
o
n
e
lay
er
s
to
e
n
ab
le
h
ig
h
-
ac
cu
r
ac
y
I
n
d
o
n
esian
co
in
r
ec
o
g
n
itio
n
d
esp
ite
a
s
m
all
d
ataset,
wh
ile
m
ain
tain
in
g
r
o
b
u
s
tn
ess
u
n
d
e
r
r
ea
l
-
wo
r
ld
v
a
r
iatio
n
s
in
lig
h
tin
g
,
o
cc
lu
s
io
n
,
an
d
c
o
in
wea
r
.
T
h
e
p
r
im
ar
y
co
n
tr
i
b
u
tio
n
s
a
r
e
to
:
(
i
)
im
p
lem
e
n
t
a
r
ea
l
-
ti
m
e
co
in
d
etec
tio
n
m
o
d
el
u
s
i
n
g
tr
a
n
s
f
er
lear
n
in
g
with
YOL
Ov
8
;
(
ii
)
d
ev
elo
p
a
n
ewly
c
o
n
s
tr
u
cted
an
n
o
tated
d
ataset
o
f
2
0
1
6
I
n
d
o
n
esi
an
co
in
s
;
(
3
)
co
n
d
u
ct
c
o
m
p
r
e
h
en
s
iv
e
ev
al
u
atio
n
to
q
u
an
tify
p
er
f
o
r
m
an
ce
an
d
h
ig
h
lig
h
t lim
itatio
n
s
.
E
ar
ly
co
m
p
u
ter
v
is
io
n
ap
p
r
o
a
ch
es
r
elied
o
n
h
a
n
d
-
cr
a
f
ted
f
e
atu
r
es
lik
e
SIFT
o
r
HOG
co
m
b
in
ed
with
SVM
clas
s
if
ier
s
[
13
]
,
[
1
4
]
.
T
h
ese
m
eth
o
d
s
wer
e
ef
f
ec
tiv
e
b
u
t
c
o
m
p
u
tatio
n
ally
cu
m
b
er
s
o
m
e
an
d
lack
ed
r
o
b
u
s
tn
ess
to
r
ea
l
-
wo
r
l
d
v
ar
ia
b
ilit
y
[
1
5
]
,
[
1
6
]
.
T
h
e
ad
v
e
n
t
o
f
d
ee
p
lear
n
in
g
r
ev
o
lu
tio
n
ized
o
b
ject
d
etec
tio
n
[
17
]
.
T
wo
-
s
tag
e
d
etec
to
r
s
lik
e
Fas
ter
R
-
C
NN
ac
h
iev
ed
h
ig
h
p
r
ec
is
io
n
b
u
t
in
tr
o
d
u
ce
d
co
m
p
u
tatio
n
al
o
v
er
h
ea
d
u
n
s
u
itab
le
f
o
r
r
ea
l
-
tim
e
d
e
p
lo
y
m
en
t
[
1
8
]
,
[
1
9
]
.
T
h
is
ca
taly
z
ed
s
in
g
le
-
s
tag
e
d
etec
to
r
s
,
with
th
e
YOL
O
f
am
ily
b
ec
o
m
in
g
th
e
d
ef
in
in
g
a
r
ch
ite
ctu
r
e.
YOL
Ov
8
p
r
o
v
id
es
o
p
ti
m
al
b
alan
ce
f
o
r
r
ea
l
-
tim
e
ap
p
licatio
n
s
th
r
o
u
g
h
its
s
in
g
le
-
s
tag
e,
en
d
-
to
-
en
d
f
r
am
e
wo
r
k
[
20
].
R
ec
e
n
t
ad
v
an
ce
s
h
ig
h
lig
h
t
t
h
e
im
p
o
r
tan
ce
o
f
lar
g
e
-
s
ca
le
p
r
e
-
tr
ain
i
n
g
f
o
r
d
ata
-
s
ca
r
c
e
s
ettin
g
s
.
Xu
et
a
l.
[
21
]
d
e
m
o
n
s
tr
ate
t
h
e
ef
f
ec
tiv
e
n
ess
o
f
p
r
e
-
tr
ain
e
d
b
ac
k
b
o
n
es
i
n
r
e
d
u
cin
g
r
eli
an
ce
o
n
ex
ten
s
iv
e
an
n
o
tatio
n
s
,
m
o
tiv
atin
g
o
u
r
f
r
o
ze
n
-
b
ac
k
b
o
n
e
tr
a
n
s
f
er
lear
n
in
g
s
tr
ateg
y
.
L
in
et
a
l.
[
22
]
s
h
o
w
s
in
g
le
-
s
tag
e
d
etec
to
r
s
r
em
ain
d
o
m
in
an
t
f
o
r
r
ea
l
-
tim
e
ap
p
licatio
n
s
d
u
e
to
ef
f
icien
t e
n
d
-
to
-
en
d
d
esig
n
.
Sev
er
al
s
tu
d
ies
h
av
e
ap
p
lied
th
ese
m
o
d
els
to
cu
r
r
en
cy
r
ec
o
g
n
itio
n
.
Pra
b
u
et
a
l.
[
23
]
d
e
m
o
n
s
tr
ated
YOL
Ov
5
f
o
r
d
etec
tin
g
an
d
r
e
co
g
n
izin
g
I
n
d
ian
co
i
n
s
.
R
o
s
alin
a
[
24
]
ap
p
lied
YOL
Ov
8
to
an
cien
t
I
n
d
o
n
esian
co
in
s
,
p
r
o
v
i
n
g
th
e
m
o
d
el
’
s
ef
f
icien
cy
wh
ile
n
o
tin
g
d
if
f
er
en
c
es
f
r
o
m
m
o
d
e
r
n
cir
cu
lated
c
o
in
s
.
An
o
th
er
s
tu
d
y
im
p
lem
en
ted
YOL
Ov
8
to
d
et
ec
t
I
n
d
o
n
esian
b
an
k
n
o
te
d
e
n
o
m
in
atio
n
s
[
25
]
,
v
alid
atin
g
t
h
e
m
o
d
el
’
s
ef
f
icac
y
wh
ile
ac
k
n
o
wled
g
i
n
g
d
if
f
er
en
t
v
is
u
al
ch
allen
g
es c
o
m
p
ar
ed
t
o
m
etallic
co
in
s
u
r
f
ac
es.
T
ab
le
1
p
r
esen
ts
a
co
m
p
ar
is
o
n
b
etwe
en
o
u
r
ap
p
r
o
ac
h
a
n
d
r
elate
d
s
tu
d
ies.
Un
lik
e
p
r
io
r
wo
r
k
s
f
o
cu
s
in
g
o
n
I
n
d
ian
co
in
s
[
23
]
,
an
cien
t
I
n
d
o
n
esian
co
in
s
[
24
]
,
o
r
I
n
d
o
n
esian
b
a
n
k
n
o
tes
[
25
]
,
th
is
s
tu
d
y
tar
g
ets
m
o
d
er
n
I
n
d
o
n
esian
co
i
n
s
f
r
o
m
th
e
2
0
1
6
s
er
ies
,
wh
ich
p
o
s
e
d
is
tin
ct
ch
allen
g
es
d
u
e
t
o
m
etallic
r
ef
lecta
n
ce
,
v
is
u
al
s
im
ilar
ity
ac
r
o
s
s
d
en
o
m
in
atio
n
s
,
an
d
cir
cu
latio
n
-
in
d
u
ce
d
wea
r
.
A
k
ey
co
n
t
r
ib
u
tio
n
is
ex
p
licit
u
s
e
o
f
tr
an
s
f
er
lear
n
in
g
with
f
r
o
ze
n
b
ac
k
b
o
n
e
lay
er
s
—
a
s
tr
ateg
y
n
o
t
ad
o
p
ted
in
c
o
m
p
ar
e
d
s
tu
d
i
es.
B
y
lev
er
ag
in
g
p
r
e
-
tr
ain
ed
r
ep
r
esen
tatio
n
s
,
o
u
r
m
eth
o
d
ac
h
iev
es
h
ig
h
ac
cu
r
a
cy
with
lim
ited
d
ata
wh
ile
m
a
in
tain
in
g
r
ea
l
-
tim
e
p
er
f
o
r
m
an
ce
.
T
h
is
s
tu
d
y
p
r
esen
ts
th
e
f
ir
s
t
ap
p
licatio
n
o
f
YOL
Ov
8
with
f
r
o
ze
n
-
b
ac
k
b
o
n
e
tr
a
n
s
f
er
lea
r
n
in
g
f
o
r
m
o
d
er
n
I
n
d
o
n
esian
co
i
n
r
ec
o
g
n
itio
n
.
C
o
m
p
ar
ativ
e
a
n
aly
s
is
co
n
f
ir
m
s
ex
is
tin
g
s
tu
d
ies
d
o
n
o
t
em
p
lo
y
tr
a
n
s
f
er
lear
n
in
g
o
n
lim
ited
co
in
d
ata
s
ets.
Ou
r
ap
p
r
o
ac
h
d
em
o
n
s
tr
ates
th
at
lev
er
ag
in
g
p
r
e
-
tr
ain
ed
f
ea
tu
r
es
en
ab
les
h
ig
h
ac
cu
r
ac
y
an
d
r
o
b
u
s
tn
ess
wh
ile
r
ed
u
cin
g
tr
ain
in
g
tim
e
a
n
d
co
m
p
u
tatio
n
al
c
o
s
t
.
T
ab
le
1
.
C
o
m
p
a
r
is
o
n
o
f
r
elev
a
n
t w
o
r
k
s
F
e
a
t
u
r
e
I
n
d
i
a
n
c
o
i
n
d
e
t
e
c
t
i
o
n
a
n
d
r
e
c
o
g
n
i
t
i
o
n
[
2
3
]
D
e
t
e
c
t
i
o
n
o
f
a
n
c
i
e
n
t
i
n
d
o
n
e
si
a
n
c
o
i
n
s [
2
4
]
D
e
n
o
mi
n
a
t
i
o
n
d
e
t
e
c
t
i
o
n
[
2
5
]
Th
i
s
r
e
sea
r
c
h
C
o
r
e
m
o
d
e
l
Y
O
LO
v
5
Y
O
LO
v
8
Y
O
LO
v
8
Y
O
LO
v
8
U
se
o
f
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
f
r
o
z
e
n
l
a
y
e
r
s
No
No
No
Y
e
s
R
e
a
l
-
t
i
me
d
e
t
e
c
t
i
o
n
No
No
Y
e
s
Y
e
s
T
a
r
g
e
t
c
u
r
r
e
n
c
y
I
n
d
i
a
n
c
o
i
n
s
c
u
r
r
e
n
c
y
(
R
u
p
e
e
)
A
n
c
i
e
n
t
i
n
d
o
n
e
si
a
n
c
o
i
n
s
I
n
d
o
n
e
si
a
n
b
a
n
k
n
o
t
e
s
c
u
r
r
e
n
c
y
i
ssu
e
d
f
r
o
m
2
0
0
9
t
o
2
0
2
3
2
0
1
6
seri
e
s
i
n
d
o
n
e
si
a
n
c
o
i
n
s
c
u
r
r
e
n
c
y
2.
M
E
T
H
O
D
W
e
ad
o
p
ted
a
s
tr
u
ctu
r
ed
m
eth
o
d
o
lo
g
y
u
s
in
g
YOL
Ov
8
with
tr
an
s
f
er
lear
n
in
g
,
e
n
co
m
p
ass
in
g
s
y
s
tem
d
esig
n
,
d
ataset
cr
ea
tio
n
,
m
o
d
e
l c
o
n
f
ig
u
r
atio
n
,
a
n
d
tr
ain
i
n
g
s
etu
p
.
2
.
1
.
Sy
s
t
e
m
o
v
er
v
iew
T
h
e
s
y
s
tem
is
a
d
esk
to
p
ap
p
licatio
n
p
r
o
v
id
in
g
an
e
n
d
-
to
-
en
d
p
ip
elin
e
p
r
o
ce
s
s
in
g
r
ea
l
-
t
im
e
v
id
eo
s
tr
ea
m
s
to
d
etec
t
an
d
co
u
n
t
c
o
in
s
.
User
s
lau
n
ch
th
e
ap
p
licatio
n
an
d
ac
tiv
ate
th
e
ca
m
er
a
f
ee
d
.
T
h
e
co
r
e
is
a
d
etec
tio
n
lo
o
p
:
th
e
ca
m
e
r
a
ca
p
tu
r
es
co
n
tin
u
o
u
s
f
r
am
es
p
a
s
s
ed
to
o
u
r
tu
n
ed
YOL
Ov
8
m
o
d
el
in
in
f
er
e
n
ce
m
o
d
e.
T
h
e
m
o
d
el
p
r
o
ce
s
s
es
ea
ch
f
r
am
e
in
a
s
in
g
le
f
o
r
war
d
p
ass
,
g
en
er
atin
g
p
r
e
d
ictio
n
s
f
o
r
d
etec
tio
n
,
class
if
icatio
n
,
an
d
d
en
o
m
in
at
io
n
.
Fo
r
ea
ch
d
etec
tio
n
,
th
e
m
o
d
el
o
u
tp
u
ts
b
o
u
n
d
in
g
b
o
x
co
o
r
d
in
ates,
class
lab
els,
an
d
co
n
f
id
en
ce
s
co
r
es.
T
h
e
ap
p
licatio
n
co
u
n
t
s
ea
ch
d
etec
tio
n
an
d
ac
c
u
m
u
lates to
tal
v
alu
e.
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.
42
,
No
.
3
,
J
u
n
e
20
2
6
:
8
5
6
-
8
6
4
858
T
h
e
p
r
o
ce
s
s
b
eg
i
n
s
with
co
in
s
p
o
s
itio
n
ed
with
in
th
e
ca
m
er
a
’
s
f
ield
o
f
v
iew.
T
h
e
s
y
s
tem
d
e
tects
ea
ch
co
in
b
ased
o
n
v
is
u
al
f
ea
t
u
r
es,
d
eter
m
in
es
d
en
o
m
in
atio
n
,
v
e
r
i
f
ies
all
co
in
s
ar
e
id
en
tifie
d
,
ca
lcu
lates
a
s
u
b
to
tal,
an
d
eith
er
r
ep
ea
ts
f
o
r
a
d
d
itio
n
al
co
in
s
o
r
co
m
p
u
tes th
e
g
r
an
d
to
tal
.
Fig
u
r
e
1
illu
s
tr
ates
th
e
c
o
m
p
lete
s
y
s
tem
wo
r
k
f
lo
w.
T
h
e
f
lo
wch
ar
t
b
e
g
in
s
with
ca
m
er
a
in
p
u
t
ca
p
tu
r
in
g
c
o
n
tin
u
o
u
s
f
r
am
es,
f
o
llo
wed
b
y
th
e
YOL
Ov
8
m
o
d
el
p
r
o
ce
s
s
in
g
ea
ch
f
r
am
e
t
o
p
r
ed
ict
b
o
u
n
d
in
g
b
o
x
es
an
d
class
lab
els.
Dete
cted
co
in
s
ar
e
co
u
n
ted
p
er
f
r
am
e
an
d
ac
cu
m
u
late
d
in
a
r
u
n
n
in
g
to
tal,
with
an
n
o
tated
f
r
am
es
a
n
d
to
tals
d
is
p
lay
ed
to
t
h
e
u
s
er
in
r
ea
l
-
ti
m
e.
T
h
e
lo
o
p
co
n
tin
u
es
u
n
til
th
e
u
s
er
s
to
p
s
th
e
d
etec
tio
n
p
r
o
ce
s
s
,
a
t w
h
ich
p
o
in
t f
in
al
r
ep
o
r
ts
ca
n
b
e
g
en
er
at
ed
.
Fig
u
r
e
1
.
Sy
s
tem
w
o
r
k
f
lo
w
2
.
2
.
Da
t
a
s
et
p
re
pa
ra
t
io
n
A
s
ig
n
if
ican
t
co
n
tr
ib
u
ti
o
n
is
d
ev
elo
p
in
g
th
e
f
i
r
s
t
an
n
o
ta
ted
d
ataset
o
f
2
0
1
6
I
n
d
o
n
esian
co
in
s
.
T
h
e
f
in
al
d
ataset
co
n
s
is
ts
o
f
1
,
5
0
0
h
i
g
h
-
r
eso
lu
ti
o
n
im
a
g
es
r
ep
r
esen
tin
g
f
o
u
r
d
en
o
m
in
atio
n
s
:
I
DR
1
0
0
,
2
0
0
,
5
0
0
,
an
d
1
,
0
0
0
(
ap
p
r
o
x
im
ate
ly
3
7
5
im
ag
es
p
er
class
)
.
T
h
e
d
ataset
wa
s
cu
r
ated
to
ca
p
tu
r
e
r
ep
r
esen
tativ
e
v
ar
iatio
n
s
in
lig
h
tin
g
,
o
cc
lu
s
io
n
,
an
d
co
in
wea
r
.
T
o
en
s
u
r
e
th
e
m
o
d
el
co
u
ld
ef
f
ec
ti
v
ely
g
en
er
alize
to
r
ea
l
-
wo
r
ld
s
ce
n
ar
io
s
,
d
ata
co
llec
tio
n
was
in
ten
tio
n
ally
d
iv
e
r
s
if
ied
ac
r
o
s
s
s
ev
er
al
d
im
en
s
io
n
s
:
−
L
ig
h
t
co
n
d
itio
n
s
:
im
ag
es
wer
e
ca
p
tu
r
e
d
in
a
r
an
g
e
o
f
lig
h
ti
n
g
c
o
n
d
itio
n
s
,
f
r
o
m
b
r
ig
h
t
s
u
n
lig
h
t
p
r
o
d
u
cin
g
g
lar
e
to
d
a
r
k
in
d
o
o
r
am
b
ien
t li
g
h
t c
r
ea
tin
g
s
h
a
d
o
ws.
−
B
ac
k
g
r
o
u
n
d
s
an
d
s
u
r
f
ac
es:
co
i
n
s
wer
e
p
r
esen
ted
o
n
f
lat,
te
x
t
u
r
ed
,
an
d
m
u
lti
-
co
lo
r
ed
b
ac
k
g
r
o
u
n
d
s
to
teac
h
r
ec
o
g
n
itio
n
in
clu
tter
ed
en
v
ir
o
n
m
en
ts
.
−
An
g
les:
m
an
y
im
ag
es c
ap
t
u
r
e
d
co
in
s
f
r
o
m
o
v
er
h
ea
d
an
d
an
g
led
p
er
s
p
ec
tiv
es.
−
C
o
in
co
n
d
itio
n
s
:
th
e
d
ataset
in
clu
d
es
c
o
in
s
f
r
o
m
f
r
esh
l
y
m
in
ted
to
h
ea
v
il
y
cir
c
u
lated
with
v
is
ib
le
s
cr
atch
es a
n
d
tar
n
is
h
.
Fig
u
r
e
2
s
h
o
ws
a
r
ep
r
esen
tati
v
e
s
am
p
le
o
f
an
n
o
tated
co
in
s
f
r
o
m
th
e
d
ataset.
E
ac
h
co
in
is
en
clo
s
ed
with
in
a
b
o
u
n
d
in
g
b
o
x
with
its
co
r
r
esp
o
n
d
in
g
class
lab
el
(
I
DR
1
0
0
,
2
0
0
,
5
0
0
,
o
r
1
,
0
0
0
)
.
T
h
e
im
a
g
e
d
em
o
n
s
tr
ates
th
e
v
ar
iety
o
f
lig
h
tin
g
c
o
n
d
itio
n
s
,
b
ac
k
g
r
o
u
n
d
s
,
an
d
co
i
n
o
r
ien
tatio
n
s
ca
p
tu
r
e
d
in
th
e
d
ataset,
as
well
as th
e
p
r
ec
is
io
n
o
f
th
e
b
o
u
n
d
in
g
b
o
x
an
n
o
tatio
n
s
th
at
ti
g
h
tly
en
clo
s
e
ea
ch
co
in
.
Fig
u
r
e
2
.
Sam
p
le
o
f
a
n
n
o
tated
co
in
s
f
r
o
m
d
ataset
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
A
tr
a
n
s
fer lea
r
n
in
g
a
p
p
r
o
a
c
h
f
o
r
r
ea
l
-
time
d
etec
tio
n
a
n
d
cla
s
s
ifica
tio
n
…
(
N
u
r
Ha
d
is
u
kma
n
a
)
859
2
.
3
.
YO
L
O
v
8
a
nd
t
ra
ns
f
er
l
ea
rning
a
pp
ro
a
ch
YOL
Ov
8
was
s
elec
ted
f
o
r
its
b
alan
ce
o
f
f
ast
in
f
er
en
ce
an
d
h
ig
h
d
etec
tio
n
ac
cu
r
ac
y
.
T
h
e
ar
ch
itectu
r
e
im
p
lem
en
ts
s
ig
n
if
ican
t
im
p
r
o
v
em
en
ts
in
clu
d
in
g
a
n
an
ch
o
r
-
f
r
ee
d
etec
tio
n
h
ea
d
an
d
a
C
2
f
m
o
d
u
le
en
a
b
lin
g
ef
f
icien
t p
r
o
ce
s
s
in
g
[
26
].
T
o
m
in
im
ize
tr
ain
in
g
co
m
p
lex
ity
with
a
s
m
all
s
p
ec
ialized
d
ataset,
we
em
p
lo
y
ed
tr
a
n
s
f
er
le
ar
n
in
g
:
−
Mo
d
el
in
itializatio
n
:
we
in
itialized
with
y
o
lo
v
8
m
.
p
t,
p
r
e
-
tr
ain
ed
o
n
th
e
c
o
m
p
r
e
h
en
s
iv
e
MS
C
OC
O
d
ataset.
−
Fre
ez
in
g
b
ac
k
b
o
n
e
lay
e
r
s
:
we
f
r
o
ze
th
e
weig
h
ts
o
f
th
e
f
ir
s
t
ten
lay
er
s
d
etec
tin
g
lo
w
-
le
v
el
f
ea
tu
r
es
(
ed
g
es,
co
r
n
er
s
,
tex
tu
r
es)
—
f
u
n
d
am
e
n
tal
an
d
tr
an
s
f
er
a
b
le
ac
r
o
s
s
v
is
u
al
task
s
.
T
h
is
p
r
eser
v
ed
g
en
er
alize
d
k
n
o
wled
g
e
w
h
ile
r
ed
u
ci
n
g
tr
ai
n
ab
le
p
ar
am
ete
r
s
,
p
r
ev
e
n
tin
g
o
v
er
f
itti
n
g
.
−
Fin
e
-
tu
n
in
g
:
th
e
r
em
ain
in
g
lay
er
s
lear
n
in
g
task
-
s
p
ec
if
ic
f
ea
tu
r
es
wer
e
tr
ain
ed
o
n
o
u
r
cu
s
to
m
co
in
d
ataset,
ad
ap
tin
g
to
v
is
u
al
ch
ar
ac
ter
is
tics
lik
e
en
g
r
av
in
g
s
,
n
u
m
b
er
s
,
a
n
d
r
ef
lectiv
e
s
u
r
f
ac
es.
Usi
n
g
th
is
tr
an
s
f
er
lear
n
in
g
a
p
p
r
o
ac
h
,
we
ac
h
iev
ed
s
ig
n
i
f
ican
t
tr
ain
in
g
tim
e
s
av
in
g
s
an
d
m
ain
tain
ed
h
ig
h
ac
cu
r
ac
y
u
n
attain
ab
le
if
t
r
ain
ed
f
r
o
m
s
cr
atch
.
2
.
4
.
T
ra
ini
ng
env
iro
nm
ent
a
nd
pa
ra
m
et
er
s
T
r
ain
in
g
u
s
ed
an
NVI
DI
A
A1
0
0
GPU
v
ia
Go
o
g
le
C
o
lab
Pr
o
.
A
Py
T
o
r
c
h
p
ip
elin
e
with
U
ltra
ly
tics
’
YOL
Ov
8
im
p
lem
en
tatio
n
was
b
u
ilt.
Key
p
a
r
am
eter
s
:
−
E
p
o
ch
s
:
u
p
t
o
1
0
0
with
ea
r
ly
s
to
p
p
in
g
(
p
atien
ce
=1
0
)
m
o
n
ito
r
in
g
v
alid
atio
n
lo
s
s
.
−
B
atch
s
ize
: 1
6
b
alan
cin
g
m
e
m
o
r
y
ef
f
icien
cy
an
d
s
tab
le
co
n
v
er
g
en
ce
.
−
I
m
ag
e
r
eso
lu
tio
n
: 6
4
0
×6
4
0
p
i
x
els m
atch
in
g
YOL
Ov
8
in
p
u
t
d
im
en
s
io
n
s
.
−
Op
tim
izer
: A
d
am
W
s
elec
ted
f
o
r
ef
f
ec
tiv
e
weig
h
t d
ec
ay
p
r
o
p
er
ties
.
−
L
ea
r
n
in
g
r
ate
: 0
.
0
0
1
,
wid
ely
a
cc
ep
ted
f
o
r
f
in
e
-
t
u
n
in
g
task
s
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
E
v
a
lua
t
i
o
n
m
et
ric
Mo
d
el
p
er
f
o
r
m
an
ce
was
ass
ess
ed
u
s
in
g
wid
ely
ad
o
p
ted
o
b
ject
d
etec
tio
n
m
etr
i
cs:
p
r
ec
is
io
n
,
r
ec
all,
an
d
m
ea
n
av
er
ag
e
p
r
ec
is
io
n
(
m
AP)
.
−
Pre
cisi
o
n
=
T
P/(T
P+FP
)
: p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
t p
o
s
itiv
e
p
r
ed
i
ctio
n
s
.
−
R
ec
all
=
T
P/(T
P+FN)
: a
b
ili
ty
to
d
etec
t a
ll r
elev
an
t in
s
tan
ce
s
.
−
m
AP@
0
.
5
:
m
AP
at
I
o
U
th
r
esh
o
ld
0
.
5
.
−
m
AP@
0
.
5
:0
.
9
5
:
m
AP
ac
r
o
s
s
I
o
U
th
r
esh
o
ld
s
0
.
5
to
0
.
9
5
.
3
.
2
.
Q
ua
ntit
a
t
iv
e
r
esu
lt
s
T
h
e
m
o
d
el
was
ev
al
u
ated
o
n
th
e
u
n
s
ee
n
test
s
et
(
1
5
0
im
a
g
es,
ap
p
r
o
x
im
ately
3
0
0
-
co
i
n
in
s
tan
ce
s
)
.
E
ar
ly
s
to
p
p
in
g
h
alted
tr
ain
in
g
at
ep
o
c
h
3
9
;
weig
h
ts
f
r
o
m
e
p
o
ch
2
9
(
lo
west
v
alid
atio
n
l
o
s
s
)
wer
e
r
etain
ed
as
th
e
f
in
al
m
o
d
el.
Fig
u
r
e
3
p
r
esen
ts
t
h
e
f
in
al
p
e
r
f
o
r
m
an
ce
m
etr
ics.
Ov
er
all
m
AP@
0
.
5
o
f
0
.
9
9
5
a
n
d
m
AP@
0
.
5
:0
.
9
5
o
f
0
.
9
8
d
em
o
n
s
tr
ate
n
ea
r
-
pe
r
f
ec
t
class
if
icatio
n
an
d
h
ig
h
ly
a
cc
u
r
ate
lo
ca
lizatio
n
.
Per
-
class
an
aly
s
is
s
h
o
wed
co
n
s
is
ten
t
p
er
f
o
r
m
an
ce
ac
r
o
s
s
all
d
en
o
m
in
atio
n
s
,
with
m
A
P@
0
.
5
ex
ce
ed
in
g
0
.
9
9
f
o
r
ea
ch
class
.
T
h
e
m
o
s
t
s
im
ilar
d
en
o
m
in
atio
n
s
(
I
DR
1
0
0
an
d
2
0
0
)
ac
h
iev
ed
m
AP@
0
.
5
:0
.
9
5
s
co
r
es
o
f
0
.
9
7
a
n
d
0
.
9
8
r
esp
ec
tiv
ely
,
co
n
f
ir
m
in
g
th
e
m
o
d
el
s
u
cc
es
s
f
u
lly
lear
n
ed
d
is
tin
g
u
is
h
in
g
ch
ar
ac
ter
is
tics
d
esp
ite
s
im
ilar
ap
p
ea
r
a
n
ce
s
.
T
h
e
r
elativ
ely
h
ig
h
p
er
f
o
r
m
an
ce
c
an
b
e
attr
ib
u
te
d
to
th
e
co
n
s
tr
a
in
ed
p
r
o
b
lem
s
co
p
e
in
v
o
l
v
in
g
o
n
ly
f
o
u
r
v
is
u
ally
d
is
tin
ct
class
e
s
u
n
d
er
s
tatic
s
in
g
le
-
co
in
s
ce
n
ar
io
s
.
Dir
ec
t
co
m
p
ar
is
o
n
v
alid
ates
o
u
r
tr
an
s
f
er
lear
n
in
g
ap
p
r
o
ac
h
.
Pra
b
u
et
al.
[
23
]
ac
h
iev
e
d
m
A
P@
0
.
5
o
f
0
.
9
8
o
n
I
n
d
ian
co
in
s
u
s
in
g
Y
OL
Ov
5
tr
ain
ed
f
r
o
m
s
cr
atch
o
n
2
,
0
0
0
im
a
g
es.
Ou
r
m
eth
o
d
ac
h
iev
es
0
.
9
9
5
with
2
5
%
less
d
ata
(
1
,
5
0
0
im
ag
es)
an
d
r
e
d
u
ce
d
t
r
ain
in
g
tim
e
d
u
e
to
f
r
o
ze
n
b
ac
k
b
o
n
e
tr
an
s
f
er
lear
n
in
g
.
R
o
s
alin
a
[
24
]
ac
h
iev
ed
0
.
9
4
o
n
a
n
cien
t
co
in
s
with
f
u
ll f
in
e
-
tu
n
in
g
o
n
1
,
2
0
0
im
ag
es,
r
ef
lectin
g
th
e
a
d
d
itio
n
al
ch
allen
g
es
o
f
ir
r
eg
u
lar
s
h
a
p
es
an
d
ex
tr
e
m
e
d
eg
r
ad
atio
n
.
Ma
d
e
et
al
.
[
25
]
r
ep
o
r
ted
0
.
9
9
o
n
b
an
k
n
o
te
s
u
s
in
g
YOL
Ov
8
o
n
1
,
8
0
0
im
ag
es,
b
u
t
b
a
n
k
n
o
tes
b
en
ef
it
f
r
o
m
d
is
tin
ct
co
lo
r
p
atter
n
s
wh
ile
o
u
r
m
o
d
el
h
a
n
d
les
m
o
r
e
d
if
f
icu
lt
m
etallic
co
in
s
with
s
u
b
tle
r
eli
ef
f
ea
tu
r
es.
T
h
ese
co
m
p
ar
is
o
n
s
co
n
f
ir
m
th
at
f
r
o
ze
n
-
b
ac
k
b
o
n
e
tr
an
s
f
er
le
ar
n
in
g
r
ed
u
ce
s
d
ata
r
e
q
u
ir
em
e
n
ts
wh
ile
m
ain
tain
in
g
co
m
p
etitiv
e
ac
cu
r
ac
y
.
3
.
3
.
Q
ua
lit
a
t
iv
e
a
na
ly
s
is
W
e
ev
alu
ated
p
r
ac
tical
d
etec
ti
o
n
ca
p
ab
ilit
ies
th
r
o
u
g
h
q
u
alitativ
e
test
s
in
v
o
lv
in
g
d
if
f
e
r
en
t
r
ea
l
-
wo
r
ld
f
ac
to
r
s
.
Fig
u
r
e
4
s
h
o
ws
d
ete
ctio
n
r
esu
lts
u
n
d
e
r
id
ea
l
c
o
n
d
itio
n
s
—
d
if
f
u
s
e
lig
h
tin
g
,
m
i
n
im
al
b
ac
k
g
r
o
u
n
d
clu
tter
,
an
d
clea
r
ly
s
ep
ar
ated
co
in
s
.
T
h
e
m
o
d
el
p
e
r
f
o
r
m
ed
o
p
tim
ally
with
co
n
f
id
e
n
ce
s
co
r
es
ex
ce
ed
in
g
0
.
9
5
,
estab
lis
h
in
g
a
s
tr
o
n
g
p
e
r
f
o
r
m
a
n
ce
b
aselin
e.
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.
42
,
No
.
3
,
J
u
n
e
20
2
6
:
8
5
6
-
8
6
4
860
Fig
u
r
e
3
.
Fin
al
m
o
d
el
p
e
r
f
o
r
m
an
ce
m
etr
ic
Fig
u
r
e
4
.
Dete
ctio
n
o
b
jects in
an
id
ea
l c
o
n
d
itio
n
3
.
4
.
Rea
l
-
wo
rld c
o
nd
it
io
ns
:
3
.
4
.
1
.
G
la
re
a
nd
direct
lig
ht
Stro
n
g
s
p
ec
u
la
r
r
ef
lectio
n
s
ca
u
s
ed
b
y
d
ir
ec
t
lig
h
tin
g
o
cc
asio
n
ally
o
b
s
cu
r
ed
s
u
r
f
ac
e
d
etails,
lead
in
g
to
s
p
o
r
ad
ic
m
is
class
if
icatio
n
s
(
Fig
u
r
e
5
)
.
T
h
e
m
o
s
t
co
m
m
o
n
co
n
f
u
s
io
n
o
b
s
er
v
ed
was
b
etwe
en
I
DR
5
0
0
an
d
I
DR
1
0
0
0
co
in
s
,
as
b
o
th
h
av
e
s
im
ilar
d
iam
eter
s
an
d
th
e
r
ef
lectiv
e
h
ig
h
lig
h
ts
ten
d
ed
to
a
p
p
ea
r
i
n
th
e
ce
n
tr
al
r
eg
io
n
wh
er
e
d
en
o
m
in
atio
n
n
u
m
er
als
ar
e
lo
ca
ted
.
W
h
en
g
l
ar
e
co
v
er
ed
m
o
r
e
th
a
n
a
p
p
r
o
x
im
ately
3
0
%
o
f
th
e
co
in
’
s
s
u
r
f
ac
e
a
r
ea
,
class
if
icatio
n
ac
cu
r
ac
y
d
r
o
p
p
ed
n
o
tice
ab
ly
wh
ile
d
etec
tio
n
co
n
f
id
en
ce
r
em
ain
ed
ab
o
v
e
0
.
9
0
.
T
h
is
in
d
icate
s
th
at
th
e
m
o
d
el
r
elies
m
o
r
e
h
ea
v
ily
o
n
f
i
n
e
s
u
r
f
ac
e
d
etails
f
o
r
class
if
icatio
n
th
an
f
o
r
i
n
itial
d
etec
tio
n
.
T
h
e
m
o
d
el
g
en
er
ally
r
etain
ed
d
etec
tio
n
ca
p
a
b
ilit
y
d
esp
ite
r
ed
u
ce
d
class
if
icatio
n
r
eliab
ilit
y
,
s
u
g
g
esti
n
g
th
at
lo
w
-
lev
el
s
h
ap
e
f
ea
tu
r
es
ar
e
m
o
r
e
r
o
b
u
s
t
to
lig
h
tin
g
v
ar
iatio
n
s
th
an
th
e
h
ig
h
-
f
r
eq
u
e
n
cy
tex
tu
r
al
d
etails r
eq
u
ir
e
d
f
o
r
f
in
e
-
g
r
ain
ed
d
en
o
m
in
atio
n
d
is
cr
im
in
atio
n
.
Fig
u
r
e
5
.
Misclass
if
icatio
n
d
u
e
to
g
lar
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:
2502
-
4
7
5
2
A
tr
a
n
s
fer lea
r
n
in
g
a
p
p
r
o
a
c
h
f
o
r
r
ea
l
-
time
d
etec
tio
n
a
n
d
cla
s
s
ifica
tio
n
…
(
N
u
r
Ha
d
is
u
kma
n
a
)
861
3
.
4
.
2
.
P
o
o
r
lig
ht
env
iro
nm
en
t
I
n
lo
w
-
lig
h
t
e
n
v
ir
o
n
m
en
ts
wit
h
illu
m
in
atio
n
lev
els
b
elo
w
5
0
lu
x
,
r
e
d
u
ce
d
v
is
u
al
co
n
tr
ast
an
d
lo
s
s
o
f
f
in
e
-
g
r
ain
e
d
f
ea
tu
r
es
s
ig
n
if
ica
n
tly
af
f
ec
te
d
p
e
r
f
o
r
m
an
ce
.
I
m
ag
e
an
aly
s
is
s
h
o
wed
co
m
p
r
ess
ed
d
y
n
a
m
ic
r
a
n
g
es
with
m
o
s
t
p
ix
el
v
alu
es
co
n
ce
n
tr
ated
in
th
e
lo
wer
i
n
ten
s
ity
s
p
ec
tr
u
m
.
W
h
ile
d
etec
tio
n
r
e
m
ain
ed
p
o
s
s
ib
le
in
s
o
m
e
ca
s
es,
m
is
class
if
icatio
n
s
in
cr
ea
s
ed
n
o
ticea
b
ly
co
m
p
ar
ed
to
id
ea
l
lig
h
tin
g
c
o
n
d
itio
n
s
co
m
p
a
r
ed
t
o
id
ea
l
lig
h
tin
g
co
n
d
itio
n
s
.
C
o
n
f
id
en
ce
s
co
r
es
f
ell
to
0
.
7
0
-
0
.
8
5
e
v
e
n
f
o
r
co
r
r
ec
t
class
if
icatio
n
s
(
Fig
u
r
e
6
)
,
in
d
icatin
g
r
ed
u
ce
d
m
o
d
el
ce
r
tain
ty
.
T
h
e
p
r
im
ar
y
ca
u
s
e
was
in
s
u
f
f
icien
t
illu
m
in
atio
n
f
ailin
g
to
r
ev
ea
l
cr
itical
e
n
g
r
av
in
g
s
an
d
d
e
n
o
m
in
atio
n
m
a
r
k
er
s
,
p
ar
ticu
lar
ly
f
o
r
th
e
I
DR
1
0
0
a
n
d
I
DR
2
0
0
c
o
in
s
wh
ich
r
el
y
o
n
s
u
b
tle
s
u
r
f
ac
e
r
elief
f
o
r
d
is
cr
im
in
atio
n
.
Fig
u
r
e
6
.
Misclass
if
icatio
n
d
u
e
to
p
o
o
r
lig
h
tin
g
3
.
4
.
3
.
O
v
er
la
pp
ing
a
nd
clut
t
er
ed
co
ins
Hea
v
ily
o
v
e
r
lap
p
in
g
o
r
clu
s
ter
ed
co
in
s
p
o
s
ed
s
ig
n
if
ican
t
ch
a
llen
g
es
f
o
r
in
s
tan
ce
s
ep
ar
atio
n
,
r
esu
ltin
g
in
m
is
s
ed
d
etec
tio
n
s
o
r
in
co
r
r
ec
t
class
if
icatio
n
s
wh
en
o
v
er
lap
ex
ce
ed
ed
a
p
p
r
o
x
im
atel
y
3
0
%
(
Fig
u
r
e
7
)
.
Qu
an
titativ
e
an
aly
s
is
r
ev
ea
led
th
at
r
ec
all
d
r
o
p
p
e
d
f
r
o
m
0
.
9
9
to
0
.
8
2
wh
e
n
o
v
er
lap
ex
ce
ed
e
d
th
is
th
r
esh
o
ld
.
I
n
ca
s
es
wh
er
e
th
r
ee
o
r
m
o
r
e
co
i
n
s
wer
e
clu
s
ter
ed
,
th
e
m
o
d
el
f
r
eq
u
en
tly
d
etec
te
d
o
n
ly
two
o
f
th
e
th
r
ee
,
m
er
g
in
g
th
e
th
ir
d
with
an
a
d
jace
n
t
d
e
tectio
n
.
T
h
is
lim
itatio
n
s
tem
s
f
r
o
m
th
e
n
o
n
-
m
ax
im
u
m
s
u
p
p
r
ess
io
n
alg
o
r
ith
m
u
s
ed
in
YOL
Ov
8
,
wh
ic
h
elim
in
ates
r
ed
u
n
d
an
t
b
o
u
n
d
in
g
b
o
x
es
b
ased
o
n
o
v
er
lap
th
r
esh
o
ld
s
.
W
h
en
co
i
n
s
ar
e
p
h
y
s
ically
s
tack
ed
,
th
e
o
v
er
la
p
cr
ea
tes
g
en
u
in
e
am
b
ig
u
ity
ab
o
u
t
o
b
ject
b
o
u
n
d
ar
ies
th
at
ca
n
n
o
t
b
e
r
eso
lv
ed
with
o
u
t a
d
d
itio
n
al
c
o
n
tex
tu
al
i
n
f
o
r
m
atio
n
o
r
m
u
lti
-
v
iew
an
al
y
s
is
.
Fig
u
r
e
7
.
Miss
ed
o
b
ject
d
etec
t
io
n
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.
42
,
No
.
3
,
J
u
n
e
20
2
6
:
8
5
6
-
8
6
4
862
3
.
4
.
4
.
Co
in
wea
r
a
nd
deg
ra
d
a
t
io
n
Hea
v
ily
tar
n
is
h
ed
o
r
d
am
a
g
ed
co
in
s
e
x
h
ib
ited
d
eg
r
ad
ed
s
u
r
f
ac
e
f
ea
tu
r
es,
a
d
v
er
s
ely
af
f
ec
tin
g
class
if
icatio
n
ac
cu
r
ac
y
.
An
a
ly
s
is
o
f
wo
r
n
co
in
s
s
h
o
we
d
th
at
wh
en
s
u
r
f
ac
e
r
elief
d
ep
th
d
ec
r
ea
s
ed
s
ig
n
if
ican
tly
,
class
if
icatio
n
ac
cu
r
ac
y
d
r
o
p
p
e
d
f
r
o
m
0
.
9
9
to
ap
p
r
o
x
im
ately
0
.
7
6
.
C
o
in
s
with
ex
ten
s
iv
e
wea
r
wer
e
m
o
s
t
co
m
m
o
n
l
y
m
is
class
if
ied
as
th
e
n
ex
t
s
m
aller
d
en
o
m
in
atio
n
—
f
o
r
ex
am
p
le,
w
o
r
n
I
DR
5
0
0
c
o
in
s
wer
e
o
f
ten
class
if
ied
as
I
D
R
2
0
0
(
Fig
u
r
e
8
)
.
T
h
is
co
n
f
u
s
io
n
p
atter
n
s
u
g
g
ests
th
at
wh
en
n
u
m
er
ic
d
ig
its
ar
e
d
eg
r
ad
e
d
,
th
e
m
o
d
el
d
ef
au
lts
to
s
ize
-
b
ased
class
if
icatio
n
,
wh
ich
ca
n
b
e
am
b
ig
u
o
u
s
g
iv
en
th
e
s
m
all
s
ize
d
if
f
er
en
ce
s
b
etwe
en
d
en
o
m
in
atio
n
s
.
T
h
is
lim
itatio
n
is
p
ar
ticu
la
r
ly
r
elev
an
t
f
o
r
r
ea
l
-
wo
r
ld
d
ep
lo
y
m
en
t,
as
cir
cu
lated
co
in
s
in
e
v
itab
ly
s
h
o
w
v
ar
y
in
g
d
eg
r
ee
s
o
f
wea
r
,
an
d
an
y
p
r
ac
tical
s
y
s
tem
m
u
s
t
m
ain
tain
r
ea
s
o
n
ab
l
e
ac
cu
r
ac
y
ac
r
o
s
s
th
e
f
u
ll sp
ec
tr
u
m
o
f
c
o
in
co
n
d
itio
n
s
.
Ov
er
all,
ex
p
e
r
im
en
tal
r
esu
lts
co
n
f
ir
m
th
e
s
y
s
tem
d
eliv
er
s
h
ig
h
ac
cu
r
ac
y
u
n
d
er
ty
p
ical
o
p
er
atin
g
co
n
d
itio
n
s
b
u
t sh
o
ws lim
itatio
n
s
u
n
d
er
ex
tr
em
e
lig
h
tin
g
,
s
ig
n
if
ican
t o
cc
lu
s
io
n
,
an
d
s
ev
er
e
co
in
wea
r
,
d
ef
in
in
g
o
p
er
atio
n
al
b
o
u
n
d
ar
ies an
d
f
u
t
u
r
e
im
p
r
o
v
em
e
n
t a
r
ea
s
.
Fig
u
r
e
8
.
Misclass
if
icatio
n
o
f
tar
n
is
h
ed
o
b
jects
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
p
r
o
p
o
s
es
a
r
ea
l
-
ti
m
e
s
y
s
tem
f
o
r
d
etec
tin
g
an
d
c
lass
if
y
in
g
2
0
1
6
I
n
d
o
n
esian
co
in
s
u
s
in
g
a
tr
an
s
f
er
lear
n
in
g
-
b
ased
YOL
Ov
8
f
r
am
ewo
r
k
.
B
y
f
r
ee
zin
g
b
ac
k
b
o
n
e
lay
e
r
s
an
d
f
i
n
e
-
tu
n
in
g
h
ig
h
e
r
-
lev
el
fe
atu
r
es,
th
e
m
o
d
el
ac
h
iev
es
h
ig
h
d
etec
tio
n
ac
c
u
r
ac
y
(
m
AP@
0
.
5
=
0
.
9
9
5
)
o
n
an
u
n
s
ee
n
test
s
et
d
esp
ite
lim
ited
d
ataset
s
ize.
R
esu
lt
s
d
em
o
n
s
tr
ate
tr
an
s
f
er
lear
n
in
g
is
an
ef
f
ec
tiv
e
s
tr
ateg
y
f
o
r
ad
ap
tin
g
g
en
er
al
-
p
u
r
p
o
s
e
o
b
ject
d
etec
tio
n
m
o
d
els
to
s
p
ec
ialized
,
lo
w
-
r
eso
u
r
ce
d
o
m
ain
s
.
Qu
alitativ
e
ev
alu
atio
n
s
r
ev
ea
l
p
er
f
o
r
m
an
ce
d
eg
r
ad
atio
n
u
n
d
er
ex
tr
em
e
lig
h
tin
g
,
s
ev
er
e
o
cc
lu
s
io
n
,
an
d
h
ea
v
y
c
o
in
wea
r
,
p
r
o
v
id
i
n
g
tr
an
s
p
ar
en
t a
s
s
ess
m
en
t o
f
th
e
s
y
s
tem
’
s
p
r
ac
tical
lim
itatio
n
s
.
Scien
tific
ally
,
th
is
wo
r
k
v
alid
ates
f
r
o
ze
n
-
b
ac
k
b
o
n
e
tr
a
n
s
f
er
lear
n
in
g
as
a
c
o
m
p
u
tatio
n
ally
ef
f
icien
t
ap
p
r
o
ac
h
f
o
r
d
o
m
ain
-
s
p
ec
if
ic
cu
r
r
en
c
y
r
ec
o
g
n
itio
n
an
d
co
n
t
r
ib
u
tes
a
n
ewly
co
n
s
tr
u
cted
a
n
n
o
tated
d
ataset
f
o
r
m
o
d
er
n
I
n
d
o
n
esian
co
in
s
,
a
d
d
r
ess
in
g
a
g
ap
in
e
x
is
tin
g
r
esear
ch
.
Fro
m
an
a
p
p
licatio
n
p
er
s
p
ec
tiv
e,
th
e
p
r
o
p
o
s
ed
s
y
s
tem
is
well
-
s
u
ite
d
f
o
r
b
an
k
i
n
g
au
to
m
atio
n
,
r
et
ail
s
y
s
tem
s
,
an
d
ass
is
tiv
e
tech
n
o
lo
g
ies
r
eq
u
i
r
in
g
r
ea
l
-
tim
e
co
in
r
ec
o
g
n
itio
n
.
Fu
t
u
r
e
wo
r
k
will f
o
cu
s
o
n
ex
p
a
n
d
in
g
d
ataset
d
iv
er
s
ity
to
in
clu
d
e
m
o
r
e
ch
allen
g
i
n
g
s
ce
n
ar
io
s
s
u
ch
as
ex
tr
em
e
lig
h
t
in
g
v
ar
iatio
n
s
,
s
ev
er
e
o
cc
lu
s
io
n
,
an
d
h
ea
v
ily
wo
r
n
co
i
n
s
,
an
d
d
ev
elo
p
in
g
a
lig
h
tweig
h
t m
o
b
ile
v
er
s
io
n
f
o
r
b
r
o
ad
e
r
ac
ce
s
s
ib
ilit
y
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
e
au
th
o
r
s
d
ec
lar
e
th
at
n
o
f
u
n
d
in
g
was r
ec
eiv
e
d
f
o
r
co
n
d
u
c
tin
g
th
e
r
esear
ch
.
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
A
tr
a
n
s
fer lea
r
n
in
g
a
p
p
r
o
a
c
h
f
o
r
r
ea
l
-
time
d
etec
tio
n
a
n
d
cla
s
s
ifica
tio
n
…
(
N
u
r
Ha
d
is
u
kma
n
a
)
863
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Nu
r
Had
is
u
k
m
an
a
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
R
.
B
.
W
ah
y
u
✓
✓
✓
✓
✓
✓
Stewar
t Q
iu
✓
✓
✓
✓
✓
✓
✓
✓
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
T
h
e
au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
ter
est r
elate
d
to
th
is
r
esear
ch
.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
aila
b
le
o
n
r
eq
u
est
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
,
[
in
itials
,
NH
]
.
T
h
e
d
a
ta,
wh
ich
co
n
tain
in
f
o
r
m
atio
n
th
at
co
u
ld
c
o
m
p
r
o
m
is
e
th
e
p
r
iv
ac
y
o
f
r
esear
c
h
p
ar
ticip
an
ts
,
ar
e
n
o
t p
u
b
licly
a
v
ailab
le
d
u
e
to
ce
r
tain
r
estrictio
n
s
.
RE
F
E
R
E
NC
E
S
[
1
]
D
.
T.
A
seff
a
,
H
.
K
a
l
l
a
,
a
n
d
S
.
M
i
s
h
r
a
,
“
E
t
h
i
o
p
i
a
n
b
a
n
k
n
o
t
e
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
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
i
t
s
p
r
o
t
o
t
y
p
e
d
e
v
e
l
o
p
me
n
t
u
si
n
g
e
m
b
e
d
d
e
d
p
l
a
t
f
o
r
m,”
J
o
u
r
n
a
l
o
f
S
e
n
s
o
rs
,
v
o
l
.
2
0
2
2
,
p
p
.
1
–
1
8
,
Ja
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
2
/
4
5
0
5
0
8
9
.
[
2
]
J.
W
.
L
e
e
,
H
.
G
.
H
o
n
g
,
K
.
W
.
K
i
m
,
a
n
d
K
.
R
.
P
a
r
k
,
“
A
s
u
r
v
e
y
o
n
b
a
n
k
n
o
t
e
r
e
c
o
g
n
i
t
i
o
n
m
e
t
h
o
d
s
b
y
v
a
r
i
o
u
s
s
e
n
s
o
r
s,”
S
e
n
s
o
r
s
(
S
w
i
t
zer
l
a
n
d
)
,
v
o
l
.
1
7
,
n
o
.
2
,
p
.
3
1
3
,
F
e
b
.
2
0
1
7
,
d
o
i
:
1
0
.
3
3
9
0
/
s1
7
0
2
0
3
1
3
.
[
3
]
S
.
G
u
p
t
a
,
P
.
A
r
b
e
l
á
e
z
,
R
.
G
i
r
s
h
i
c
k
,
a
n
d
J.
M
a
l
i
k
,
“
I
n
d
o
o
r
sce
n
e
u
n
d
e
r
st
a
n
d
i
n
g
w
i
t
h
R
G
B
-
D
i
ma
g
e
s
:
b
o
t
t
o
m
-
u
p
s
e
g
me
n
t
a
t
i
o
n
,
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
sem
a
n
t
i
c
se
g
me
n
t
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
Vi
s
i
o
n
,
v
o
l
.
1
1
2
,
n
o
.
2
,
p
p
.
1
3
3
–
1
4
9
,
A
p
r
.
2
0
1
5
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
2
6
3
-
014
-
0
7
7
7
-
6.
[
4
]
A
.
K
r
i
z
h
e
v
s
k
y
,
I
.
S
u
t
s
k
e
v
e
r
,
a
n
d
G
.
E.
H
i
n
t
o
n
,
“
I
mag
e
N
e
t
c
l
a
ssi
f
i
c
a
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
s
,
”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
o
f
t
h
e
A
C
M
,
v
o
l
.
6
0
,
n
o
.
6
,
p
p
.
8
4
–
9
0
,
M
a
y
2
0
1
7
,
d
o
i
:
1
0
.
1
1
4
5
/
3
0
6
5
3
8
6
.
[
5
]
A
.
V
o
u
l
o
d
i
mo
s,
N
.
D
o
u
l
a
m
i
s,
A
.
D
o
u
l
a
m
i
s,
a
n
d
E.
P
r
o
t
o
p
a
p
a
d
a
k
i
s
,
“
D
e
e
p
l
e
a
r
n
i
n
g
f
o
r
c
o
m
p
u
t
e
r
v
i
si
o
n
:
a
b
r
i
e
f
r
e
v
i
e
w
,
”
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
N
e
u
r
o
sc
i
e
n
c
e
,
v
o
l
.
2
0
1
8
,
p
p
.
1
–
1
3
,
2
0
1
8
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
1
8
/
7
0
6
8
3
4
9
.
[
6
]
J.
G
u
e
t
a
l
.
,
“
R
e
c
e
n
t
a
d
v
a
n
c
e
s
i
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
,
”
Pa
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
v
o
l
.
7
7
,
p
p
.
3
5
4
–
3
7
7
,
M
a
y
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
c
o
g
.
2
0
1
7
.
1
0
.
0
1
3
.
[
7
]
T.
D
i
w
a
n
,
G
.
A
n
i
r
u
d
h
,
a
n
d
J
.
V
.
T
e
mb
h
u
r
n
e
,
“
O
b
j
e
c
t
d
e
t
e
c
t
i
o
n
u
si
n
g
Y
O
LO
:
c
h
a
l
l
e
n
g
e
s
,
a
r
c
h
i
t
e
c
t
u
r
a
l
s
u
c
c
e
sso
r
s,
d
a
t
a
s
e
t
s
a
n
d
a
p
p
l
i
c
a
t
i
o
n
s,
”
M
u
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
2
,
n
o
.
6
,
p
p
.
9
2
4
3
–
9
2
7
5
,
M
a
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
0
4
2
-
0
2
2
-
1
3
6
4
4
-
y.
[
8
]
J.
R
e
d
m
o
n
,
S
.
D
i
v
v
a
l
a
,
R
.
G
i
r
s
h
i
c
k
,
a
n
d
A
.
F
a
r
h
a
d
i
,
“
Y
o
u
o
n
l
y
l
o
o
k
o
n
c
e
:
u
n
i
f
i
e
d
,
r
e
a
l
-
t
i
m
e
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
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
R
e
c
o
g
n
i
t
i
o
n
,
I
EEE,
Ju
n
.
2
0
1
6
,
p
p
.
7
7
9
–
7
8
8
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
9
1
.
[
9
]
J.
T
e
r
v
e
n
,
D
.
M
.
C
ó
r
d
o
v
a
-
Esp
a
r
z
a
,
a
n
d
J.
A
.
R
o
mer
o
-
G
o
n
z
á
l
e
z
,
“
A
c
o
mp
r
e
h
e
n
si
v
e
r
e
v
i
e
w
o
f
Y
O
LO
a
r
c
h
i
t
e
c
t
u
r
e
s
i
n
c
o
m
p
u
t
e
r
v
i
s
i
o
n
:
f
r
o
m
Y
O
LO
v
1
t
o
Y
O
LO
v
8
a
n
d
Y
O
LO
-
N
A
S
,
”
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
K
n
o
w
l
e
d
g
e
Ex
t
r
a
c
t
i
o
n
,
v
o
l
.
5
,
n
o
.
4
,
p
p
.
1
6
8
0
–
1
7
1
6
,
N
o
v
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
ma
k
e
5
0
4
0
0
8
3
.
[
1
0
]
C
.
Y
.
W
a
n
g
,
A
.
B
o
c
h
k
o
v
sk
i
y
,
a
n
d
H
.
Y
.
M
.
Li
a
o
,
“
Y
O
LO
v
7
:
t
r
a
i
n
a
b
l
e
b
a
g
-
of
-
f
r
e
e
b
i
e
s
se
t
s
n
e
w
st
a
t
e
-
of
-
t
h
e
-
a
r
t
f
o
r
r
e
a
l
-
t
i
me
o
b
j
e
c
t
d
e
t
e
c
t
o
r
s,”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
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
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
,
I
EEE
,
Ju
n
.
2
0
2
3
,
p
p
.
7
4
6
4
–
7
4
7
5
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
5
2
7
2
9
.
2
0
2
3
.
0
0
7
2
1
.
[
1
1
]
F
.
Zh
u
a
n
g
e
t
a
l
.
,
“
A
C
o
m
p
r
e
h
e
n
s
i
v
e
S
u
r
v
e
y
o
n
Tr
a
n
sf
e
r
L
e
a
r
n
i
n
g
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
,
v
o
l
.
1
0
9
,
n
o
.
1
,
p
p
.
4
3
–
7
6
,
Ja
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
JP
R
O
C
.
2
0
2
0
.
3
0
0
4
5
5
5
.
[
1
2
]
M
.
Tr
i
g
k
a
a
n
d
E.
D
r
i
t
s
a
s,
“
A
C
o
m
p
r
e
h
e
n
si
v
e
S
u
r
v
e
y
o
f
M
a
c
h
i
n
e
Le
a
r
n
i
n
g
Te
c
h
n
i
q
u
e
s
a
n
d
M
o
d
e
l
s
f
o
r
O
b
j
e
c
t
D
e
t
e
c
t
i
o
n
,
”
S
e
n
so
rs
,
v
o
l
.
2
5
,
n
o
.
1
,
p
.
2
1
4
,
Ja
n
.
2
0
2
5
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
5
0
1
0
2
1
4
.
[
1
3
]
N
.
D
a
l
a
l
a
n
d
B
.
Tr
i
g
g
s
,
“
H
i
st
o
g
r
a
ms
o
f
o
r
i
e
n
t
e
d
g
r
a
d
i
e
n
t
s
f
o
r
h
u
m
a
n
d
e
t
e
c
t
i
o
n
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
-
2
0
0
5
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
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
Pa
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
C
VPR
2
0
0
5
,
I
EEE,
2
0
0
5
,
p
p
.
8
8
6
–
8
9
3
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
0
5
.
1
7
7
.
[
1
4
]
N
.
C
a
r
i
o
n
,
F
.
M
a
ssa,
G
.
S
y
n
n
a
e
v
e
,
N
.
U
su
n
i
e
r
,
A
.
K
i
r
i
l
l
o
v
,
a
n
d
S
.
Z
a
g
o
r
u
y
k
o
,
“
E
n
d
-
to
-
e
n
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
t
r
a
n
sf
o
r
mers
,
”
i
n
Eu
r
o
p
e
a
n
c
o
n
f
e
re
n
c
e
o
n
c
o
m
p
u
t
e
r
v
i
s
i
o
n
,
2
0
2
0
,
p
p
.
2
1
3
–
2
2
9
.
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
0
-
5
8
4
5
2
-
8
_
1
3
.
[
1
5
]
Z.
Z
o
u
,
K
.
C
h
e
n
,
Z.
S
h
i
,
Y
.
G
u
o
,
a
n
d
J.
Y
e
,
“
O
b
j
e
c
t
d
e
t
e
c
t
i
o
n
i
n
2
0
y
e
a
r
s
:
a
su
r
v
e
y
,
”
Pr
o
c
e
e
d
i
n
g
s o
f
t
h
e
I
E
EE
,
v
o
l
.
1
1
1
,
n
o
.
3
,
p
p
.
2
5
7
–
2
7
6
,
M
a
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
JP
R
O
C
.
2
0
2
3
.
3
2
3
8
5
2
4
.
[
1
6
]
Y
.
S
u
n
,
Z.
S
u
n
,
a
n
d
W
.
C
h
e
n
,
“
T
h
e
e
v
o
l
u
t
i
o
n
o
f
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
me
t
h
o
d
s,”
E
n
g
i
n
e
e
r
i
n
g
A
p
p
l
i
c
a
t
i
o
n
s
o
f
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
3
3
,
p
.
1
0
8
4
5
8
,
J
u
l
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
n
g
a
p
p
a
i
.
2
0
2
4
.
1
0
8
4
5
8
.
[
1
7
]
A
.
B
.
A
mj
o
u
d
a
n
d
M
.
A
mr
o
u
c
h
,
“
O
b
j
e
c
t
d
e
t
e
c
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
C
N
N
s
a
n
d
v
i
si
o
n
t
r
a
n
sf
o
r
m
e
r
s:
a
r
e
v
i
e
w
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
3
5
4
7
9
–
3
5
5
1
6
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
6
6
0
9
3
.
[
1
8
]
S
.
R
e
n
,
K
.
H
e
,
R
.
G
i
r
s
h
i
c
k
,
a
n
d
J
.
S
u
n
,
“
F
a
s
t
e
r
R
-
C
N
N
:
t
o
w
a
r
d
s
r
e
a
l
-
t
i
me
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
r
e
g
i
o
n
p
r
o
p
o
sa
l
n
e
t
w
o
r
k
s,”
I
EE
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
Pa
t
t
e
rn
A
n
a
l
y
s
i
s
a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
9
,
n
o
.
6
,
p
p
.
1
1
3
7
–
1
1
4
9
,
J
u
n
.
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
1
6
.
2
5
7
7
0
3
1
.
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.
42
,
No
.
3
,
J
u
n
e
20
2
6
:
8
5
6
-
8
6
4
864
[
1
9
]
R
.
G
i
r
s
h
i
c
k
,
“
F
a
st
R
-
C
N
N
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
I
n
t
e
rn
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
e
r
Vi
s
i
o
n
,
I
EE
E,
D
e
c
.
2
0
1
5
,
p
p
.
1
4
4
0
–
1
4
4
8
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
.
2
0
1
5
.
1
6
9
.
[
2
0
]
M
.
S
o
h
a
n
,
T.
S
a
i
R
a
m,
a
n
d
C
.
V
.
R
a
mi
R
e
d
d
y
,
“
A
r
e
v
i
e
w
o
n
Y
O
LO
v
8
a
n
d
i
t
s
a
d
v
a
n
c
e
m
e
n
t
s,
”
i
n
Al
g
o
ri
t
h
m
s
f
o
r
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
,
2
0
2
4
,
p
p
.
5
2
9
–
5
4
5
.
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
9
8
1
-
99
-
7
9
6
2
-
2
_
3
9
.
[
2
1
]
H
.
X
u
,
L
.
L
e
n
g
,
Z.
Y
a
n
g
,
A
.
B
.
J
.
Te
o
h
,
a
n
d
Z.
J
i
n
,
“
M
u
l
t
i
-
t
a
sk
p
r
e
-
t
r
a
i
n
i
n
g
w
i
t
h
s
o
f
t
b
i
o
me
t
r
i
c
s
f
o
r
t
r
a
n
sf
e
r
-
l
e
a
r
n
i
n
g
p
a
l
m
p
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
,
”
N
e
u
r
a
l
Pr
o
c
e
ss
i
n
g
L
e
t
t
e
rs
,
v
o
l
.
5
5
,
n
o
.
3
,
p
p
.
2
3
4
1
–
2
3
5
8
,
J
u
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
0
6
3
-
0
2
2
-
1
0
8
2
2
-
9.
[
2
2
]
W
.
L
i
n
,
J.
C
h
u
,
L.
L
e
n
g
,
J
.
M
i
a
o
,
a
n
d
L.
W
a
n
g
,
“
F
e
a
t
u
r
e
d
i
s
e
n
t
a
n
g
l
e
m
e
n
t
i
n
o
n
e
-
st
a
g
e
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
,
”
Pa
t
t
e
rn
R
e
c
o
g
n
i
t
i
o
n
,
v
o
l
.
1
4
5
,
p
.
1
0
9
8
7
8
,
J
a
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
c
o
g
.
2
0
2
3
.
1
0
9
8
7
8
.
[
2
3
]
S
.
P
r
a
b
u
,
N
.
Jawa
l
i
,
K
.
J.
A
.
S
u
n
d
a
r
,
K
.
S
h
a
r
v
a
n
i
,
G
.
S
h
a
n
mu
k
h
a
n
j
a
l
i
,
a
n
d
V
.
N
i
r
ma
l
a
,
“
I
n
d
i
a
n
c
o
i
n
d
e
t
e
c
t
i
o
n
a
n
d
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m,”
i
n
Pr
o
c
e
e
d
i
n
g
s o
f
2
0
2
2
6
t
h
Asi
a
n
C
o
n
f
e
r
e
n
c
e
o
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
T
e
c
h
n
o
l
o
g
y
,
A
C
AI
T
2
0
2
2
,
I
EEE,
D
e
c
.
2
0
2
2
,
p
p
.
1
–
6
.
d
o
i
:
1
0
.
1
1
0
9
/
A
C
A
I
T5
6
2
1
2
.
2
0
2
2
.
1
0
1
3
7
9
4
0
.
[
2
4
]
R
o
s
a
l
i
n
a
,
“
A
u
t
o
ma
t
e
d
d
e
t
e
c
t
i
o
n
o
f
a
n
c
i
e
n
t
i
n
d
o
n
e
si
a
n
c
o
i
n
s
f
o
r
h
i
st
o
r
i
c
a
l
n
u
mi
sm
a
t
i
c
i
n
v
e
s
t
i
g
a
t
i
o
n
s
,
”
J
I
T
E
(
J
o
u
rn
a
l
o
f
I
n
f
o
rm
a
t
i
c
s A
n
d
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
En
g
i
n
e
e
ri
n
g
)
,
v
o
l
.
8
,
n
o
.
1
,
p
p
.
7
5
–
8
2
,
2
0
2
4
,
d
o
i
:
1
0
.
3
1
2
8
9
/
j
i
t
e
.
v
8
i
1
.
1
1
2
4
1
.
[
2
5
]
I
.
M
.
S
.
K
u
mara
,
G
.
P
.
R
.
S
.
J
a
t
i
,
a
n
d
N
.
P
.
W
.
Y
u
n
i
a
r
i
,
“
I
n
t
e
g
r
a
t
e
Y
o
l
o
v
8
a
l
g
o
r
i
t
h
m
f
o
r
r
u
p
i
a
h
d
e
n
o
mi
n
a
t
i
o
n
d
e
t
e
c
t
i
o
n
i
n
a
l
l
-
in
-
o
n
e
sm
a
r
t
c
a
n
e
f
o
r
v
i
s
u
a
l
l
y
i
m
p
a
i
r
e
d
,
”
T
e
c
h
n
o
.
C
om
,
v
o
l
.
2
3
,
n
o
.
1
,
p
p
.
1
7
6
–
1
8
6
,
F
e
b
.
2
0
2
4
,
d
o
i
:
1
0
.
6
2
4
1
1
/
t
c
.
v
2
3
i
1
.
9
7
3
4
.
[
2
6
]
G
.
L
i
u
,
J
.
D
i
,
Q
.
W
a
n
g
,
Y
.
Z
h
a
o
,
a
n
d
Y
.
Y
a
n
g
,
“
A
n
E
n
h
a
n
c
e
d
a
n
d
L
i
g
h
t
w
e
i
g
h
t
Y
O
LO
v
8
-
b
a
s
e
d
m
o
d
e
l
f
o
r
a
c
c
u
r
a
t
e
r
i
c
e
p
e
s
t
d
e
t
e
c
t
i
o
n
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
3
,
p
p
.
9
1
0
4
6
–
9
1
0
6
4
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
5
.
3
5
6
9
8
1
9
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Nur
H
a
d
isu
k
m
a
n
a
is
As
sista
n
t
P
r
o
fe
ss
o
r
a
t
De
p
a
rtme
n
t
o
f
I
n
f
o
rm
a
ti
c
s,
F
a
c
u
lt
y
o
f
Co
m
p
u
ter S
c
ien
c
e
,
P
re
sid
e
n
t
U
n
iv
e
rsit
y
,
In
d
o
n
e
sia
.
Th
r
o
u
g
h
o
u
t
h
is ca
re
e
r,
h
e
h
a
s tak
e
n
o
n
a
d
m
in
istrativ
e
re
sp
o
n
si
b
il
it
ies
,
sh
a
p
in
g
p
o
li
c
y
a
n
d
g
u
i
d
in
g
o
r
g
a
n
iza
ti
o
n
a
l
d
e
v
e
lo
p
m
e
n
t
with
in
th
e
d
e
p
a
rtme
n
t
fro
m
2
0
0
8
to
2
0
2
3
.
He
h
a
s
e
a
rn
e
d
M
a
ste
r
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ie
n
c
e
a
t
Co
ll
e
g
e
o
f
Art
&
S
c
ien
c
e
,
Ok
lah
o
m
a
S
tate
Un
iv
e
rsit
y
.
Th
is
in
tern
a
ti
o
n
a
l
a
c
a
d
e
m
ic
e
x
p
e
rien
c
e
n
o
t
o
n
l
y
b
ro
a
d
e
n
e
d
h
is
tec
h
n
ica
l
e
x
p
e
rti
se
b
u
t
a
ls
o
e
n
rich
e
d
h
is
p
e
rsp
e
c
ti
v
e
o
n
g
lo
b
a
l
re
se
a
rc
h
sta
n
d
a
rd
s
a
n
d
m
e
th
o
d
o
l
o
g
ies
.
His
re
se
a
rc
h
a
re
a
s
a
re
S
e
c
u
rit
y
S
y
ste
m
,
wit
h
a
stro
n
g
f
o
c
u
s
o
n
Cry
p
to
g
ra
p
h
y
,
e
n
su
ri
n
g
d
a
ta
in
te
g
rit
y
a
n
d
c
o
n
fid
e
n
ti
a
li
t
y
i
n
d
i
g
it
a
l
c
o
m
m
u
n
ica
ti
o
n
s,
M
o
b
il
e
Co
m
p
u
ti
n
g
wh
ich
e
x
p
l
o
re
s
th
e
d
y
n
a
m
ic
wo
rld
o
f
p
o
r
tab
le
tec
h
n
o
l
o
g
ies
a
n
d
th
e
ir
a
p
p
li
c
a
ti
o
n
,
M
a
c
h
i
n
e
Lea
rn
i
n
g
th
a
t
d
e
l
v
e
s
in
to
i
n
telli
g
e
n
t
s
y
ste
m
s
c
a
p
a
b
le
o
f
lea
rn
i
n
g
a
n
d
a
d
a
p
ti
n
g
,
a
n
d
Alg
o
rit
h
m
a
s
t
h
e
c
o
re
o
f
c
o
m
p
u
tatio
n
a
l
p
ro
b
lem
-
so
lv
i
n
g
.
He
h
a
s
a
u
t
h
o
re
d
o
r
c
o
a
u
th
o
re
d
m
o
re
t
h
a
n
4
0
p
u
b
li
c
a
ti
o
n
s
in
re
p
u
ta
b
le
p
r
o
c
e
e
d
in
g
s
a
n
d
j
o
u
r
n
a
ls.
His
c
u
rre
n
t
re
se
a
rc
h
i
n
tere
sts
a
re
in
o
b
jec
t
d
e
tec
ti
o
n
,
Io
T
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
n
u
r
h
a
d
isu
k
m
a
n
a
@p
re
sid
e
n
t
.
a
c
.
id
.
R.
B.
W
a
h
y
u
is
a
lec
t
u
re
r
a
t
t
h
e
De
p
a
rtme
n
t
o
f
I
n
fo
rm
a
ti
c
s,
F
a
c
u
lt
y
o
f
Co
m
p
u
ter
S
c
ien
c
e
,
P
re
si
d
e
n
t
U
n
iv
e
rsit
y
,
I
n
d
o
n
e
sia
.
He
e
a
rn
e
d
h
is
M
a
ste
r’s
d
e
g
re
e
i
n
Co
m
p
u
ter
S
c
ien
c
e
fro
m
Cu
rti
n
Un
iv
e
rsity
,
P
e
rt
h
,
Wes
tern
Au
stra
li
a
,
a
n
d
c
o
m
p
lete
d
h
is
Do
c
to
ra
l
d
e
g
re
e
a
t
th
e
S
c
h
o
o
l
o
f
El
e
c
tri
c
a
l
a
n
d
Co
m
p
u
ter
En
g
in
e
e
rin
g
(S
T
EI)
,
Ba
n
d
u
n
g
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
(IT
B),
I
n
d
o
n
e
sia
.
In
a
d
d
it
i
o
n
t
o
tea
c
h
in
g
a
t
P
re
sid
e
n
t
Un
i
v
e
rsity
,
h
e
a
lso
lec
tu
re
s
a
t
th
e
F
a
c
u
lt
y
o
f
P
u
b
li
c
He
a
lt
h
,
U
n
iv
e
rsit
y
o
f
In
d
o
n
e
sia
,
in
t
h
e
M
a
ste
r’s
d
e
g
re
e
p
ro
g
ra
m
.
F
u
rth
e
rm
o
re
,
h
e
se
rv
e
s
a
s
a
t
h
e
sis
e
x
a
m
in
e
r
fo
r
th
e
Do
c
to
ra
l
p
r
o
g
ra
m
in
Nu
c
lea
r
a
n
d
Co
m
p
u
t
a
ti
o
n
a
l
P
h
y
sic
s
a
t
Ba
n
d
u
n
g
I
n
stit
u
te
o
f
Tec
h
n
o
lo
g
y
.
His
re
se
a
rc
h
a
re
a
s
in
c
lu
d
e
Co
m
p
u
ter
Visi
o
n
,
P
ro
jec
t
M
a
n
a
g
e
m
e
n
t,
a
n
d
Da
ta
Ce
n
ters
.
He
h
a
s
a
u
th
o
re
d
o
r
c
o
-
a
u
th
o
re
d
m
o
re
th
a
n
3
0
p
u
b
li
c
a
ti
o
n
s
in
r
e
p
u
tab
le
p
ro
c
e
e
d
i
n
g
s
a
n
d
jo
u
rn
a
ls.
His
c
u
rre
n
t
re
se
a
rc
h
in
ter
e
sts
fo
c
u
s
o
n
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
d
a
ta
c
e
n
ter
tec
h
n
o
lo
g
ies
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t:
rb
.
wa
h
y
u
@p
re
sid
e
n
t.
a
c
.
id
.
S
te
wa
r
t
is
a
g
ra
d
u
a
ti
n
g
st
u
d
e
n
t
a
t
De
p
a
r
tme
n
t
o
f
I
n
fo
rm
a
ti
c
s,
F
a
c
u
lt
y
o
f
Co
m
p
u
ter
S
c
ien
c
e
,
P
re
si
d
e
n
t
U
n
iv
e
rsit
y
.
His
p
rima
ry
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
c
o
m
p
u
ter
sc
ien
c
e
,
a
rti
ficia
l
in
telli
g
e
n
c
e
,
a
n
d
m
a
c
h
in
e
lea
rn
i
n
g
.
Th
is
p
a
p
e
r
re
p
re
se
n
ts
h
is
first
f
o
rm
a
l
e
n
try
i
n
to
th
e
fiel
d
,
wh
e
re
h
e
p
lay
e
d
a
sig
n
if
ica
n
t
ro
le as
o
n
e
o
f
th
e
p
rin
c
i
p
a
l
c
o
n
tr
ib
u
to
rs.
His
in
v
o
lv
e
m
e
n
t
sp
a
n
n
e
d
th
e
f
u
ll
sc
o
p
e
o
f
th
e
p
ro
jec
t,
in
c
lu
d
in
g
d
a
tas
e
t
p
re
p
a
ra
ti
o
n
,
a
n
d
so
ftwa
re
imp
lem
e
n
tatio
n
.
In
c
o
ll
a
b
o
ra
ti
o
n
with
o
t
h
e
r
a
u
t
h
o
rs,
h
e
a
c
ti
v
e
ly
p
a
rti
c
ip
a
ted
in
d
a
ta
c
o
ll
e
c
ti
o
n
a
n
d
a
n
a
ly
sis,
a
n
d
c
o
n
tri
b
u
te
d
sig
n
ifi
c
a
n
tl
y
to
t
h
e
writi
n
g
a
n
d
re
fin
e
m
e
n
t
o
f
th
e
m
a
n
u
sc
rip
t
fro
m
i
n
it
ial
p
lan
n
i
n
g
t
h
ro
u
g
h
to
p
u
b
li
c
a
ti
o
n
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
ste
wa
rtq
iu
@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.