I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
,
p
p
.
680
~
689
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ee
cs
.v
41
.
i
2
.
pp
680
-
6
8
9
680
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Ro
bust pal
m
prin
t
bio
m
e
tric
so
lutio
n f
o
r sec
ure
m
o
bi
le
a
uthent
ica
tion
So
n Ng
uy
en,
Art
ho
rn
L
ua
ng
s
o
ds
a
i,
P
a
t
t
a
ra
s
inee
B
ha
t
t
a
r
a
k
o
s
o
l
D
e
p
a
r
t
me
n
t
o
f
M
a
t
h
e
mat
i
c
s
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
,
C
h
u
l
a
l
o
n
g
k
o
r
n
U
n
i
v
e
r
si
t
y
,
B
a
n
g
k
o
k
,
T
h
a
i
l
a
n
d
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
4
,
2025
R
ev
i
s
ed
J
an
3
,
2
0
2
6
A
cc
ep
ted
J
an
11
,
2
0
2
6
S
m
a
rtp
h
o
n
e
s
in
c
re
a
sin
g
ly
re
l
y
o
n
b
i
o
m
e
tri
c
a
u
th
e
n
ti
c
a
ti
o
n
f
o
r
a
c
c
e
ss
to
f
in
a
n
c
ial
a
n
d
p
e
rso
n
a
l
se
rv
ice
s,
c
re
a
ti
n
g
a
n
e
e
d
f
o
r
p
a
lm
p
rin
t
re
c
o
g
n
it
io
n
th
a
t
is
a
c
c
u
ra
te,
f
a
st,
a
n
d
d
e
p
lo
y
a
b
le
o
n
d
e
v
ice
.
T
h
is
p
a
p
e
r
p
ro
p
o
s
e
s
a
n
e
n
d
-
to
-
e
n
d
sm
a
rtp
h
o
n
e
p
a
lm
p
rin
t
a
u
th
e
n
ti
c
a
ti
o
n
f
ra
m
e
w
o
rk
th
a
t
in
teg
ra
tes
g
u
id
e
d
m
o
b
il
e
im
a
g
e
c
a
p
tu
re
,
lan
d
m
a
rk
-
b
a
se
d
re
g
io
n
-
of
-
in
tere
st
(
ROI
)
e
x
trac
ti
o
n
,
a
n
d
c
o
m
p
a
c
t
e
m
b
e
d
d
in
g
in
f
e
re
n
c
e
.
A
Re
sN
e
t
-
1
8
tea
c
h
e
r
is
f
irst
train
e
d
w
it
h
se
lf
-
su
p
e
rv
ise
d
c
o
n
t
ra
stiv
e
lea
rn
in
g
to
re
d
u
c
e
d
e
p
e
n
d
e
n
c
e
o
n
la
b
e
led
b
io
m
e
tri
c
d
a
ta,
th
e
n
d
isti
l
led
in
t
o
a
li
g
h
tw
e
ig
h
t
M
o
b
il
e
Ne
tV
3
st
u
d
e
n
t
f
o
r
e
ff
ici
e
n
t
m
o
b
il
e
d
e
p
l
o
y
m
e
n
t.
T
h
e
lea
rn
e
d
e
m
b
e
d
d
in
g
s
su
p
p
o
r
t
b
o
th
o
n
-
d
e
v
ice
v
e
ri
f
ica
ti
o
n
a
n
d
larg
e
-
s
c
a
le
id
e
n
ti
f
ica
ti
o
n
u
sin
g
a
n
a
p
p
ro
x
im
a
te
n
e
a
re
st
n
e
ig
h
b
o
r
i
n
d
e
x
(F
A
IS
S
).
Ex
p
e
rim
e
n
ts
o
n
a
p
u
b
li
c
Ka
g
g
le
p
a
lm
d
a
tas
e
t
a
c
h
iev
e
9
9
.
2
%
a
c
c
u
ra
c
y
w
it
h
a
0
.
1
5
%
e
q
u
a
l
e
rro
r
ra
te
(
E
ER
)
.
On
a
n
iP
h
o
n
e
1
3
,
t
h
e
e
n
d
-
to
-
e
n
d
p
i
p
e
li
n
e
ru
n
s
i
n
8
7
.
0
m
s
w
it
h
a
1
2
.
4
M
B
stu
d
e
n
t
m
o
d
e
l.
F
o
r
a
1
m
il
li
o
n
-
e
n
try
g
a
l
ler
y
,
F
A
IS
S
p
ro
v
id
e
s
3
2
m
s
q
u
e
ry
late
n
c
y
w
h
i
le
m
a
in
tain
in
g
9
9
.
5
%
Re
c
a
ll
@1
.
L
im
it
a
ti
o
n
s
in
c
lu
d
e
e
v
a
lu
a
t
io
n
u
n
d
e
r
m
o
stl
y
c
o
n
tro
ll
e
d
c
a
p
t
u
re
c
o
n
d
it
i
o
n
s
a
n
d
t
h
e
a
b
se
n
c
e
o
f
a
n
e
x
p
li
c
i
t
li
v
e
n
e
ss
o
r
p
re
se
n
tati
o
n
a
tt
a
c
k
d
e
tec
ti
o
n
(
P
A
D
)
m
o
d
u
le;
f
u
tu
re
w
o
rk
w
il
l
a
d
d
re
ss
u
n
c
o
n
stra
in
e
d
tes
ti
n
g
a
n
d
a
n
ti
-
s
p
o
o
f
in
g
in
teg
ra
ti
o
n
.
K
ey
w
o
r
d
s
:
B
io
m
etr
ic
s
ec
u
r
it
y
C
lo
u
d
co
m
p
u
tin
g
C
o
n
tr
ast
iv
e
lear
n
in
g
Mo
b
ile
s
ec
u
r
it
y
P
alm
p
r
i
n
t a
u
t
h
e
n
ticatio
n
Self
-
s
u
p
er
v
i
s
ed
lear
n
i
n
g
Vec
to
r
d
atab
ase
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
:
P
attar
asin
ee
B
h
attar
ak
o
s
o
l
Dep
ar
t
m
en
t o
f
Ma
th
e
m
at
ics a
n
d
C
o
m
p
u
ter
Scien
ce
,
Fac
u
lt
y
o
f
Scien
ce
,
C
h
u
lalo
n
g
k
o
r
n
U
n
iv
er
s
it
y
B
an
g
k
o
k
,
T
h
ailan
d
E
m
ail:
p
attar
asi
n
ee
.
b
@
c
h
u
la.
a
c.
th
1.
I
NT
RO
D
UCT
I
O
N
S
m
ar
tp
h
o
n
es
h
a
v
e
b
ec
o
m
e
t
h
e
p
r
i
m
ar
y
g
a
te
w
a
y
to
d
ig
i
tal
s
er
v
ice
s
s
u
c
h
as
m
o
b
ile
b
an
k
i
n
g
,
elec
tr
o
n
ic
p
a
y
m
en
ts
,
tele
m
ed
i
cin
e,
an
d
r
e
m
o
te
w
o
r
k
p
lat
f
o
r
m
s
.
As
a
r
es
u
lt,
a
u
t
h
en
ticatio
n
m
ec
h
a
n
i
s
m
s
m
u
s
t
ac
h
iev
e
b
o
th
s
tr
o
n
g
s
ec
u
r
it
y
an
d
ac
ce
p
tab
le
u
s
er
ex
p
er
ien
ce
u
n
d
er
tig
h
t
o
n
-
d
ev
ice
co
n
s
tr
ain
t
s
.
B
io
m
etr
ic
au
th
e
n
tica
tio
n
is
attr
ac
ti
v
e
b
ec
au
s
e
it
ca
n
r
ed
u
ce
r
elian
ce
o
n
p
ass
w
o
r
d
s
an
d
m
it
ig
ate
r
i
s
k
s
ass
o
ciate
d
w
it
h
cr
ed
en
tial
t
h
ef
t;
h
o
w
ev
er
,
p
r
ac
tical
d
ep
lo
y
m
e
n
t
also
r
ai
s
e
s
co
n
ce
r
n
s
ab
o
u
t
p
r
iv
ac
y
,
s
p
o
o
f
in
g
,
an
d
s
ec
u
r
e
te
m
p
late
s
to
r
ag
e.
W
ith
in
t
h
i
s
co
n
tex
t,
p
al
m
p
r
in
t
r
ec
o
g
n
iti
o
n
is
a
co
m
p
elli
n
g
m
o
d
alit
y
b
ec
au
s
e
t
h
e
p
al
m
co
n
tain
s
d
en
s
e
an
d
d
is
cr
i
m
i
n
ati
v
e
te
x
tu
r
e
p
atter
n
s
(
p
r
in
cip
al
lin
es,
w
r
i
n
k
les,
a
n
d
f
i
n
e
cr
ea
s
es)
t
h
at
ar
e
r
elativ
el
y
s
tab
le
o
v
er
ti
m
e,
an
d
it
ca
n
b
e
ca
p
tu
r
ed
in
a
co
n
ta
ctless
m
a
n
n
er
u
s
i
n
g
co
m
m
o
d
i
t
y
ca
m
er
as.
R
ec
e
n
t
s
u
r
v
e
y
s
an
d
r
ev
ie
w
s
co
n
s
i
s
t
en
tl
y
r
ep
o
r
t
th
at
m
o
d
er
n
p
al
m
p
r
in
t
s
y
s
te
m
s
esp
ec
iall
y
d
ee
p
-
lear
n
i
n
g
-
b
ased
ap
p
r
o
ac
h
es
h
av
e
ac
h
ie
v
ed
s
u
b
s
tan
tial
g
ai
n
s
i
n
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
an
d
r
o
b
u
s
tn
e
s
s
co
m
p
ar
ed
w
i
th
ea
r
lier
h
an
d
cr
af
ted
p
ip
elin
es,
w
h
ile
also
h
ig
h
li
g
h
ti
n
g
th
at
r
ea
l
is
tic
d
ep
lo
y
m
e
n
t
r
e
m
ai
n
s
c
h
alle
n
g
i
n
g
d
u
e
to
ac
q
u
is
itio
n
v
ar
iab
ilit
y
a
n
d
o
p
e
r
atio
n
al
co
n
s
tr
ai
n
ts
[
1
]
,
[
2
]
.
Fro
m
a
tech
n
ical
p
er
s
p
ec
tiv
e,
t
w
o
a
s
p
ec
ts
s
tr
o
n
g
l
y
i
n
f
l
u
en
ce
p
al
m
p
r
i
n
t
p
er
f
o
r
m
a
n
c
e
in
r
ea
l
ap
p
licatio
n
s
: r
eliab
le
r
eg
io
n
-
of
-
in
ter
e
s
t (
R
OI
)
e
x
tr
ac
tio
n
an
d
d
is
cr
i
m
i
n
ati
v
e
f
ea
t
u
r
e
lear
n
i
n
g
.
I
n
u
n
co
n
s
tr
ain
ed
co
n
tactles
s
ca
p
tu
r
e,
ch
an
g
es
i
n
h
a
n
d
p
o
s
e,
d
is
tan
ce
to
ca
m
e
r
a,
illu
m
i
n
atio
n
,
an
d
p
ar
tia
l
o
c
clu
s
io
n
ca
n
d
eg
r
ad
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
R
o
b
u
s
t p
a
lmp
r
in
t
b
io
metric so
lu
tio
n
fo
r
s
ec
u
r
e
mo
b
ile
a
u
th
en
tica
tio
n
(
S
o
n
N
g
u
ye
n
)
681
R
OI
co
n
s
is
te
n
c
y
a
n
d
d
o
w
n
s
tr
ea
m
r
ec
o
g
n
itio
n
.
R
ec
e
n
t
r
e
s
ea
r
ch
h
a
s
t
h
er
ef
o
r
e
e
m
p
h
a
s
ized
co
m
p
lete
an
d
s
tab
le
R
OI
s
tr
ateg
ie
s
f
o
r
u
n
co
n
s
tr
ain
ed
p
al
m
p
r
in
t
r
ec
o
g
n
i
ti
o
n
an
d
d
em
o
n
s
tr
ated
m
ea
s
u
r
ab
le
r
o
b
u
s
tn
e
s
s
i
m
p
r
o
v
e
m
en
t
s
u
n
d
er
v
ar
iab
le
ac
q
u
i
s
itio
n
co
n
d
itio
n
s
[
3
]
.
I
n
p
ar
allel,
d
ee
p
m
o
d
els
w
it
h
f
ea
tu
r
e
f
u
s
io
n
o
r
g
ati
n
g
m
ec
h
a
n
i
s
m
s
h
av
e
b
ee
n
i
n
tr
o
d
u
ce
d
to
en
h
an
ce
d
is
cr
i
m
i
n
ati
v
e
r
ep
r
esen
tatio
n
lear
n
in
g
,
p
ar
t
icu
lar
l
y
w
h
e
n
i
n
tr
a
-
class
v
ar
iat
io
n
is
n
o
n
-
tr
i
v
ial
[
4
]
.
No
n
eth
eless
,
m
an
y
h
i
g
h
-
p
er
f
o
r
m
in
g
a
ca
d
e
m
ic
m
o
d
els
s
till
ass
u
m
e
w
ell
-
co
n
tr
o
lled
ca
p
tu
r
e
an
d
d
o
n
o
t
ex
p
licitl
y
r
ep
o
r
t
en
d
-
to
-
e
n
d
s
y
s
te
m
b
e
h
av
io
r
f
o
r
s
m
ar
tp
h
o
n
e
d
ep
lo
y
m
e
n
t.
Fo
r
m
o
b
ile
s
ett
in
g
s
,
e
f
f
icie
n
c
y
-
o
r
ien
ted
p
al
m
p
r
i
n
t
d
esi
g
n
s
h
a
v
e
b
ee
n
ex
p
lo
r
ed
,
s
u
c
h
a
s
co
m
p
ac
t
co
d
in
g
o
r
h
as
h
i
n
g
-
b
ased
r
e
p
r
esen
tatio
n
s
th
at
r
ed
u
ce
m
atc
h
i
n
g
co
s
t
an
d
s
to
r
ag
e
o
v
er
h
ea
d
w
h
ile
m
ai
n
tai
n
in
g
co
m
p
etiti
v
e
v
er
if
ica
tio
n
p
er
f
o
r
m
a
n
ce
[
5
]
.
C
las
s
ical
co
d
in
g
ap
p
r
o
ac
h
es
r
e
m
ai
n
r
ele
v
an
t
as
lig
h
t
w
e
ig
h
t
b
aselin
e
s
a
n
d
as
a
r
ef
er
en
ce
p
o
in
t
f
o
r
u
n
d
er
s
ta
n
d
in
g
w
h
at
p
er
f
o
r
m
an
ce
is
ac
h
ie
v
ab
le
w
it
h
lo
w
co
m
p
u
tatio
n
a
l
co
m
p
lex
it
y
[
6
]
.
I
n
ad
d
itio
n
,
in
ter
p
r
etab
ilit
y
is
in
cr
ea
s
in
g
l
y
i
m
p
o
r
tan
t
f
o
r
b
io
m
etr
ic
s
y
s
te
m
s
u
s
ed
in
s
ec
u
r
it
y
-
s
e
n
s
i
tiv
e
ap
p
licatio
n
s
;
ex
p
lai
n
ab
le
p
alm
p
r
in
t
r
ec
o
g
n
itio
n
p
ip
elin
e
s
h
av
e
b
ee
n
s
tu
d
ied
to
p
r
o
v
id
e
h
u
m
an
-
u
n
d
er
s
tan
d
a
b
l
e
ev
id
en
ce
ab
o
u
t
w
h
ic
h
r
eg
io
n
s
o
r
p
atter
n
s
co
n
tr
ib
u
te
m
o
s
t to
th
e
d
ec
is
io
n
[
7
]
.
B
ey
o
n
d
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
,
r
ea
l
-
w
o
r
ld
p
al
m
p
r
in
t
au
t
h
en
ticatio
n
m
u
s
t
al
s
o
ad
d
r
ess
em
er
g
i
n
g
s
ec
u
r
it
y
th
r
ea
t
s
an
d
p
r
iv
ac
y
c
o
n
s
tr
ain
ts
.
Ge
n
er
ativ
e
m
o
d
eli
n
g
h
as
b
ee
n
ac
ti
v
el
y
s
t
u
d
ied
f
o
r
p
al
m
i
m
a
g
er
y
,
in
cl
u
d
in
g
th
e
u
s
e
o
f
G
A
N
-
b
ased
p
ip
elin
es
in
p
al
m
p
r
in
t
r
ec
o
g
n
itio
n
r
esear
ch
a
n
d
an
al
y
s
i
s
[
8
]
,
as
w
e
ll
as
t
h
e
ex
p
licit
p
r
o
b
le
m
o
f
d
etec
ti
n
g
s
y
n
t
h
etic
o
r
m
a
n
ip
u
lated
p
alm
i
m
a
g
es
[
9
]
.
R
ec
en
t
w
o
r
k
h
as
also
s
h
o
w
n
th
at
d
if
f
u
s
io
n
m
o
d
els
ca
n
g
e
n
er
ate
r
ea
lis
tic
co
n
tactless
p
al
m
p
r
i
n
t
s
,
w
h
ic
h
f
u
r
t
h
er
m
o
ti
v
ates
th
e
n
ee
d
f
o
r
s
y
s
te
m
a
tic
an
t
i
-
s
p
o
o
f
in
g
e
v
alu
a
tio
n
w
h
en
d
ep
lo
y
i
n
g
p
alm
p
r
i
n
t
au
t
h
e
n
ticatio
n
o
u
ts
id
e
co
n
tr
o
lled
en
v
ir
o
n
m
e
n
t
s
[
1
0
]
.
Mo
r
e
b
r
o
ad
ly
,
b
io
m
e
tr
ics
ar
e
k
n
o
w
n
t
o
b
e
v
u
ln
er
ab
le
to
ad
v
er
s
ar
ia
l
m
an
ip
u
latio
n
an
d
p
r
esen
tatio
n
attac
k
s
,
an
d
r
ec
en
t
s
t
u
d
ies
o
n
co
m
b
in
ed
attac
k
s
ag
ai
n
s
t
r
ec
o
g
n
itio
n
an
d
p
r
esen
tatio
n
attac
k
d
etec
tio
n
(
P
A
D)
e
m
p
h
a
s
ize
t
h
at
s
tr
o
n
g
v
er
if
ica
tio
n
ac
c
u
r
ac
y
alo
n
e
is
in
s
u
f
f
icie
n
t
to
g
u
ar
an
tee
s
ec
u
r
it
y
[
1
1
]
.
A
t
th
e
te
m
p
late
le
v
el,
o
n
ce
b
io
m
etr
ic
te
m
p
lates
ar
e
co
m
p
r
o
m
is
ed
,
t
h
e
y
ca
n
n
o
t
b
e
“
r
e
s
e
t”
lik
e
p
as
s
w
o
r
d
s
;
ca
n
ce
llab
le
te
m
p
late
g
en
er
ati
o
n
is
th
er
e
f
o
r
e
an
i
m
p
o
r
tan
t
d
ir
ec
tio
n
to
r
ed
u
ce
t
h
e
i
m
p
ac
t
o
f
te
m
p
la
te
lea
k
ag
e
an
d
en
ab
le
r
e
v
o
ca
tio
n
a
n
d
r
e
-
is
s
u
an
ce
[
1
2
]
.
I
n
ad
d
itio
n
,
s
in
c
e
b
io
m
e
tr
ic
d
ata
is
s
e
n
s
iti
v
e,
p
r
iv
ac
y
-
p
r
eser
v
i
n
g
tr
ain
i
n
g
a
n
d
u
p
d
ati
n
g
p
ar
ad
ig
m
s
s
u
c
h
as
f
ed
er
ated
lear
n
i
n
g
h
a
v
e
b
ee
n
p
r
o
p
o
s
ed
f
o
r
p
al
m
v
er
if
icatio
n
to
r
ed
u
ce
th
e
n
ee
d
to
ce
n
tr
alize
r
a
w
d
ata
d
u
r
i
n
g
m
o
d
el
i
m
p
r
o
v
e
m
en
t c
y
cle
s
[
1
3
]
.
T
h
ese
co
n
s
id
er
atio
n
s
b
ec
o
m
e
m
o
r
e
cr
itical
o
n
m
o
b
ile
an
d
ed
g
e
p
latf
o
r
m
s
.
E
d
g
e
i
n
telli
g
e
n
ce
r
esear
ch
e
m
p
h
a
s
izes
th
a
t
th
e
o
p
er
ati
o
n
al
co
n
s
tr
ain
t
s
o
f
o
n
-
d
e
v
ice
A
I
laten
c
y
,
m
e
m
o
r
y
f
o
o
tp
r
in
t,
en
er
g
y
u
s
a
g
e,
an
d
r
eliab
ilit
y
u
n
d
er
h
eter
o
g
en
eo
u
s
h
ar
d
w
ar
e
o
f
ten
d
o
m
i
n
ate
s
y
s
te
m
f
ea
s
ib
ilit
y
a
n
d
u
s
er
ac
ce
p
tan
ce
[
1
4
]
.
T
h
er
ef
o
r
e,
m
o
b
ile
p
al
m
p
r
i
n
t
au
t
h
en
ticati
o
n
r
eq
u
ir
es
a
co
m
p
lete
w
o
r
k
f
lo
w
t
h
at
in
cl
u
d
es
s
tab
le
ac
q
u
is
itio
n
,
f
a
s
t
p
r
ep
r
o
ce
s
s
in
g
,
an
d
ef
f
icie
n
t
i
n
f
er
en
ce
,
r
ath
er
th
a
n
an
ac
cu
r
ac
y
-
o
n
l
y
e
v
al
u
atio
n
.
P
r
ac
tical
m
o
b
ile
v
is
io
n
d
ep
lo
y
m
e
n
t
h
a
s
b
en
ef
i
ted
f
r
o
m
r
ea
l
-
ti
m
e
p
er
ce
p
tio
n
f
r
a
m
e
w
o
r
k
s
th
at
s
u
p
p
o
r
t
f
ast
lan
d
m
ar
k
d
etec
tio
n
an
d
s
tr
ea
m
li
n
ed
o
n
-
d
ev
ice
p
ip
elin
e
s
,
en
ab
lin
g
r
o
b
u
s
t
R
OI
p
r
o
ce
s
s
in
g
u
n
d
er
in
ter
ac
ti
v
e
ca
m
er
a
ca
p
tu
r
e
co
n
d
itio
n
s
[
1
5
]
.
A
t
t
h
e
s
a
m
e
t
i
m
e,
co
ll
ec
tin
g
lar
g
e
-
s
ca
le
lab
eled
b
i
o
m
e
t
r
ic
d
ataset
s
i
s
ex
p
en
s
iv
e
a
n
d
r
aise
s
et
h
ical
a
n
d
leg
a
l c
h
a
llen
g
e
s
.
Sel
f
-
s
u
p
er
v
is
ed
lear
n
in
g
(
S
S
L
)
o
f
f
er
s
a
p
r
o
m
i
s
in
g
ap
p
r
o
ac
h
to
r
ed
u
ce
lab
el
d
e
p
en
d
en
ce
b
y
lear
n
in
g
v
ie
w
-
i
n
v
ar
ia
n
t
r
ep
r
esen
tatio
n
s
f
r
o
m
u
n
lab
ele
d
d
ata;
co
n
tr
asti
v
e
lear
n
in
g
f
r
a
m
e
w
o
r
k
s
s
u
c
h
as
Si
m
C
L
R
p
r
o
v
id
e
a
w
id
el
y
ad
o
p
ted
f
o
u
n
d
atio
n
f
o
r
th
is
p
ar
ad
ig
m
[
1
6
]
.
SS
L
h
a
s
also
d
e
m
o
n
s
tr
ated
v
al
u
e
in
b
i
o
m
e
tr
ic
-
ad
j
ac
en
t
d
o
m
ai
n
s
s
u
c
h
as
f
ac
e
r
ep
r
esen
tatio
n
lear
n
in
g
,
m
o
ti
v
ati
n
g
it
s
u
s
e
a
s
a
d
ata
-
ef
f
icie
n
t
p
r
etr
ai
n
in
g
s
tr
ate
g
y
w
h
en
lab
els
ar
e
li
m
ited
o
r
c
o
s
tl
y
[
1
7
]
.
Si
m
ilar
S
SL
p
r
i
n
cip
les
h
a
v
e
b
ee
n
ap
p
lied
in
o
th
er
s
tr
u
ctu
r
ed
r
ec
o
g
n
itio
n
p
r
o
b
lem
s
,
s
u
p
p
o
r
tin
g
th
e
b
r
o
ad
er
claim
th
at
r
ep
r
esen
tatio
n
q
u
alit
y
ca
n
b
e
i
m
p
r
o
v
ed
b
y
le
ar
n
in
g
f
r
o
m
co
n
te
x
t a
n
d
i
n
v
ar
i
an
ce
s
r
ath
er
t
h
a
n
ex
p
licit lab
el
s
alo
n
e
[
1
8
]
.
I
n
ad
d
i
tio
n
to
d
ata
ef
f
icie
n
c
y
,
m
o
b
ile
f
ea
s
ib
ilit
y
r
eq
u
i
r
es
co
m
p
ac
t
n
eu
r
al
ar
c
h
itec
tu
r
es
an
d
co
m
p
r
es
s
io
n
tech
n
iq
u
es.
Mo
b
ile
-
o
p
ti
m
ized
b
ac
k
b
o
n
es
(
e.
g
.
,
Mo
b
ileNet
-
f
a
m
il
y
m
o
d
els)
h
av
e
b
ee
n
d
esig
n
ed
ex
p
licitl
y
to
i
m
p
r
o
v
e
th
e
ac
cu
r
ac
y
–
late
n
c
y
tr
ad
e
-
o
f
f
o
n
s
m
ar
tp
h
o
n
es
a
n
d
e
m
b
ed
d
ed
d
ev
ices
[
1
9
]
.
L
i
g
h
t
w
ei
g
h
t
n
et
w
o
r
k
d
e
s
ig
n
s
tr
ateg
ie
s
t
h
at
g
e
n
er
ate
m
o
r
e
f
ea
tu
r
es
f
r
o
m
in
e
x
p
en
s
i
v
e
o
p
er
atio
n
s
f
u
r
th
er
s
tr
en
g
th
e
n
th
i
s
d
ir
ec
tio
n
,
p
r
o
v
id
in
g
a
n
ad
d
itio
n
al
o
p
tio
n
f
o
r
i
m
p
r
o
v
i
n
g
ef
f
ic
ien
c
y
w
it
h
o
u
t
p
r
o
h
ib
iti
v
e
ac
cu
r
ac
y
lo
s
s
[
2
0
]
.
Kn
o
w
l
ed
g
e
d
is
till
atio
n
is
a
p
r
ac
tical
co
m
p
r
ess
io
n
m
ec
h
an
is
m
f
o
r
tr
an
s
f
er
r
in
g
r
ep
r
esen
tatio
n
al
ca
p
ab
ilit
y
f
r
o
m
a
lar
g
er
teac
h
er
to
a
s
m
a
ller
s
tu
d
e
n
t
m
o
d
el,
o
f
ten
p
r
eser
v
i
n
g
v
er
i
f
icatio
n
p
er
f
o
r
m
a
n
ce
w
h
ile
r
ed
u
ci
n
g
m
o
d
el
s
ize
a
n
d
in
f
er
en
ce
t
i
m
e
[
2
1
]
.
Qu
an
tizatio
n
a
n
d
in
te
g
er
-
o
n
l
y
i
n
f
er
en
c
e
m
et
h
o
d
s
ca
n
f
u
r
t
h
er
r
ed
u
ce
r
u
n
ti
m
e
an
d
m
e
m
o
r
y
co
s
t
s
,
b
u
t
m
u
s
t
b
e
ap
p
lied
ca
r
ef
u
ll
y
to
av
o
id
u
n
ac
ce
p
tab
le
ac
cu
r
ac
y
d
eg
r
ad
atio
n
i
n
b
io
m
etr
ic
v
er
if
icat
io
n
s
et
tin
g
s
[
2
2
]
.
Fin
all
y
,
m
o
b
ile
p
al
m
p
r
in
t
s
y
s
te
m
s
m
u
s
t
s
u
p
p
o
r
t
b
o
th
v
er
if
icat
io
n
(
1
:1
)
an
d
id
en
tif
icat
io
n
(
1
:N)
.
W
h
ile
m
a
n
y
ac
ad
e
m
ic
s
tu
d
i
es
f
o
cu
s
o
n
v
er
if
ica
tio
n
,
r
ea
l
d
ep
lo
y
m
e
n
t
s
m
a
y
r
eq
u
ir
e
s
ea
r
ch
in
g
a
lar
g
e
en
r
o
ll
m
e
n
t
d
atab
ase
(
e.
g
.
,
ca
m
p
u
s
ac
ce
s
s
,
en
ter
p
r
is
e
id
e
n
ti
t
y
,
o
r
m
u
lti
-
u
s
er
d
ev
ice
s
ce
n
a
r
io
s
)
.
A
t
s
ca
le,
n
a
ïv
e
ex
h
a
u
s
tiv
e
m
a
tch
i
n
g
b
ec
o
m
e
s
i
m
p
r
ac
tical,
m
o
ti
v
ati
n
g
ap
p
r
o
x
i
m
ate
n
ea
r
est
n
ei
g
h
b
o
r
(
A
N
N)
in
d
ex
in
g
.
F
A
I
SS
p
r
o
v
id
es
a
p
r
ac
tical
an
d
w
id
ely
u
s
ed
to
o
lk
it
f
o
r
s
i
m
ilar
it
y
s
e
ar
ch
an
d
in
d
ex
i
n
g
t
h
at
s
u
p
p
o
r
ts
h
i
g
h
-
t
h
r
o
u
g
h
p
u
t
r
etr
iev
al
p
ip
elin
es
[
2
3
]
.
GP
U
-
ac
ce
ler
ated
s
i
m
ilar
it
y
s
ea
r
ch
h
as
d
e
m
o
n
s
tr
ated
th
at
lar
g
e
-
s
ca
le
m
atch
in
g
ca
n
b
e
ac
h
iev
ed
ef
f
icie
n
tl
y
,
en
ab
lin
g
b
illi
o
n
-
s
ca
le
r
etr
iev
al
u
n
d
er
s
u
itab
le
in
d
ex
i
n
g
an
d
h
ar
d
w
ar
e
co
n
f
ig
u
r
atio
n
s
[
2
4
]
.
R
ec
en
t
A
N
N
r
esear
ch
co
n
ti
n
u
e
s
to
s
tu
d
y
t
h
e
tr
ad
e
-
o
f
f
s
b
et
w
ee
n
r
ec
all,
laten
c
y
,
an
d
m
e
m
o
r
y
,
w
h
ic
h
ar
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
6
8
0
-
689
682
cr
itical
to
s
y
s
te
m
d
esi
g
n
w
h
e
n
th
e
g
aller
y
s
ize
g
r
o
w
s
s
i
g
n
i
f
ica
n
tl
y
[
2
5
]
.
R
elate
d
w
o
r
k
in
o
th
er
s
m
ar
tp
h
o
n
e
b
io
m
etr
ic
s
,
s
u
c
h
as
v
is
ib
le
-
li
g
h
t
ir
is
r
ec
o
g
n
itio
n
,
f
u
r
t
h
er
co
n
f
ir
m
s
th
at
h
i
g
h
-
ac
cu
r
ac
y
b
io
m
etr
ic
au
th
e
n
ticatio
n
ca
n
b
e
f
ea
s
ib
le
o
n
co
n
s
u
m
er
d
ev
ices
w
h
e
n
s
y
s
te
m
d
esi
g
n
i
s
t
ig
h
tl
y
al
ig
n
ed
w
it
h
m
o
b
ile
co
n
s
tr
ain
t
s
[
2
6
]
.
2.
T
H
E
P
RO
P
O
SE
D
M
E
T
H
O
D
T
h
is
s
ec
tio
n
p
r
esen
t
s
t
h
e
p
r
o
p
o
s
ed
en
d
-
to
-
e
n
d
p
al
m
p
r
i
n
t
au
t
h
en
ticatio
n
f
r
a
m
e
w
o
r
k
,
d
esig
n
ed
to
j
o
in
tl
y
ad
d
r
ess
t
h
r
ee
p
r
ac
tical
d
ep
lo
y
m
en
t
r
eq
u
ir
e
m
e
n
t
s
:
(
i)
r
o
b
u
s
t
p
al
m
r
ep
r
ese
n
tatio
n
lea
r
n
in
g
w
i
th
r
ed
u
ce
d
r
elian
ce
o
n
lab
eled
b
io
m
etr
ic
d
ata,
(
ii)
ef
f
icien
t
o
n
-
d
e
v
ice
in
f
er
en
ce
f
o
r
r
ea
l
-
ti
m
e
s
m
ar
tp
h
o
n
e
u
s
e,
an
d
(
iii)
s
ca
lab
le
m
atc
h
i
n
g
t
h
at
s
u
p
p
o
r
ts
b
o
th
1
:1
v
er
if
ica
tio
n
a
n
d
1
:N
id
en
tific
atio
n
.
T
h
e
o
v
er
all
w
o
r
k
f
lo
w
i
s
s
u
m
m
ar
ized
in
Fi
g
u
r
e
1
,
an
d
th
e
m
o
b
ile
ac
q
u
is
it
io
n
i
n
ter
f
ac
es
an
d
R
OI
lo
ca
lizat
io
n
o
u
tp
u
ts
ar
e
s
h
o
w
n
i
n
Fig
u
r
e
2
an
d
Fig
u
r
e
3
,
r
esp
ec
tiv
el
y
.
2
.
1
.
Sy
s
t
em
w
o
rk
f
lo
w
o
v
er
v
i
ew
T
h
e
s
y
s
te
m
f
o
llo
w
s
a
m
o
b
ile
-
f
ir
s
t
p
ip
elin
e
,
a
s
d
ep
icted
in
Fig
u
r
e
1
.
Du
r
i
n
g
e
n
r
o
ll
m
e
n
t
,
th
e
u
s
er
ca
p
tu
r
es
p
al
m
i
m
a
g
es
u
s
in
g
th
e
s
m
ar
tp
h
o
n
e
ca
m
er
a
w
i
th
in
-
ap
p
g
u
id
an
ce
;
t
h
e
ca
p
tu
r
ed
f
r
a
m
es
ar
e
p
r
ep
r
o
ce
s
s
ed
o
n
-
d
ev
ice
to
ex
tr
ac
t
a
s
tan
d
ar
d
ized
p
alm
R
O
I
,
an
d
a
co
m
p
ac
t
C
NN
g
e
n
er
ates
a
f
ix
ed
-
le
n
g
th
e
m
b
ed
d
in
g
t
h
at
i
s
s
to
r
ed
as
th
e
u
s
e
r
te
m
p
late.
D
u
r
i
n
g
a
u
t
h
e
n
ticatio
n
,
th
e
s
a
m
e
p
r
ep
r
o
ce
s
s
in
g
a
n
d
e
m
b
ed
d
in
g
s
tep
s
ar
e
ap
p
lied
to
th
e
q
u
er
y
s
a
m
p
le.
T
h
e
q
u
er
y
e
m
b
ed
d
in
g
is
t
h
en
m
atc
h
ed
eit
h
er
lo
ca
ll
y
(
1
:1
v
er
if
icatio
n
)
o
r
s
en
t to
a
clo
u
d
b
ac
k
en
d
f
o
r
lar
g
e
-
s
ca
le
r
etr
iev
al
(
1
:N
id
en
t
if
icatio
n
)
.
Of
f
li
n
e,
th
e
f
ea
t
u
r
e
ex
tr
ac
to
r
is
tr
ain
ed
in
t
w
o
s
tag
e
s
:
a
s
el
f
-
s
u
p
er
v
is
ed
teac
h
er
m
o
d
el
is
f
ir
s
t
lear
n
ed
f
r
o
m
u
n
lab
eled
p
al
m
R
O
I
s
an
d
th
en
d
is
t
illed
in
to
a
lig
h
t
w
ei
g
h
t st
u
d
en
t
m
o
d
el
s
u
itab
le
f
o
r
d
ep
lo
y
m
en
t o
n
r
eso
u
r
ce
-
co
n
s
tr
ain
ed
d
ev
ice
s
[
1
4
]
.
Fig
u
r
e
1
.
W
o
r
k
f
lo
w
p
ip
eli
n
e
2
.
2
.
M
o
bil
e
i
m
a
g
e
a
cquis
it
io
n a
nd
re
a
l
-
t
i
m
e
u
s
er
g
uid
a
nc
e
T
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
ac
q
u
ir
es
co
n
tactless
p
al
m
i
m
ag
e
s
u
s
i
n
g
s
tan
d
ar
d
s
m
ar
tp
h
o
n
e
ca
m
er
as.
T
o
s
tan
d
ar
d
ize
ca
p
tu
r
e
q
u
alit
y
a
n
d
r
ed
u
ce
u
s
er
-
i
n
d
u
ce
d
v
ar
ia
tio
n
,
th
e
m
o
b
ile
ap
p
licatio
n
p
r
o
v
id
es
r
ea
l
-
ti
m
e
f
ee
d
b
ac
k
b
ef
o
r
e
allo
w
i
n
g
ca
p
tu
r
e.
Fi
g
u
r
e
2
ill
u
s
tr
ate
s
t
h
r
ee
in
ter
f
ac
e
s
tate
s
u
s
ed
in
t
h
i
s
w
o
r
k
:
Fi
g
u
r
e
2
(
a)
in
d
icate
s
t
h
at
th
e
p
al
m
is
s
tab
le
an
d
co
r
r
ec
tly
p
o
s
itio
n
ed
f
o
r
ca
p
tu
r
e;
Fig
u
r
e
2
(
b
)
w
ar
n
s
th
at
th
e
p
al
m
R
OI
is
to
o
s
m
a
ll
(
t
y
p
ical
l
y
w
h
e
n
th
e
u
s
er
is
to
o
f
ar
f
r
o
m
t
h
e
ca
m
er
a)
,
p
r
o
m
p
ti
n
g
t
h
e
u
s
er
to
m
o
v
e
clo
s
er
;
a
n
d
Fig
u
r
e
2
(
c)
in
d
icate
s
t
h
at
n
o
p
al
m
h
as
b
ee
n
d
etec
ted
in
t
h
e
c
u
r
r
en
t
v
ie
w
.
T
h
is
g
u
id
an
ce
m
ec
h
an
i
s
m
i
m
p
r
o
v
es
R
OI
co
n
s
is
te
n
c
y
an
d
r
ed
u
ce
s
lo
w
-
q
u
ali
t
y
s
a
m
p
le
s
ca
u
s
ed
b
y
m
o
tio
n
b
lu
r
,
i
n
co
r
r
ec
t
d
is
tan
ce
,
o
r
m
is
s
in
g
h
an
d
s
.
2
.
3
.
P
re
pro
ce
s
s
ing
a
nd
RO
I
ex
t
ra
ct
io
n
Af
ter
ca
p
t
u
r
e,
ea
ch
R
GB
f
r
am
e
i
s
s
ta
n
d
ar
d
ized
o
n
d
ev
ice
to
p
r
o
d
u
ce
a
co
n
s
is
ten
t
p
alm
R
OI
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
.
T
h
e
p
r
ep
r
o
ce
s
s
i
n
g
s
eq
u
e
n
ce
i
s
d
esi
g
n
ed
t
o
b
e
l
ig
h
t
w
eig
h
t
w
h
ile
p
r
eser
v
in
g
d
i
s
cr
i
m
i
n
ati
v
e
p
al
m
li
n
e
s
tr
u
ct
u
r
es.
F
ir
s
t,
t
h
e
in
p
u
t
is
co
n
v
er
ted
to
g
r
a
y
s
ca
le
an
d
co
n
tr
a
s
t
i
s
n
o
r
m
aliz
ed
u
s
i
n
g
co
n
tr
as
t
-
li
m
ited
ad
ap
tiv
e
h
i
s
to
g
r
a
m
e
q
u
aliza
tio
n
(
C
L
AHE
)
.
A
m
e
d
ian
f
il
ter
is
t
h
en
ap
p
lied
to
r
ed
u
ce
s
en
s
o
r
n
o
is
e
w
h
ile
m
ain
tain
in
g
ed
g
e
an
d
l
in
e
d
etails.
R
OI
lo
ca
lizatio
n
i
s
p
er
f
o
r
m
ed
u
s
i
n
g
a
li
g
h
t
w
ei
g
h
t
h
an
d
la
n
d
m
ar
k
d
etec
to
r
im
p
le
m
en
ted
in
Me
d
iaP
ip
e
[
1
5
]
.
T
h
e
d
etec
to
r
r
e
tu
r
n
s
2
1
h
an
d
k
e
y
p
o
in
t
s
;
a
p
o
ly
g
o
n
al
r
eg
io
n
is
co
n
s
tr
u
cted
f
r
o
m
a
s
u
b
s
et
o
f
s
tab
le
p
al
m
lan
d
m
ar
k
s
to
is
o
late
th
e
ce
n
tr
al
p
al
m
ar
ea
.
T
h
e
cr
o
p
p
ed
R
OI
is
r
esized
to
2
2
4
×2
2
4
an
d
in
te
n
s
it
y
-
n
o
r
m
alize
d
b
e
f
o
r
e
b
ein
g
p
ass
ed
to
th
e
n
e
u
r
al
f
ea
t
u
r
e
ex
tr
ac
to
r
.
Fig
u
r
e
3
s
h
o
w
s
a
n
ex
a
m
p
le
o
f
t
h
e
la
n
d
m
ar
k
o
u
tp
u
t
u
s
ed
to
co
m
p
u
te
t
h
e
R
OI
.
2
.
4
.
St
a
g
e
1
:
s
elf
-
s
up
er
v
is
ed
re
presenta
t
io
n lea
rning
(
t
ea
cher
m
o
del)
T
o
r
ed
u
ce
d
ep
en
d
en
c
y
o
n
lab
eled
b
io
m
etr
ic
d
ata,
th
e
f
r
am
e
w
o
r
k
tr
ain
s
a
h
ig
h
-
ca
p
ac
it
y
teac
h
er
n
et
w
o
r
k
u
s
i
n
g
s
el
f
-
s
u
p
er
v
i
s
ed
co
n
tr
asti
v
e
l
ea
r
n
in
g
.
A
R
esN
et
-
1
8
b
ac
k
b
o
n
e
is
ad
o
p
ted
as
th
e
teac
h
er
f
ea
t
u
r
e
ex
tr
ac
to
r
,
an
d
ea
ch
R
OI
s
a
m
p
le
is
tr
a
n
s
f
o
r
m
ed
i
n
to
t
w
o
a
u
g
m
e
n
ted
v
ie
w
s
(
e.
g
.
,
r
a
n
d
o
m
cr
o
p
,
m
ild
r
o
tatio
n
,
b
lu
r
/n
o
is
e)
to
f
o
r
m
a
p
o
s
itiv
e
p
air
.
T
h
e
o
b
j
ec
tiv
e
en
co
u
r
ag
es
v
ie
w
-
in
v
ar
ia
n
t
r
ep
r
esen
t
ati
o
n
s
b
y
m
ax
i
m
izin
g
ag
r
ee
m
e
n
t
b
et
w
ee
n
t
h
e
t
w
o
v
ie
w
s
w
h
ile
s
ep
ar
atin
g
t
h
e
m
f
r
o
m
o
th
er
s
a
m
p
les
i
n
th
e
b
atch
.
T
h
e
tr
ain
in
g
f
o
llo
w
s
a
Si
m
C
L
R
-
s
t
y
le
co
n
tr
asti
v
e
f
o
r
m
u
la
tio
n
u
s
in
g
t
h
e
n
o
r
m
alize
d
te
m
p
er
at
u
r
e
-
s
c
aled
cr
o
s
s
-
e
n
tr
o
p
y
(
NT
-
Xen
t)
lo
s
s
[
1
6
]
.
T
o
im
p
r
o
v
e
r
ep
r
esen
tatio
n
q
u
alit
y
f
o
r
co
n
tr
asti
v
e
lear
n
i
n
g
,
th
e
l
o
s
s
is
ap
p
lied
in
a
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
R
o
b
u
s
t p
a
lmp
r
in
t
b
io
metric so
lu
tio
n
fo
r
s
ec
u
r
e
mo
b
ile
a
u
th
en
tica
tio
n
(
S
o
n
N
g
u
ye
n
)
683
p
r
o
j
ec
tio
n
s
p
ac
e
p
r
o
d
u
ce
d
b
y
a
s
m
all
M
L
P
h
ea
d
attac
h
ed
to
th
e
b
ac
k
b
o
n
e,
co
n
s
is
te
n
t
w
it
h
s
tan
d
ar
d
Si
m
C
L
R
p
r
ac
tice.
T
h
e
tr
ain
in
g
is
g
u
id
ed
b
y
th
e
NT
-
Xen
t
lo
s
s
f
u
n
ct
io
n
(
1
)
,
w
h
ich
is
ce
n
tr
al
to
th
e
Si
m
C
L
R
f
r
a
m
e
w
o
r
k
[
1
6
]
.
T
h
e
o
b
j
ec
tiv
e
o
f
t
h
is
lo
s
s
i
s
to
m
ax
i
m
ize
t
h
e
s
i
m
ilar
it
y
b
et
w
ee
n
t
h
e
e
m
b
ed
d
in
g
s
o
f
th
e
p
o
s
iti
v
e
p
air
w
h
ile
s
i
m
u
lta
n
eo
u
s
l
y
m
i
n
i
m
iz
in
g
t
h
eir
s
i
m
ilar
it
y
to
all
n
eg
a
tiv
e
p
air
s
.
T
h
e
N
T
-
Xen
t
lo
s
s
f
o
r
a
p
o
s
itiv
e
p
air
(
,
)
is
f
o
r
m
u
lated
as:
,
=
−
l
og
ex
p
(
(
,
)
⁄
)
∑
1
[
≠
]
2
=
1
ex
p
(
(
,
)
⁄
)
(
1
)
w
h
er
e
a
n
d
ar
e
th
e
e
m
b
ed
d
in
g
v
ec
to
r
s
o
f
t
h
e
p
o
s
iti
v
e
p
air
,
(
⋅
)
is
t
h
e
co
s
i
n
e
s
i
m
ilar
i
t
y
f
u
n
c
tio
n
,
τ
i
s
a
te
m
p
er
atu
r
e
p
ar
a
m
eter
th
at
co
n
tr
o
ls
th
e
s
ep
ar
at
io
n
o
f
n
eg
ativ
e
s
a
m
p
le
s
,
an
d
N
is
th
e
b
atch
s
ize.
A
lo
w
er
te
m
p
er
atu
r
e
i
n
cr
ea
s
e
s
t
h
e
p
en
alt
y
f
o
r
h
ar
d
-
to
-
d
is
tin
g
u
is
h
n
e
g
ati
v
e
s
a
m
p
les,
f
o
r
ci
n
g
th
e
m
o
d
el
to
lear
n
m
o
r
e
d
is
cr
i
m
i
n
ati
v
e
f
ea
t
u
r
es.
Du
r
i
n
g
t
h
is
s
ta
g
e,
a
t
w
o
-
la
y
er
ML
P
p
r
o
j
ec
tio
n
h
ea
d
is
a
d
d
ed
to
th
e
R
esNet
-
18
en
co
d
er
.
(
a)
(
b
)
(
c)
Fig
u
r
e
2
.
E
x
a
m
p
les o
f
r
ea
l
-
ti
m
e
u
s
er
g
u
id
an
ce
i
n
th
e
m
o
b
il
e
ap
p
licatio
n
.
Fro
m
le
f
t to
r
ig
h
t (
a)
Su
cc
ess
f
u
l
d
etec
tio
n
w
ith
a
g
r
ee
n
o
v
er
la
y
in
d
icatin
g
t
h
e
p
al
m
i
s
s
tab
le
a
n
d
r
ea
d
y
(
b
)
A
w
ar
n
i
n
g
t
h
at
t
h
e
p
alm
ar
ea
is
to
o
s
m
al
l,
p
r
o
m
p
ti
n
g
t
h
e
u
s
er
to
m
o
v
e
clo
s
er
an
d
(
c)
A
n
o
ti
f
icat
i
o
n
th
at
n
o
p
al
m
h
a
s
b
ee
n
d
ete
cted
in
th
e
ca
m
er
a
'
s
v
ie
w
Fig
u
r
e
3
.
P
alm
la
n
d
m
ar
k
d
etec
tio
n
f
o
r
R
OI
e
x
tr
ac
tio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
6
8
0
-
689
684
2
.
5
.
St
a
g
e
2
:
k
no
w
ledg
e
dis
t
illa
t
io
n f
o
r
m
o
bil
e
dep
lo
y
m
e
nt
(
s
t
ud
ent
m
o
del)
Af
ter
tr
ain
i
n
g
t
h
e
teac
h
er
n
et
w
o
r
k
,
its
r
ep
r
esen
tatio
n
is
tr
an
s
f
er
r
ed
to
a
co
m
p
ac
t
s
tu
d
en
t
m
o
d
el
to
en
ab
le
lo
w
-
laten
c
y
i
n
f
er
en
ce
o
n
s
m
ar
tp
h
o
n
e
s
.
A
Mo
b
ileNet
V3
b
ac
k
b
o
n
e
is
ad
o
p
ted
as
t
h
e
s
tu
d
e
n
t
d
u
e
to
it
s
f
av
o
r
ab
le
ac
cu
r
ac
y
–
late
n
c
y
tr
ad
e
-
o
f
f
o
n
m
o
b
ile
h
ar
d
w
ar
e
[
1
9
]
.
T
h
e
s
tu
d
en
t
is
tr
ain
ed
to
m
i
m
ic
th
e
teac
h
er
e
m
b
ed
d
in
g
f
o
r
th
e
s
a
m
e
R
OI
in
p
u
t
u
s
i
n
g
a
d
is
till
a
tio
n
o
b
jectiv
e
th
a
t
alig
n
s
t
h
e
s
t
u
d
en
t
’
s
e
m
b
ed
d
in
g
s
p
ac
e
w
it
h
t
h
e
teac
h
er
’
s
e
m
b
ed
d
in
g
s
p
ac
e.
T
h
is
ap
p
r
o
ac
h
is
co
n
s
is
ten
t
w
it
h
co
m
m
o
n
d
ep
lo
y
m
en
t
p
r
ac
tice
w
h
er
e
d
is
till
atio
n
is
u
s
ed
to
p
r
eser
v
e
r
ec
o
g
n
itio
n
ca
p
ab
ilit
y
w
h
ile
r
ed
u
cin
g
m
o
d
el
s
ize
a
n
d
in
f
er
en
ce
co
s
t
[
2
1
]
.
I
n
ad
d
itio
n
,
lig
h
t
w
e
ig
h
t
ar
ch
itec
tu
r
al
p
r
in
cip
les
(
e.
g
.
,
g
en
er
ati
n
g
m
o
r
e
f
ea
t
u
r
es
u
s
in
g
in
e
x
p
en
s
i
v
e
o
p
er
atio
n
s
)
ca
n
b
e
in
co
r
p
o
r
ated
to
f
u
r
th
e
r
im
p
r
o
v
e
m
o
b
ile
e
f
f
ic
ien
c
y
w
h
e
n
n
ee
d
ed
[
2
0
]
.
I
f
ad
d
itio
n
al
co
m
p
r
ess
io
n
is
r
eq
u
ir
ed
,
in
teg
er
-
o
n
l
y
q
u
a
n
tiz
atio
n
ca
n
b
e
ap
p
lied
af
ter
tr
ai
n
in
g
;
h
o
w
e
v
er
,
q
u
an
t
izatio
n
s
h
o
u
ld
b
e
v
alid
ated
ca
r
ef
u
ll
y
b
ec
au
s
e
b
io
m
etr
ic
v
e
r
if
icatio
n
ac
c
u
r
ac
y
ca
n
b
e
s
en
s
itiv
e
to
e
m
b
ed
d
in
g
p
er
tu
r
b
atio
n
s
[
2
2
]
.
2
.
6
.
H
y
brid
m
a
t
chi
ng
f
o
r
v
e
rif
ica
t
io
n a
nd
s
ca
la
ble ident
i
f
ica
t
io
n
T
h
e
d
e
p
lo
y
ed
s
tu
d
e
n
t
m
o
d
el
o
u
tp
u
ts
a
f
i
x
ed
-
d
i
m
e
n
s
io
n
al
em
b
ed
d
i
n
g
(
2
5
6
-
D
in
th
i
s
w
o
r
k
)
f
o
r
ea
ch
R
OI
.
T
h
is
em
b
ed
d
in
g
s
ize
is
s
elec
ted
as
a
b
alan
ce
b
etw
ee
n
d
is
cr
im
i
n
at
iv
e
p
o
w
er
an
d
o
n
-
d
ev
ice
ef
f
ic
ien
c
y
.
Fo
r
lo
ca
l
au
th
e
n
ticatio
n
,
th
e
s
y
s
te
m
p
er
f
o
r
m
s
1
:1
v
er
i
f
icat
io
n
b
y
co
m
p
u
tin
g
co
s
i
n
e
s
i
m
ilar
it
y
b
et
w
ee
n
th
e
q
u
er
y
e
m
b
ed
d
in
g
an
d
th
e
en
r
o
lled
tem
p
late,
f
o
llo
w
ed
b
y
a
th
r
esh
o
ld
d
ec
is
io
n
.
Fo
r
ap
p
licatio
n
s
r
eq
u
ir
i
n
g
id
en
ti
f
icatio
n
a
m
o
n
g
m
an
y
e
n
r
o
lled
u
s
er
s
,
t
h
e
q
u
er
y
e
m
b
ed
d
in
g
is
tr
a
n
s
m
itted
to
a
cl
o
u
d
b
ac
k
en
d
th
at
p
er
f
o
r
m
s
A
NN
s
ea
r
c
h
o
v
er
th
e
e
m
b
ed
d
in
g
d
atab
ase.
B
ec
au
s
e
ex
h
a
u
s
ti
v
e
s
ea
r
ch
b
ec
o
m
es
in
e
f
f
ic
ien
t
f
o
r
lar
g
e
g
aller
ies,
t
h
e
b
ac
k
e
n
d
u
s
e
s
F
A
I
S
S
in
d
e
x
i
n
g
to
s
u
p
p
o
r
t
f
as
t
s
i
m
ilar
it
y
s
ea
r
ch
at
s
ca
le
[
2
3
]
.
T
h
e
d
esig
n
f
o
llo
w
s
estab
lis
h
ed
ANN
p
r
ac
tice
f
o
r
h
ig
h
-
d
i
m
en
s
io
n
al
r
etr
ie
v
al,
w
h
er
e
r
ec
all
–
laten
c
y
tr
ad
e
-
o
f
f
s
ar
e
tu
n
ed
to
m
ee
t
ap
p
licatio
n
co
n
s
tr
ai
n
t
s
[
2
5
]
.
Fo
r
v
er
y
lar
g
e
d
ep
lo
y
m
e
n
ts
,
GP
U
-
ac
ce
ler
ated
s
i
m
ilar
it
y
s
e
ar
ch
ca
n
b
e
u
s
ed
to
i
m
p
r
o
v
e
th
r
o
u
g
h
p
u
t
w
h
e
n
ap
p
r
o
p
r
iate
in
f
r
astru
ct
u
r
e
is
a
v
ail
ab
le
[
2
4
]
.
3.
M
E
T
H
O
D
T
h
is
s
ec
tio
n
d
escr
ib
es
t
h
e
ex
p
er
i
m
e
n
tal
d
esi
g
n
u
s
ed
to
ev
alu
a
te
th
e
p
r
o
p
o
s
ed
f
r
a
m
e
w
o
r
k
i
n
a
r
ep
r
o
d
u
cib
le
m
a
n
n
er
,
i
n
cl
u
d
in
g
t
h
e
d
ataset
u
s
a
g
e
co
n
s
tr
ai
n
t
,
tr
ain
in
g
co
n
f
ig
u
r
atio
n
,
e
v
al
u
atio
n
m
etr
ics,
a
n
d
th
e
m
ea
s
u
r
e
m
e
n
t p
r
o
to
co
l f
o
r
m
o
b
ile
la
ten
c
y
an
d
lar
g
e
-
s
ca
le
id
en
tific
atio
n
.
3
.
1
.
Da
t
a
s
et
a
nd
ex
peri
m
ent
a
l pro
t
o
co
l
A
ll
e
x
p
er
i
m
e
n
ts
i
n
th
is
s
t
u
d
y
w
er
e
co
n
d
u
cted
u
s
i
n
g
o
n
l
y
t
h
e
p
u
b
lic
Kag
g
le
d
ata
s
et
“
P
al
m
R
ec
o
g
n
itio
n
Data
s
et
f
o
r
A
u
th
en
ticatio
n
S
y
s
te
m
”
(
P
al
m
Dat
aset)
.
T
h
e
d
ataset
co
n
tain
s
ap
p
r
o
x
i
m
atel
y
1
2
,
0
0
0
p
al
m
i
m
a
g
es
a
n
d
is
d
is
tr
ib
u
t
ed
p
r
im
ar
il
y
i
n
T
I
FF
f
o
r
m
at.
B
ec
au
s
e
p
u
b
lic
b
io
m
etr
ic
d
atasets
o
f
te
n
lac
k
a
u
n
i
v
er
s
al
l
y
e
n
f
o
r
ce
d
p
r
o
t
o
co
l,
w
e
ad
o
p
t
a
co
n
s
is
te
n
t
e
n
r
o
ll
m
en
t
–
p
r
o
b
e
ev
alu
atio
n
s
tr
ateg
y
th
a
t
m
atc
h
es
r
ea
l
au
th
e
n
tica
tio
n
u
s
ag
e.
E
ac
h
id
e
n
tit
y
is
r
ep
r
esen
ted
b
y
m
u
ltip
l
e
s
a
m
p
les.
On
e
s
a
m
p
le
p
er
id
e
n
tit
y
is
s
elec
ted
as
th
e
en
r
o
ll
m
en
t
te
m
p
late,
w
h
ile
th
e
r
e
m
ain
i
n
g
s
a
m
p
le
s
ar
e
tr
ea
ted
as
p
r
o
b
e
attem
p
ts
.
Gen
u
i
n
e
co
m
p
ar
is
o
n
s
ar
e
f
o
r
m
ed
b
y
m
atc
h
in
g
ea
ch
p
r
o
b
e
to
its
en
r
o
lled
tem
p
late
o
f
th
e
s
a
m
e
id
en
tit
y
,
an
d
i
m
p
o
s
t
o
r
co
m
p
ar
is
o
n
s
ar
e
f
o
r
m
ed
b
y
m
atc
h
i
n
g
p
r
o
b
es a
g
ain
s
t
en
r
o
lled
te
m
p
lates
f
r
o
m
o
th
er
id
en
tit
ies.
T
h
is
p
r
o
d
u
ce
s
s
co
r
e
d
is
tr
ib
u
tio
n
s
r
eq
u
ir
ed
f
o
r
FAR
/F
R
R
an
al
y
s
i
s
an
d
E
E
R
co
m
p
u
tatio
n
.
T
h
e
s
a
m
e
id
e
n
tit
y
g
r
o
u
p
i
n
g
is
al
s
o
u
s
ed
f
o
r
r
ep
o
r
tin
g
id
en
ti
f
icatio
n
-
s
t
y
le
“
ac
c
u
r
ac
y
”
(
to
p
-
1
m
atc
h
o
f
a
p
r
o
b
e
a
g
ain
s
t
t
h
e
en
r
o
lled
g
aller
y
)
t
o
k
ee
p
th
e
r
ep
o
r
ted
ac
cu
r
ac
y
a
n
d
E
E
R
lo
g
icall
y
c
o
n
s
is
ten
t
w
i
th
i
n
a
s
i
n
g
le
p
r
o
to
co
l.
3
.
2
.
I
m
ple
m
e
nta
t
io
n
det
a
ils
A
ll
p
al
m
i
m
ag
e
s
w
er
e
p
r
o
ce
s
s
ed
u
s
i
n
g
t
h
e
s
a
m
e
o
n
-
d
ev
ice
p
r
ep
r
o
ce
s
s
in
g
an
d
R
OI
ex
tr
ac
tio
n
p
ip
elin
e
d
escr
ib
ed
in
Sect
io
n
2
.
E
ac
h
in
p
u
t
f
r
a
m
e
w
a
s
co
n
v
er
ted
to
g
r
a
y
s
ca
le,
e
n
h
an
ce
d
u
s
i
n
g
C
L
AHE
,
an
d
d
en
o
is
ed
u
s
i
n
g
a
m
ed
ia
n
f
ilter
.
A
lig
h
t
w
e
i
g
h
t
Me
d
iaP
ip
e
h
a
n
d
lan
d
m
ar
k
d
etec
to
r
w
a
s
th
e
n
ap
p
lied
to
o
b
tain
k
e
y
p
o
in
ts
f
o
r
co
n
s
tr
u
cti
n
g
a
p
o
ly
g
o
n
al
p
al
m
r
eg
io
n
,
w
h
ic
h
w
a
s
cr
o
p
p
ed
an
d
r
esized
t
o
2
2
4
×2
2
4
b
ef
o
r
e
f
ea
t
u
r
e
ex
tr
ac
tio
n
[
1
5
]
.
Fo
r
r
ep
r
esen
tatio
n
lear
n
i
n
g
,
a
R
e
s
N
et
-
1
8
teac
h
er
m
o
d
el
w
as
tr
ain
ed
u
s
i
n
g
a
Si
m
C
L
R
-
s
t
y
le
co
n
tr
asti
v
e
o
b
j
ec
tiv
e
w
i
th
NT
-
Xen
t
lo
s
s
,
w
h
er
e
t
w
o
au
g
m
e
n
ted
v
ie
w
s
w
er
e
g
e
n
er
ated
p
er
R
OI
an
d
o
p
tim
ized
to
m
ax
i
m
ize
ag
r
ee
m
en
t
f
o
r
p
o
s
itiv
e
p
air
s
w
h
ile
s
ep
ar
atin
g
n
eg
a
tiv
e
s
w
i
th
i
n
t
h
e
b
atch
[
1
6
]
.
Fo
r
m
o
b
ile
d
ep
lo
y
m
en
t,
th
e
lear
n
e
d
teac
h
er
r
e
p
r
esen
tatio
n
w
as
t
r
an
s
f
er
r
ed
to
a
lig
h
t
w
ei
g
h
t
M
o
b
ileNetV3
s
tu
d
en
t
n
et
w
o
r
k
[
1
9
]
u
s
i
n
g
k
n
o
w
led
g
e
d
is
till
atio
n
b
y
a
lig
n
i
n
g
s
t
u
d
en
t
e
m
b
ed
d
in
g
s
to
teac
h
er
e
m
b
ed
d
in
g
s
f
o
r
th
e
s
a
m
e
R
OI
in
p
u
t
(
e
m
b
ed
d
in
g
-
a
lig
n
m
e
n
t
lo
s
s
)
[
2
1
]
.
Un
less
o
th
er
w
is
e
s
tated
,
th
e
s
t
u
d
en
t
e
m
b
ed
d
in
g
d
i
m
e
n
s
io
n
w
a
s
f
i
x
ed
at
2
5
6
f
o
r
v
er
if
icati
o
n
an
d
id
en
ti
f
icatio
n
ex
p
er
i
m
e
n
ts
.
3
.
3
.
E
v
a
lua
t
i
o
n
m
et
rics
Ver
if
icatio
n
p
er
f
o
r
m
a
n
ce
is
q
u
an
t
if
ied
u
s
in
g
(
i)
au
th
e
n
ticati
o
n
ac
cu
r
ac
y
u
n
d
er
th
e
en
r
o
ll
m
en
t
–
p
r
o
b
e
id
en
ti
f
icatio
n
p
r
o
to
co
l
an
d
(
ii)
E
E
R
.
E
E
R
i
s
t
h
e
o
p
er
atin
g
p
o
in
t
w
h
er
e
th
e
f
a
ls
e
ac
ce
p
ta
n
c
e
r
ate
(
F
A
R
)
eq
u
a
ls
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
R
o
b
u
s
t p
a
lmp
r
in
t
b
io
metric so
lu
tio
n
fo
r
s
ec
u
r
e
mo
b
ile
a
u
th
en
tica
tio
n
(
S
o
n
N
g
u
ye
n
)
685
th
e
f
alse
r
ej
ec
tio
n
r
ate
(
FR
R
)
,
an
d
it
is
w
id
el
y
u
s
ed
in
b
io
m
etr
ic
v
er
i
f
icat
io
n
s
t
u
d
ies
[
1
]
.
Fo
r
co
m
p
lete
n
ess
,
F
A
R
a
n
d
F
R
R
ar
e
co
m
p
u
ted
b
y
s
w
ee
p
i
n
g
t
h
e
s
i
m
ilar
it
y
th
r
es
h
o
ld
o
v
er
th
e
g
e
n
u
i
n
e
a
n
d
i
m
p
o
s
to
r
s
co
r
e
d
is
tr
ib
u
tio
n
s
; E
E
R
i
s
th
e
t
h
r
es
h
o
ld
w
h
er
e
F
AR
an
d
FR
R
i
n
te
r
s
ec
t.
Fo
r
s
ca
lab
ilit
y
,
1
:N
id
en
tif
icat
io
n
p
er
f
o
r
m
a
n
ce
is
ev
al
u
ated
u
s
i
n
g
R
ec
all
@
1
,
d
ef
in
ed
as
th
e
f
r
ac
tio
n
o
f
p
r
o
b
es
w
h
o
s
e
co
r
r
ec
t
en
r
o
lled
id
en
tit
y
ap
p
ea
r
s
as
th
e
to
p
-
r
an
k
ed
r
etr
iev
al
r
esu
lt.
T
h
is
alig
n
s
w
it
h
s
tan
d
ar
d
id
en
ti
f
icatio
n
e
v
al
u
atio
n
p
r
ac
ti
ce
an
d
s
u
p
p
o
r
ts
a
p
r
ac
tical
in
t
er
p
r
etatio
n
o
f
s
y
s
te
m
b
e
h
av
io
r
at
s
ca
le
[
2
3
]
,
[
2
5
]
.
3.
4
.
M
o
bil
e
la
t
ency
a
nd
la
rg
e
-
s
ca
le
identif
ica
t
io
n
m
ea
s
ur
e
m
e
nt
T
o
ev
alu
ate
d
ep
lo
y
ab
ilit
y
,
on
-
d
ev
ice
p
er
f
o
r
m
a
n
ce
is
m
ea
s
u
r
ed
o
n
an
A
p
p
le
iP
h
o
n
e
1
3
(
A1
5
B
i
o
n
ic
ch
ip
s
et,
4
GB
R
A
M)
,
co
n
s
i
s
ten
t
w
ith
t
h
e
d
ep
lo
y
m
en
t
tar
g
et
d
escr
ib
ed
in
th
e
R
esu
lt
s
s
ec
tio
n
.
L
ate
n
c
y
i
s
r
ep
o
r
ted
as
en
d
-
to
-
en
d
r
u
n
ti
m
e
(
p
r
ep
r
o
ce
s
s
in
g
+
R
OI
e
x
tr
a
ctio
n
+
m
o
d
el
i
n
f
er
en
ce
+
s
i
m
ilar
it
y
m
atch
in
g
)
,
an
d
co
m
p
o
n
en
t
-
le
v
el
ti
m
i
n
g
is
also
r
ec
o
r
d
e
d
to
attr
ib
u
te
th
e
to
tal
co
s
t to
ea
ch
s
tag
e.
T
o
ev
alu
ate
lar
g
e
-
s
ca
le
id
en
ti
f
icatio
n
,
w
e
co
n
s
tr
u
ct
a
s
i
m
u
l
ated
g
aller
y
o
f
o
n
e
m
i
llio
n
e
m
b
ed
d
i
n
g
s
an
d
co
m
p
ar
e
ex
h
a
u
s
ti
v
e
s
ea
r
c
h
ag
ai
n
s
t
a
n
A
NN
in
d
e
x
.
F
A
I
SS
is
u
s
ed
as
th
e
v
ec
to
r
s
ea
r
c
h
b
ac
k
en
d
,
an
d
th
e
HNSW
in
d
ex
i
s
s
elec
ted
d
u
e
to
its
p
r
ac
tical
s
p
ee
d
–
r
ec
all
tr
a
d
e
-
o
f
f
in
m
ed
i
u
m
-
to
-
lar
g
e
s
ca
l
e
s
ettin
g
s
[
2
3
]
.
T
h
e
r
ep
o
r
ted
r
ec
all/laten
c
y
tr
ad
e
-
o
f
f
is
in
ter
p
r
eted
as
an
o
p
er
atio
n
al
d
esig
n
ch
o
ice,
w
h
er
e
a
s
m
all
lo
s
s
in
r
etr
iev
al
ac
cu
r
ac
y
ca
n
y
ield
lar
g
e
i
m
p
r
o
v
e
m
e
n
ts
i
n
q
u
er
y
ti
m
e,
co
n
s
i
s
ten
t
w
i
th
e
s
tab
lis
h
ed
A
NN
b
eh
av
io
r
[
2
5
]
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
ev
a
lu
ate
s
t
h
e
p
r
o
p
o
s
ed
en
d
-
to
-
e
n
d
p
al
m
p
r
i
n
t
a
u
th
e
n
tica
tio
n
f
r
a
m
e
w
o
r
k
i
n
te
r
m
s
o
f
(
i)
v
er
if
ica
tio
n
p
er
f
o
r
m
an
ce
(
ac
cu
r
ac
y
a
n
d
E
E
R
)
,
(
ii)
m
o
b
il
e
d
ep
lo
y
ab
ilit
y
(
e
n
d
-
to
-
en
d
l
aten
c
y
an
d
m
o
d
el
f
o
o
tp
r
in
t)
,
(
iii)
r
o
b
u
s
tn
ess
u
n
d
er
p
r
ac
tical
ca
p
tu
r
e
v
ar
iatio
n
s
,
an
d
(
iv
)
s
ca
lab
ili
t
y
f
o
r
lar
g
e
-
g
aller
y
id
en
ti
f
icatio
n
.
4
.
1
.
Aut
hentic
a
t
io
n
a
cc
ura
c
y
a
nd
er
ro
r
ra
t
es
T
ab
le
1
r
e
p
o
r
ts
v
er
if
icatio
n
p
er
f
o
r
m
an
ce
o
n
th
e
Ka
g
g
le
p
al
m
d
ataset
u
s
i
n
g
th
e
s
a
m
e
e
n
r
o
ll
m
en
t
–
p
r
o
b
e
p
r
o
t
o
co
l
f
o
r
b
o
th
th
e
s
u
p
er
v
is
ed
Mo
b
ileNetV3
b
aselin
e
an
d
t
h
e
p
r
o
p
o
s
ed
f
r
a
m
e
w
o
r
k
.
T
h
e
b
aselin
e
ac
h
iev
e
s
9
7
.
5
%
ac
cu
r
ac
y
w
it
h
1
.
8
0
%
E
E
R
,
w
h
ile
t
h
e
p
r
o
p
o
s
e
d
s
elf
-
s
u
p
er
v
is
ed
p
r
etr
ain
in
g
p
lu
s
d
is
till
ed
Mo
b
ileNetV3
i
m
p
r
o
v
es
ac
cu
r
ac
y
to
9
9
.
2
%
an
d
r
ed
u
ce
s
E
E
R
to
0
.
1
5
%.
T
h
is
co
r
r
esp
o
n
d
s
to
a
+1
.
7
p
er
ce
n
tag
e
-
p
o
in
t
g
ai
n
i
n
ac
c
u
r
ac
y
a
n
d
a
1
.
6
5
p
er
ce
n
tag
e
-
p
o
i
n
t
ab
s
o
l
u
te
r
ed
u
ctio
n
i
n
E
E
R
,
in
d
icati
n
g
th
at
th
e
lear
n
ed
e
m
b
ed
d
in
g
b
ec
o
m
e
s
m
o
r
e
d
is
cr
i
m
i
n
ati
v
e
an
d
s
tab
l
e
w
h
ile
r
e
m
a
in
i
n
g
s
u
itab
le
f
o
r
m
o
b
ile
d
ep
lo
y
m
e
n
t.
Fo
r
b
r
o
a
d
er
co
n
tex
t,
T
ab
le
1
also
in
clu
d
e
s
r
ep
r
esen
tati
v
e
r
esu
lt
s
r
ep
o
r
ted
in
r
ec
en
t
p
al
m
p
r
in
t
liter
atu
r
e
GL
G
A
Net
[
4
]
an
d
a
C
C
Net
r
es
u
lt
s
u
m
m
ar
ized
i
n
th
e
p
al
m
p
r
i
n
t
d
ee
p
-
lear
n
i
n
g
s
u
r
v
e
y
[
1
]
.
T
h
ese
liter
atu
r
e
v
al
u
es
ar
e
lis
ted
to
g
et
h
er
w
it
h
t
h
eir
o
r
ig
in
al
d
atasets
/p
r
o
to
co
ls
(
e.
g
.
,
T
o
n
g
j
i/P
o
ly
U
-
s
t
y
le
b
en
ch
m
ar
k
s
)
an
d
ar
e
p
r
o
v
id
ed
f
o
r
p
o
s
itio
n
in
g
r
ath
er
t
h
an
d
ir
ec
t
n
u
m
er
ical
co
m
p
ar
is
o
n
,
b
ec
au
s
e
p
al
m
p
r
in
t
p
er
f
o
r
m
a
n
ce
i
s
s
tr
o
n
g
l
y
a
f
f
ec
ted
b
y
d
ataset
c
h
ar
ac
t
er
is
tics
,
ca
p
tu
r
e
co
n
d
itio
n
s
,
an
d
ev
al
u
atio
n
p
r
o
to
co
ls
[
1
]
,
[
2
]
.
W
ith
in
th
is
f
r
a
m
i
n
g
,
th
e
co
n
tr
ib
u
tio
n
o
f
th
is
w
o
r
k
i
s
n
o
t
li
m
ited
to
r
ec
o
g
n
itio
n
ac
c
u
r
ac
y
;
i
t
is
t
h
e
d
e
m
o
n
s
tr
atio
n
o
f
a
n
e
n
d
-
to
-
en
d
,
s
m
ar
tp
h
o
n
e
-
o
r
ien
ted
p
ip
elin
e
th
at
co
u
p
l
es
lab
el
-
e
f
f
icie
n
t
r
ep
r
esen
tatio
n
lear
n
in
g
a
n
d
a
co
m
p
ac
t
d
ep
lo
y
ed
m
o
d
el
w
i
th
m
ea
s
u
r
ab
le
o
n
-
d
e
v
ice
late
n
c
y
a
n
d
a
s
ca
lab
le
id
en
ti
f
icatio
n
b
ac
k
e
n
d
,
w
h
ile
m
ai
n
tai
n
in
g
v
er
i
f
icatio
n
p
er
f
o
r
m
an
ce
w
it
h
i
n
th
e
h
i
g
h
-
ac
c
u
r
ac
y
r
eg
i
m
e
r
ep
o
r
ted
b
y
r
ec
e
n
t st
u
d
ies [
1
]
,
[
4
]
.
T
ab
le
1
.
C
o
m
p
ar
is
o
n
o
f
au
t
h
e
n
ticatio
n
ac
cu
r
ac
y
a
n
d
E
E
R
with
s
ta
te
-
of
-
th
e
-
ar
t
m
eth
o
d
s
M
e
t
h
o
d
T
r
a
i
n
i
n
g
D
a
t
a
se
t
A
c
c
u
r
a
c
y
(
%)
EER
(
%)
S
u
p
e
r
v
i
se
d
M
o
b
i
l
e
N
e
t
V
3
(
B
a
se
l
i
n
e
)
S
u
p
e
r
v
i
se
d
K
a
g
g
l
e
9
7
.
5
1
.
8
0
P
r
o
p
o
se
d
S
S
L
+
D
i
st
i
l
l
e
d
M
o
b
i
l
e
N
e
t
V
3
S
S
L
+
d
i
s
t
i
l
l
a
t
i
o
n
K
a
g
g
l
e
9
9
.
2
0
.
1
5
G
L
G
A
N
e
t
[
4
]
S
u
p
e
r
v
i
se
d
T
o
n
g
j
i
/
P
o
l
y
U
9
8
.
5
/
9
9
.
5
N
o
t
r
e
p
o
r
t
e
d
C
C
N
e
t
[
1
]
S
u
p
e
r
v
i
se
d
T
o
n
g
j
i
1
0
0
.
0
0
.
0
0
0
0
4
4
.
2
.
On
-
dev
ice
perf
o
r
m
a
nce
a
nd
deplo
y
m
ent
re
lev
a
nce
A
m
o
b
ile
b
io
m
etr
ic
s
y
s
te
m
m
u
s
t
b
e
ac
cu
r
ate
an
d
r
esp
o
n
s
i
v
e
u
n
d
er
s
m
ar
tp
h
o
n
e
co
n
s
tr
ai
n
t
s
.
T
a
b
le
2
r
ep
o
r
ts
a
f
u
l
l
en
d
-
to
-
e
n
d
late
n
c
y
o
f
8
7
.
0
m
s
o
n
iP
h
o
n
e
1
3
,
i
n
clu
d
i
n
g
p
r
ep
r
o
ce
s
s
in
g
/R
OI
e
x
tr
ac
tio
n
(
4
8
.
0
m
s
)
,
s
tu
d
e
n
t
in
f
er
en
ce
(
2
1
.
5
m
s
)
,
an
d
co
s
in
e
m
atc
h
i
n
g
(
1
7
.
5
m
s
)
.
I
m
p
o
r
tan
tl
y
,
th
is
m
ea
s
u
r
e
m
en
t
r
e
f
lects
t
h
e
co
m
p
le
te
p
ip
eli
n
e
r
at
h
er
th
a
n
m
o
d
el
in
f
er
en
ce
alo
n
e,
w
h
ich
b
etter
r
ep
r
esen
ts
r
ea
l
u
s
er
ex
p
er
ien
ce
in
m
o
b
ile
au
th
e
n
tica
tio
n
.
T
h
e
d
is
till
atio
n
s
ta
g
e
i
s
t
h
e
k
e
y
f
ac
to
r
en
ab
li
n
g
d
ep
lo
y
ab
ili
t
y
:
t
h
e
s
t
u
d
en
t
m
o
d
el
r
ed
u
ce
s
in
f
er
en
ce
ti
m
e
b
y
ap
p
r
o
x
i
m
atel
y
3
.
5
×
r
elativ
e
t
o
t
h
e
teac
h
er
(
7
5
.
2
m
s
t
o
2
1
.
5
m
s
)
an
d
r
ed
u
ce
s
t
h
e
m
o
d
el
f
o
o
tp
r
in
t
f
r
o
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
6
8
0
-
689
686
4
5
.
1
MB
to
1
2
.
4
MB
.
T
h
is
is
co
n
s
is
te
n
t
w
i
th
t
h
e
b
r
o
ad
er
ed
g
e
in
telli
g
e
n
ce
p
er
s
p
ec
tiv
e
t
h
at
m
o
d
el
co
m
p
r
es
s
io
n
an
d
s
y
s
te
m
-
lev
e
l
laten
c
y
ar
e
p
r
im
ar
y
f
ea
s
ib
il
it
y
co
n
s
tr
ai
n
ts
f
o
r
r
ea
l
-
w
o
r
ld
ed
g
e
A
I
,
n
o
t
o
n
l
y
ac
cu
r
ac
y
[
1
4
]
.
W
h
ile
s
o
m
e
p
al
m
p
r
in
t
s
t
u
d
ies
e
m
p
h
asize
r
e
co
g
n
itio
n
p
er
f
o
r
m
a
n
ce
[
4
]
o
r
in
ter
p
r
etab
ilit
y
[
7
]
,
m
an
y
d
o
n
o
t
r
ep
o
r
t
co
m
p
let
e
o
n
-
d
e
v
ice
ti
m
i
n
g
a
n
d
m
e
m
o
r
y
m
ea
s
u
r
e
m
en
ts
,
m
a
k
in
g
it
d
if
f
icu
lt
to
as
s
es
s
d
ep
lo
y
m
en
t
r
ea
d
in
es
s
.
B
y
r
ep
o
r
tin
g
en
d
-
to
-
e
n
d
r
u
n
ti
m
e
w
i
th
e
x
p
licit
m
o
d
u
le
b
r
ea
k
d
o
w
n
,
t
h
is
w
o
r
k
co
n
tr
ib
u
tes
o
p
er
atio
n
al
ev
id
e
n
ce
t
h
at
t
h
e
p
r
o
p
o
s
ed
f
r
a
m
e
w
o
r
k
ca
n
b
e
in
te
g
r
ated
i
n
to
a
p
r
ac
tical
m
o
b
il
e
ap
p
licatio
n
.
T
ab
le
2
.
On
-
d
ev
ice
p
er
f
o
r
m
a
n
ce
ev
alu
at
io
n
o
n
iP
h
o
n
e
1
3
C
o
mp
o
n
e
n
t
L
a
t
e
n
c
y
(
ms)
M
o
d
e
l
S
i
z
e
(
M
B
)
P
r
e
p
r
o
c
e
ssi
n
g
P
i
p
e
l
i
n
e
(
i
n
c
l
.
R
O
I
Ex
t
r
a
c
t
i
o
n
)
4
8
.
0
-
R
e
sN
e
t
-
1
8
T
e
a
c
h
e
r
(
T
h
e
o
r
e
t
i
c
a
l
I
n
f
e
r
e
n
c
e
)
7
5
.
2
4
5
.
1
M
o
b
i
l
e
N
e
t
V
3
S
t
u
d
e
n
t
(
O
n
-
D
e
v
i
c
e
I
n
f
e
r
e
n
c
e
)
2
1
.
5
1
2
.
4
C
o
si
n
e
S
i
mi
l
a
r
i
t
y
M
a
t
c
h
i
n
g
1
7
.
5
-
F
u
l
l
En
d
-
to
-
En
d
A
u
t
h
e
n
t
i
c
a
t
i
o
n
8
7
.
0
-
4
.
3
.
Ro
bu
s
t
nes
s
t
o
re
a
l
-
w
o
rl
d c
o
nd
it
io
ns
T
ab
le
3
ev
alu
ates r
o
b
u
s
tn
e
s
s
u
n
d
er
v
ar
iatio
n
s
t
h
at
f
r
eq
u
e
n
tl
y
o
cc
u
r
i
n
s
m
ar
tp
h
o
n
e
ca
p
tu
r
e:
lo
w
l
ig
h
t,
p
ar
tial
o
cc
lu
s
io
n
,
a
n
d
m
ild
r
o
tatio
n
.
T
h
e
p
r
o
p
o
s
ed
SS
L
m
o
d
el
co
n
s
i
s
te
n
tl
y
o
u
tp
er
f
o
r
m
s
t
h
e
s
u
p
er
v
is
ed
b
aselin
e
u
n
d
er
all
s
tr
ess
co
n
d
i
tio
n
s
,
w
i
th
p
ar
ticu
lar
l
y
n
o
tab
le
g
ai
n
s
u
n
d
er
p
ar
tial o
cc
lu
s
io
n
an
d
lo
w
l
ig
h
t.
T
h
i
s
b
eh
av
io
r
ali
g
n
s
w
it
h
t
h
e
m
o
t
i
v
atio
n
o
f
co
n
tr
as
tiv
e
lear
n
i
n
g
,
w
h
ic
h
lear
n
s
i
n
v
ar
ia
n
ce
s
t
h
r
o
u
g
h
au
g
m
e
n
tatio
n
an
d
en
co
u
r
a
g
es r
ep
r
esen
tat
io
n
s
th
at
r
e
m
ain
s
tab
le
u
n
d
er
p
er
tu
r
b
atio
n
s
t
h
at
r
ese
m
b
le
r
ea
l c
ap
tu
r
e
n
o
is
e
[
1
6
]
.
Fro
m
a
liter
atu
r
e
p
er
s
p
ec
ti
v
e,
R
OI
q
u
al
it
y
i
s
w
id
el
y
r
ec
o
g
n
ized
as
a
d
o
m
i
n
an
t
f
ac
to
r
in
u
n
co
n
s
tr
ai
n
ed
p
al
m
p
r
in
t
r
ec
o
g
n
itio
n
,
an
d
r
ec
en
t
R
OI
r
e
s
ea
r
ch
ex
p
licitl
y
tar
g
et
s
r
o
b
u
s
t
n
es
s
u
n
d
er
less
co
n
tr
o
lled
ac
q
u
is
itio
n
[
3
]
.
I
n
o
u
r
s
y
s
te
m
,
r
o
b
u
s
t
n
es
s
i
m
p
r
o
v
e
m
e
n
t
s
ar
is
e
f
r
o
m
t
w
o
co
u
p
led
d
esig
n
c
h
o
ices:
g
u
id
ed
ac
q
u
is
it
io
n
(
Fig
u
r
e
2
)
r
ed
u
ce
s
s
ev
er
e
ca
p
tu
r
e
e
r
r
o
r
s
b
ef
o
r
e
in
f
er
en
ce
,
an
d
lan
d
m
ar
k
-
b
ased
R
OI
ex
tr
ac
tio
n
(
Fig
u
r
e
3
)
p
r
o
v
id
es
co
n
s
is
te
n
t
p
al
m
al
ig
n
m
e
n
t
at
lo
w
r
u
n
ti
m
e
co
s
t.
W
h
ile
o
u
r
R
OI
m
eth
o
d
is
d
esig
n
ed
to
b
e
lig
h
t
w
e
ig
h
t
f
o
r
m
o
b
ile
d
ep
lo
y
m
e
n
t,
th
e
r
es
u
l
ts
in
T
ab
le
3
s
u
g
g
est
th
at
co
m
b
in
i
n
g
s
tab
le
R
OI
ex
tr
ac
tio
n
w
i
th
S
S
L
is
a
p
r
ac
tical
p
ath
to
i
m
p
r
o
v
ed
r
o
b
u
s
tn
e
s
s
w
it
h
o
u
t i
n
cr
ea
s
i
n
g
o
n
-
d
ev
ic
e
co
m
p
u
tatio
n
.
T
ab
le
3
.
R
o
b
u
s
tn
es
s
ev
al
u
atio
n
u
n
d
er
ch
alle
n
g
in
g
co
n
d
itio
n
s
ac
cu
r
ac
y
C
o
n
d
i
t
i
o
n
S
u
p
e
r
v
i
se
d
M
o
b
i
l
e
N
e
t
V
3
(
%)
P
r
o
p
o
se
d
S
S
L
mo
d
e
l
(
%)
N
o
r
mal
l
i
g
h
t
i
n
g
9
7
.
5
9
9
.
2
L
o
w
l
i
g
h
t
9
4
.
1
9
7
.
8
P
a
r
t
i
a
l
o
c
c
l
u
si
o
n
9
2
.
7
9
7
.
8
R
o
t
a
t
i
o
n
(
±
1
5
d
e
g
r
e
e
s)
9
5
.
2
9
8
.
5
4
.
4
.
Sca
la
bil
it
y
f
o
r
la
r
g
e
-
s
ca
le
identif
ica
t
io
n
T
ab
le
4
ev
alu
ates
t
h
e
s
y
s
te
m
in
a
lar
g
e
-
s
ca
le
id
e
n
ti
f
icatio
n
s
etti
n
g
u
s
i
n
g
a
o
n
e
m
ill
io
n
e
m
b
ed
d
i
n
g
g
aller
y
.
E
x
h
au
s
ti
v
e
s
ea
r
c
h
r
e
q
u
ir
es
ap
p
r
o
x
i
m
ate
l
y
3
5
4
0
m
s
p
er
q
u
er
y
,
w
h
ic
h
is
n
o
t
f
ea
s
ib
le
f
o
r
i
n
ter
ac
ti
v
e
id
en
ti
f
icatio
n
.
I
n
co
n
tr
ast,
F
AI
SS
w
it
h
an
HNSW
in
d
ex
r
ed
u
ce
s
laten
c
y
to
ap
p
r
o
x
i
m
atel
y
3
2
m
s
p
er
q
u
er
y
w
h
ile
m
ai
n
tai
n
in
g
9
9
.
5
%
R
e
ca
ll@
1
.
T
h
is
is
co
n
s
is
ten
t
w
i
th
t
h
e
f
u
n
d
a
m
e
n
tal
A
NN
tr
ad
e
-
o
f
f
:
a
s
m
all
a
n
d
tu
n
ab
le
r
ed
u
ct
io
n
i
n
ex
a
ctn
es
s
ca
n
y
ield
o
r
d
er
s
-
of
-
m
a
g
n
it
u
d
e
s
p
ee
d
u
p
s
f
o
r
h
ig
h
-
d
i
m
en
s
io
n
a
l
r
etr
iev
al
[
2
3
]
,
[
2
5
]
.
P
r
i
o
r
w
o
r
k
also
s
h
o
w
s
t
h
at
s
i
m
ilar
it
y
s
ea
r
ch
th
r
o
u
g
h
p
u
t
ca
n
b
e
f
u
r
t
h
er
i
m
p
r
o
v
ed
w
it
h
GP
U
ac
ce
ler
atio
n
w
h
e
n
i
n
f
r
astru
ct
u
r
e
p
er
m
its
,
r
ei
n
f
o
r
c
in
g
th
at
s
ca
lab
le
id
en
tific
at
io
n
is
a
tr
ac
tab
le
co
m
p
o
n
e
n
t o
f
r
ea
l d
ep
lo
y
m
en
t
s
[
2
4
]
.
T
h
is
s
ca
lab
ilit
y
r
esu
lt
is
a
k
e
y
p
ar
t
o
f
th
e
co
n
tr
ib
u
tio
n
b
ec
au
s
e
m
a
n
y
p
al
m
p
r
i
n
t
p
ap
er
s
f
o
cu
s
p
r
im
ar
il
y
o
n
v
er
i
f
icat
io
n
p
er
f
o
r
m
an
ce
w
it
h
o
u
t
d
e
m
o
n
s
tr
atin
g
s
y
s
te
m
b
eh
a
v
io
r
w
h
e
n
th
e
g
a
l
ler
y
s
ize
in
cr
ea
s
e
s
s
u
b
s
ta
n
tial
l
y
.
B
y
i
n
teg
r
at
in
g
A
N
N
-
b
ased
r
etr
ie
v
al
i
n
to
t
h
e
ar
ch
itect
u
r
e
an
d
q
u
an
t
if
y
i
n
g
laten
c
y
/r
ec
all
at
a
r
ea
lis
tic
s
ca
le,
t
h
e
p
r
o
p
o
s
ed
f
r
a
m
e
w
o
r
k
m
o
v
es
b
e
y
o
n
d
ac
cu
r
ac
y
-
o
n
l
y
ev
a
lu
at
io
n
to
w
ar
d
d
ep
lo
y
m
e
n
t
-
g
r
ad
e
id
en
ti
f
icatio
n
ca
p
ab
ilit
y
.
T
ab
le
4
.
Scalab
ilit
y
p
er
f
o
r
m
an
ce
f
o
r
1
:N
I
d
en
tif
icatio
n
(
1
m
i
llio
n
g
al
ler
y
s
ize)
S
e
a
r
c
h
me
t
h
od
S
e
a
r
c
h
l
a
t
e
n
c
y
(
ms p
e
r
q
u
e
r
y
)
R
e
c
a
l
l
@
1
(
%)
B
r
u
t
e
-
f
o
r
c
e
(
e
x
h
a
u
st
i
v
e
)
~
3
5
4
0
1
0
0
.
0
F
A
I
S
S
(
H
N
S
W
I
n
d
e
x
)
~
3
2
9
9
.
5
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
R
o
b
u
s
t p
a
lmp
r
in
t
b
io
metric so
lu
tio
n
fo
r
s
ec
u
r
e
mo
b
ile
a
u
th
en
tica
tio
n
(
S
o
n
N
g
u
ye
n
)
687
4.
5
.
Dis
cus
s
io
n,
li
m
it
a
t
io
ns
,
a
nd
f
uture
w
o
rk
T
ak
en
to
g
eth
er
,
T
ab
les
1
-
4
d
em
o
n
s
tr
ate
t
h
at
t
h
e
f
r
a
m
e
wo
r
k
ac
h
iev
e
s
a
b
alan
ce
d
p
r
o
f
ile
ac
r
o
s
s
ac
cu
r
ac
y
,
r
o
b
u
s
t
n
es
s
,
m
o
b
ile
ef
f
icien
c
y
,
a
n
d
s
ca
lab
ilit
y
.
T
h
e
p
r
i
m
ar
y
co
n
tr
ib
u
tio
n
r
elativ
e
to
r
ep
r
esen
tativ
e
p
al
m
p
r
in
t
s
t
u
d
ies
is
n
o
t
m
er
e
l
y
a
m
ar
g
i
n
al
m
etr
ic
g
ai
n
,
b
u
t
an
en
d
-
to
-
en
d
d
esi
g
n
t
h
at
in
teg
r
ate
s
(
i)
lab
el
-
ef
f
icien
t
tr
ain
in
g
v
ia
SSL
[
1
6
]
,
(
ii)
m
o
b
ile
-
f
r
ie
n
d
l
y
in
f
er
e
n
ce
v
ia
d
is
til
latio
n
(
T
ab
le
2
)
,
an
d
(
iii)
s
ca
lab
le
id
en
ti
f
icatio
n
v
ia
A
N
N
in
d
ex
i
n
g
[
2
3
]
,
[
2
5
]
.
T
h
is
in
teg
r
atio
n
is
r
ar
ely
r
ep
o
r
ted
as
a
s
in
g
le
v
alid
ated
p
ip
elin
e
in
p
r
io
r
p
alm
p
r
in
t
w
o
r
k
,
w
h
ic
h
ten
d
s
to
e
m
p
h
asize
eit
h
er
r
ec
o
g
n
itio
n
m
o
d
eli
n
g
[
4
]
,
R
OI
r
o
b
u
s
tn
es
s
[
3
]
,
o
r
ef
f
icien
c
y
-
o
r
ien
ted
r
ep
r
esen
tat
io
n
s
[
5
]
.
Nev
er
th
e
less
,
li
m
itat
io
n
s
r
e
m
ai
n
.
First,
th
e
Kag
g
le
d
ata
s
et
r
ef
lects
m
o
s
tl
y
co
n
tr
o
lled
o
r
s
em
i
-
co
n
tr
o
lled
ac
q
u
is
itio
n
,
s
o
g
e
n
e
r
aliza
tio
n
to
f
u
ll
y
u
n
co
n
s
tr
ai
n
ed
o
u
td
o
o
r
s
ettin
g
s
an
d
cr
o
s
s
-
d
ev
ice
v
ar
iab
ilit
y
i
s
n
o
t
y
et
estab
lis
h
ed
;
R
OI
s
tu
d
ies
in
d
icate
th
i
s
r
e
m
ai
n
s
a
c
o
r
e
ch
allen
g
e
in
u
n
co
n
s
tr
ai
n
ed
p
al
m
p
r
in
t
r
ec
o
g
n
itio
n
[
3
]
.
Seco
n
d
,
th
e
cu
r
r
en
t
s
y
s
te
m
d
o
es
n
o
t
i
n
clu
d
e
an
ex
p
licit
li
v
e
n
ess
o
r
P
A
D
co
m
p
o
n
e
n
t.
T
h
is
is
i
m
p
o
r
tan
t
b
ec
au
s
e
r
ec
en
t
w
o
r
k
s
d
e
m
o
n
s
tr
ate
b
o
th
th
e
f
ea
s
i
b
ilit
y
o
f
d
etec
tin
g
d
ee
p
f
ak
e
p
al
m
i
m
a
g
er
y
[
9
]
an
d
th
e
in
cr
ea
s
i
n
g
r
ea
lis
m
o
f
g
e
n
e
r
ated
p
alm
p
r
in
ts
u
s
i
n
g
d
if
f
u
s
i
o
n
m
o
d
els
[
1
0
]
,
w
h
ile
b
r
o
ad
er
b
io
m
etr
ic
s
ec
u
r
it
y
r
esear
ch
s
h
o
w
s
t
h
at
co
m
b
in
e
d
attac
k
s
ca
n
tar
g
et
b
o
th
r
ec
o
g
n
itio
n
a
n
d
P
A
D
m
o
d
u
le
s
[
1
1
]
.
T
h
ir
d
,
tem
p
late
p
r
o
tectio
n
is
n
o
t
y
et
i
m
p
le
m
e
n
ted
;
ca
n
ce
llab
le
te
m
p
lates
p
r
o
v
id
e
a
co
n
cr
ete
d
ir
ec
tio
n
f
o
r
r
ev
o
ca
tio
n
an
d
r
e
-
is
s
u
an
ce
i
f
te
m
p
lates
ar
e
co
m
p
r
o
m
i
s
ed
[
1
2
]
.
Fin
all
y
,
p
r
iv
ac
y
-
p
r
eser
v
i
n
g
m
o
d
el
u
p
d
ates
ar
e
n
o
t
y
et
ex
p
lo
r
ed
in
th
is
i
m
p
le
m
e
n
tat
io
n
;
f
ed
er
ated
lear
n
in
g
is
a
r
elev
a
n
t
p
ath
w
a
y
f
o
r
r
ed
u
cin
g
ce
n
tr
ali
ze
d
r
aw
-
b
io
m
e
tr
ic
ex
p
o
s
u
r
e
d
u
r
i
n
g
u
p
d
ates [
1
3
]
.
5.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
p
r
esen
ted
a
s
m
ar
tp
h
o
n
e
-
o
r
ien
ted
p
al
m
p
r
in
t
au
th
en
ticatio
n
f
r
a
m
e
w
o
r
k
t
h
at
in
te
g
r
ates
g
u
id
ed
m
o
b
ile
ac
q
u
is
itio
n
,
l
an
d
m
ar
k
-
b
ased
R
OI
ex
tr
ac
ti
o
n
,
co
m
p
ac
t
e
m
b
ed
d
in
g
in
f
er
en
ce
,
an
d
h
y
b
r
id
m
atc
h
in
g
f
o
r
b
o
th
o
n
-
d
ev
ice
v
er
if
icatio
n
a
n
d
lar
g
e
-
s
ca
le
id
en
ti
f
icatio
n
.
T
h
e
r
esu
lts
s
h
o
w
t
h
at
co
m
b
in
in
g
s
el
f
-
s
u
p
er
v
i
s
ed
co
n
tr
asti
v
e
p
r
etr
ain
in
g
w
it
h
k
n
o
w
led
g
e
d
is
till
ati
o
n
ca
n
y
ield
a
s
tr
o
n
g
p
alm
r
e
p
r
esen
tatio
n
w
h
ile
m
ee
ti
n
g
p
r
ac
tical
m
o
b
ile
co
n
s
tr
ain
ts
,
an
d
t
h
at
A
NN
-
b
ased
r
etr
iev
al
u
s
in
g
F
A
I
S
S
p
r
o
v
id
es
a
f
ea
s
ib
le
p
ath
to
lo
w
-
late
n
c
y
id
e
n
ti
f
icatio
n
w
h
e
n
th
e
g
aller
y
s
ize
b
ec
o
m
es lar
g
e.
Sev
er
al
li
m
ita
tio
n
s
s
h
o
u
l
d
b
e
ac
k
n
o
w
led
g
ed
.
First,
t
h
e
e
x
p
er
im
e
n
tal
e
v
al
u
atio
n
r
elies
o
n
a
p
u
b
lic
d
ataset
t
h
at
r
ef
lects
m
o
s
tl
y
co
n
tr
o
lled
o
r
s
e
m
i
-
co
n
tr
o
ll
ed
ca
p
tu
r
e
co
n
d
itio
n
s
;
t
h
er
e
f
o
r
e,
th
e
r
ep
o
r
ted
p
er
f
o
r
m
a
n
ce
m
a
y
n
o
t
f
u
ll
y
r
e
p
r
esen
t
m
o
r
e
ch
al
len
g
i
n
g
r
ea
l
-
w
o
r
ld
s
ce
n
ar
io
s
s
u
c
h
a
s
o
u
td
o
o
r
lig
h
ti
n
g
,
s
e
v
er
e
m
o
tio
n
b
lu
r
,
lar
g
e
p
o
s
e
v
ar
iatio
n
,
an
d
cr
o
s
s
-
d
ev
ice
ca
m
er
a
d
if
f
er
e
n
ce
s
.
Seco
n
d
,
th
e
cu
r
r
e
n
t
s
y
s
te
m
d
o
es
n
o
t
in
cl
u
d
e
an
ex
p
licit
liv
e
n
es
s
o
r
P
A
D
m
o
d
u
le,
w
h
ich
is
i
m
p
o
r
t
an
t
f
o
r
s
ec
u
r
it
y
b
ec
a
u
s
e
s
y
n
t
h
etic
o
r
m
an
ip
u
lated
p
al
m
i
m
a
g
er
y
is
b
ec
o
m
i
n
g
i
n
cr
ea
s
i
n
g
l
y
r
ea
lis
tic.
T
h
ir
d
,
t
e
m
p
late
p
r
o
tectio
n
an
d
r
ev
o
c
ab
ilit
y
ar
e
n
o
t
y
e
t
i
m
p
le
m
en
ted
,
w
h
ic
h
i
s
a
r
ele
v
an
t
co
n
s
id
er
atio
n
f
o
r
an
y
s
y
s
te
m
t
h
at
s
to
r
es
e
m
b
ed
d
in
g
s
f
o
r
au
th
e
n
ticatio
n
o
r
id
en
ti
f
icatio
n
.
Fu
t
u
r
e
w
o
r
k
w
ill
th
er
e
f
o
r
e
f
o
cu
s
o
n
(
1
)
b
r
o
ad
er
v
alid
atio
n
u
n
d
er
u
n
co
n
s
tr
ain
ed
a
n
d
cr
o
s
s
-
d
ev
ice
ca
p
tu
r
e
s
ettin
g
s
w
i
th
a
co
n
s
is
ten
t
p
r
o
to
co
l,
(
2
)
in
teg
r
ati
n
g
a
li
g
h
t
w
eig
h
t
an
ti
-
s
p
o
o
f
i
n
g
co
m
p
o
n
e
n
t
a
n
d
q
u
an
ti
f
y
i
n
g
its
s
ec
u
r
it
y
–
late
n
c
y
tr
ad
e
-
o
f
f
,
an
d
(
3
)
s
tr
en
g
th
e
n
i
n
g
te
m
p
late
s
ec
u
r
it
y
t
h
r
o
u
g
h
ca
n
ce
l
lab
le
te
m
p
lates
a
n
d
p
r
iv
ac
y
-
p
r
eser
v
in
g
m
o
d
el
u
p
d
ate
s
tr
ateg
ie
s
,
w
h
ile
al
s
o
ex
p
lo
r
in
g
ad
d
itio
n
a
l
co
m
p
r
ess
io
n
(
e.
g
.
,
q
u
an
tizat
io
n
)
u
n
d
er
ca
r
ef
u
l
v
e
r
if
icatio
n
s
tab
ilit
y
ch
ec
k
s
.
ACK
NO
WL
E
D
G
M
E
NT
S
I
w
o
u
ld
lik
e
to
ex
p
r
ess
m
y
s
in
ce
r
e
g
r
atit
u
d
e
to
Dr
.
B
ell
o
Mu
s
a
Yak
u
b
u
f
o
r
h
is
v
al
u
ab
le
tim
e,
in
s
i
g
h
t
f
u
l
co
m
m
en
t
s
,
an
d
co
n
s
tr
u
cti
v
e
f
ee
d
b
ac
k
i
n
r
ev
ie
w
i
n
g
th
is
j
o
u
r
n
al.
His
g
u
id
an
ce
g
r
e
atl
y
co
n
tr
ib
u
ted
to
i
m
p
r
o
v
i
n
g
t
h
e
clar
it
y
a
n
d
q
u
al
it
y
o
f
t
h
is
w
o
r
k
.
F
UNDIN
G
I
NF
O
RM
AT
I
O
N
T
h
is
r
esear
ch
w
a
s
s
u
p
p
o
r
ted
b
y
a
s
c
h
o
lar
s
h
ip
f
r
o
m
t
h
e
I
n
t
er
n
atio
n
al
Stu
d
e
n
ts
Sch
o
lar
s
h
ip
,
Facu
lt
y
o
f
Scien
ce
,
C
h
u
lalo
n
g
k
o
r
n
U
n
iv
er
s
it
y
,
B
an
g
k
o
k
,
T
h
ailan
d
.
AUTHO
R
CO
NT
RIB
UT
I
O
NS ST
A
T
E
M
E
NT
T
h
e
s
p
ec
if
ic
co
n
tr
ib
u
tio
n
s
o
f
ea
c
h
a
u
th
o
r
to
t
h
i
s
r
esear
ch
ar
e
o
u
t
lin
ed
in
th
e
f
o
llo
w
i
n
g
tab
le,
f
o
llo
w
in
g
t
h
e
C
R
ed
iT
(
C
o
n
tr
ib
u
to
r
R
o
les T
ax
o
n
o
m
y
)
f
r
a
m
e
w
o
r
k
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
6
8
0
-
689
688
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
So
n
Ng
u
y
en
A
r
t
h
o
r
n
L
u
a
n
g
s
o
d
s
ai
P
attar
asin
ee
B
h
attar
ak
o
s
o
l
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
si
s
I
:
I
n
v
e
st
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
si
o
n
P
:
P
r
o
j
e
c
t
a
d
m
i
n
i
s
t
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
s
i
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
t
h
o
r
s
s
tate
n
o
co
n
f
lic
t o
f
i
n
t
er
est
.
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
i
n
g
s
o
f
th
i
s
s
t
u
d
y
ar
e
o
p
en
l
y
a
v
ailab
le
i
n
P
al
m
R
ec
o
g
n
itio
n
Data
s
et
f
o
r
Au
t
h
en
t
icatio
n
S
y
s
te
m
a
t
h
ttp
s
://
www
.
k
a
g
g
le.
co
m
/d
ataset
s
/
s
aq
ib
s
h
o
aib
d
z/p
al
m
-
d
ataset/d
a
ta.
RE
F
E
R
E
NC
E
S
[
1
]
C
.
G
a
o
,
Z
.
Y
a
n
g
,
W
.
Ji
a
,
L
.
L
e
n
g
,
B
.
Z
h
a
n
g
,
a
n
d
A
.
B
.
J.
T
e
o
h
,
“
D
e
e
p
l
e
a
r
n
i
n
g
i
n
p
a
l
mp
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
-
a
c
o
m
p
r
e
h
e
n
si
v
e
su
r
v
e
y
,
”
O
c
t
.
2
0
2
5
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s
/
2
5
0
1
.
0
1
1
6
6
[
2
]
S
.
Z
h
a
o
,
L
.
F
e
i
,
a
n
d
J.
W
e
n
,
“
M
u
l
t
i
v
i
e
w
-
l
e
a
r
n
i
n
g
-
b
a
se
d
g
e
n
e
r
i
c
p
a
l
mp
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
:
a
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
,
”
M
a
t
h
e
m
a
t
i
c
s
,
v
o
l
.
1
1
,
n
o
.
5
,
p
.
1
2
6
1
,
M
a
r
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
m
a
t
h
1
1
0
5
1
2
6
1
.
[
3
]
L
.
S
u
,
L
.
F
e
i
,
B
.
Z
h
a
n
g
,
S
.
Z
h
a
o
,
J.
W
e
n
,
a
n
d
Y
.
X
u
,
“
C
o
mp
l
e
t
e
r
e
g
i
o
n
o
f
i
n
t
e
r
e
st
f
o
r
u
n
c
o
n
st
r
a
i
n
e
d
p
a
l
mp
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
,
”
I
EE
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
m
a
g
e
Pro
c
e
ss
i
n
g
,
v
o
l
.
3
3
,
p
p
.
3
6
6
2
–
3
6
7
5
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
T
I
P
.
2
0
2
4
.
3
4
0
7
6
6
6
.
[
4
]
K
.
Z
h
a
n
g
,
G
.
X
u
,
Y
.
K
.
Ji
n
,
G
.
Q
i
,
X
.
Y
a
n
g
,
a
n
d
L
.
B
a
i
,
“
P
a
l
mp
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
b
a
se
d
o
n
g
a
t
i
n
g
me
c
h
a
n
i
sm
a
n
d
a
d
a
p
t
i
v
e
f
e
a
t
u
r
e
f
u
si
o
n
,
”
Fro
n
t
i
e
r
s i
n
N
e
u
r
o
ro
b
o
t
i
c
s
,
v
o
l
.
1
7
,
M
a
y
2
0
2
3
,
d
o
i
:
1
0
.
3
3
8
9
/
f
n
b
o
t
.
2
0
2
3
.
1
2
0
3
9
6
2
.
[
5
]
H
.
S
h
a
o
,
D
.
Z
h
o
n
g
,
a
n
d
X
.
D
u
,
“
Ef
f
i
c
i
e
n
t
d
e
e
p
p
a
l
mp
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
v
i
a
d
i
s
t
i
l
l
e
d
h
a
s
h
i
n
g
c
o
d
i
n
g
,
”
i
n
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
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
W
o
r
k
sh
o
p
s
,
I
EEE,
Ju
n
.
2
0
1
9
,
p
p
.
7
1
4
–
7
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
W
.
2
0
1
9
.
0
0
0
9
8
.
[
6
]
H
.
W
a
n
g
a
n
d
X
.
C
a
o
,
“
P
a
l
mp
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
b
i
f
u
r
c
a
t
i
o
n
l
i
n
e
d
i
r
e
c
t
i
o
n
c
o
d
i
n
g
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
3
,
p
p
.
7
0
3
6
6
–
7
0
3
7
7
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
5
.
3
5
6
2
6
4
8
.
[
7
]
A
.
B
h
a
n
d
a
r
y
,
C
.
B
h
a
t
t
,
a
n
d
R
.
J
a
i
n
,
“
A
n
i
n
t
e
r
p
r
e
t
a
b
l
e
p
a
l
m
p
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
a
p
p
r
o
a
c
h
u
si
n
g
a
C
N
N
a
n
d
e
x
p
l
a
i
n
a
b
l
e
-
AI
f
r
a
mew
o
r
k
s,”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
2
4
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
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
Em
e
r
g
i
n
g
T
e
c
h
n
o
l
o
g
y
,
G
l
o
b
a
l
AI
S
u
m
m
i
t
2
0
2
4
,
I
EEE,
S
e
p
.
2
0
2
4
,
p
p
.
3
2
9
–
3
3
4
,
d
o
i
:
1
0
.
1
1
0
9
/
G
l
o
b
a
l
A
I
S
u
mm
i
t
6
2
1
5
6
.
2
0
2
4
.
1
0
9
4
7
9
8
7
.
[
8
]
S
.
S
i
n
h
a
a
n
d
S
.
B
.
V
e
r
ma,
“
H
a
r
n
e
ssi
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
G
A
N
t
e
c
h
n
o
l
o
g
i
e
s
f
o
r
p
a
l
mp
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
2
0
2
4
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
y
b
e
r
n
a
t
i
o
n
a
n
d
C
o
m
p
u
t
a
t
i
o
n
,
C
Y
BERC
O
M
2
0
2
4
,
I
EEE,
N
o
v
.
2
0
2
4
,
p
p
.
6
8
–
76
,
d
o
i
:
1
0
.
1
1
0
9
/
C
Y
B
E
R
C
O
M
6
3
6
8
3
.
2
0
2
4
.
1
0
8
0
3
2
4
5
.
[
9
]
T
.
M
i
n
-
Je
n
a
n
d
C
.
C
h
e
n
g
-
T
a
o
,
“
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
d
e
t
e
c
t
i
n
g
d
e
e
p
f
a
k
e
p
a
l
mp
r
i
n
t
i
mag
e
s,”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
1
0
3
4
0
5
–
1
0
3
4
1
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
4
3
3
4
9
7
.
[
1
0
]
S
.
A
.
G
r
o
sz
a
n
d
A
.
K
.
Jai
n
,
“
G
e
n
P
a
l
m:
c
o
n
t
a
c
t
l
e
ss
p
a
l
m
p
r
i
n
t
g
e
n
e
r
a
t
i
o
n
w
i
t
h
d
i
f
f
u
s
i
o
n
mo
d
e
l
s,”
i
n
Pro
c
e
e
d
i
n
g
s
-
2
0
2
4
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
r
e
n
c
e
o
n
B
i
o
m
e
t
ri
c
s,
I
J
C
B
2
0
2
4
,
I
EEE,
S
e
p
.
2
0
2
4
,
p
p
.
1
–
9
,
d
o
i
:
1
0
.
1
1
0
9
/
I
JC
B
6
2
1
7
4
.
2
0
2
4
.
1
0
7
4
4
4
9
3
.
[
1
1
]
S
.
Z
e
d
d
a
,
S
.
M
.
L
a
C
a
v
a
,
S
.
C
a
r
t
a
,
R
.
C
a
su
l
a
,
a
n
d
G
.
L
.
M
a
r
c
i
a
l
i
s
,
“
C
o
mb
i
n
e
d
a
d
v
e
r
sari
a
l
a
t
t
a
c
k
s
f
o
r
b
r
e
a
k
i
n
g
f
i
n
g
e
r
p
r
i
n
t
r
e
c
o
g
n
i
t
i
o
n
a
n
d
p
r
e
se
n
t
a
t
i
o
n
a
t
t
a
c
k
d
e
t
e
c
t
i
o
n
,
”
i
n
Pr
o
c
.
2
0
2
5
3
3
rd
Eu
ro
p
e
a
n
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
C
o
n
f
.
(
EU
S
I
PC
O
)
,
2
0
2
5
,
p
p
.
1
–
5.
[
1
2
]
M
.
I
mr
a
n
,
M
.
S
.
U
mar,
a
n
d
S
.
M
a
l
h
o
t
r
a
,
“
P
r
i
v
a
c
y
p
r
e
se
r
v
i
n
g
c
a
n
c
e
l
l
a
b
l
e
t
e
mp
l
a
t
e
g
e
n
e
r
a
t
i
o
n
f
o
r
c
r
y
p
t
o
-
b
i
o
me
t
r
i
c
a
u
t
h
e
n
t
i
c
a
t
i
o
n
sy
st
e
m,”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
3
,
p
p
.
1
5
8
3
2
2
–
1
5
8
3
3
9
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
5
.
3
6
0
2
7
9
5
.
[
1
3
]
Z
.
Y
a
n
g
,
A
.
B
.
J.
T
e
o
h
,
B
.
Z
h
a
n
g
,
L
.
L
e
n
g
,
a
n
d
Y
.
Z
h
a
n
g
,
“
P
h
y
si
c
s
-
d
r
i
v
e
n
s
p
e
c
t
r
u
m
-
c
o
n
s
i
st
e
n
t
f
e
d
e
r
a
t
e
d
l
e
a
r
n
i
n
g
f
o
r
p
a
l
mp
r
i
n
t
v
e
r
i
f
i
c
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r
V
i
si
o
n
,
v
o
l
.
1
3
2
,
n
o
.
1
0
,
p
p
.
4
2
5
3
–
4
2
6
8
,
O
c
t
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
2
6
3
-
024
-
0
2
0
7
7
-
9.
[
1
4
]
S
.
D
e
n
g
,
H
.
Z
h
a
o
,
W
.
F
a
n
g
,
J.
Y
i
n
,
S
.
D
u
st
d
a
r
,
a
n
d
A
.
Y
.
Z
o
may
a
,
“
Ed
g
e
i
n
t
e
l
l
i
g
e
n
c
e
:
t
h
e
c
o
n
f
l
u
e
n
c
e
o
f
e
d
g
e
c
o
mp
u
t
i
n
g
a
n
d
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
,
”
I
EEE
I
n
t
e
rn
e
t
o
f
T
h
i
n
g
s
J
o
u
rn
a
l
,
v
o
l
.
7
,
n
o
.
8
,
p
p
.
7
4
5
7
–
7
4
6
9
,
A
u
g
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
JI
O
T
.
2
0
2
0
.
2
9
8
4
8
8
7
.
[
1
5
]
C
.
L
u
g
a
r
e
si
e
t
a
l
.
,
“
M
e
d
i
a
P
i
p
e
:
a
f
r
a
me
w
o
r
k
f
o
r
b
u
i
l
d
i
n
g
p
e
r
c
e
p
t
i
o
n
p
i
p
e
l
i
n
e
s,”
J
u
n
.
2
0
1
9
,
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
a
r
x
i
v
.
o
r
g
/
a
b
s/
1
9
0
6
.
0
8
1
7
2
[
1
6
]
T
.
C
h
e
n
,
S
.
K
o
r
n
b
l
i
t
h
,
M
.
N
o
r
o
u
z
i
,
a
n
d
G
.
H
i
n
t
o
n
,
“
A
si
mp
l
e
f
r
a
mew
o
r
k
f
o
r
c
o
n
t
r
a
st
i
v
e
l
e
a
r
n
i
n
g
o
f
v
i
su
a
l
r
e
p
r
e
se
n
t
a
t
i
o
n
s,”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
Re
sea
r
c
h
,
2
0
2
0
,
p
p
.
1
5
9
7
–
1
6
0
7
.
[
1
7
]
Z
.
G
a
o
a
n
d
I
.
P
a
t
r
a
s,
“
S
e
l
f
-
su
p
e
r
v
i
sed
f
a
c
i
a
l
r
e
p
r
e
se
n
t
a
t
i
o
n
l
e
a
r
n
i
n
g
w
i
t
h
f
a
c
i
a
l
r
e
g
i
o
n
a
w
a
r
e
n
e
ss,”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
E
E
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
Pa
t
t
e
rn
R
e
c
o
g
n
i
t
i
o
n
,
I
EEE,
J
u
n
.
2
0
2
4
,
p
p
.
2
0
8
1
–
2
0
9
2
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
5
2
7
3
3
.
2
0
2
4
.
0
0
2
0
3
.
[
1
8
]
W
.
Z
h
o
u
,
L
.
K
o
n
g
,
Y
.
H
a
n
,
J.
Q
i
n
,
a
n
d
Z
.
S
u
n
,
“
C
o
n
t
e
x
t
u
a
l
i
z
e
d
r
e
l
a
t
i
o
n
p
r
e
d
i
c
t
i
v
e
mo
d
e
l
f
o
r
se
l
f
-
su
p
e
r
v
i
se
d
g
r
o
u
p
a
c
t
i
v
i
t
y
r
e
p
r
e
se
n
t
a
t
i
o
n
l
e
a
r
n
i
n
g
,
”
I
E
EE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
M
u
l
t
i
m
e
d
i
a
,
v
o
l
.
2
6
,
p
p
.
3
5
3
–
3
6
6
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
T
M
M
.
2
0
2
3
.
3
2
6
5
2
8
0
.
[
1
9
]
A
.
H
o
w
a
r
d
e
t
a
l
.
,
“
S
e
a
r
c
h
i
n
g
f
o
r
mo
b
i
l
e
N
e
t
V
3
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EE
E
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
e
r V
i
si
o
n
,
I
EEE,
O
c
t
.
2
0
1
9
,
p
p
.
1
3
1
4
–
1
3
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
.
2
0
1
9
.
0
0
1
4
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
R
o
b
u
s
t p
a
lmp
r
in
t
b
io
metric so
lu
tio
n
fo
r
s
ec
u
r
e
mo
b
ile
a
u
th
en
tica
tio
n
(
S
o
n
N
g
u
ye
n
)
689
[
2
0
]
K
.
H
a
n
,
Y
.
W
a
n
g
,
Q
.
T
i
a
n
,
J
.
G
u
o
,
C
.
X
u
,
a
n
d
C
.
X
u
,
“
G
h
o
st
N
e
t
:
mo
r
e
f
e
a
t
u
r
e
s
f
r
o
m
c
h
e
a
p
o
p
e
r
a
t
i
o
n
s,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
IE
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
0
,
p
p
.
1
5
7
7
–
1
5
8
6
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
4
2
6
0
0
.
2
0
2
0
.
0
0
1
6
5
.
[
2
1
]
N
.
İ
b
r
a
h
i
mo
ğ
u
,
M
.
C
.
A
y
t
e
k
i
n
,
a
n
d
F
.
Y
ı
l
d
ı
z
,
“
K
n
o
w
l
e
d
g
e
d
i
st
i
l
l
a
t
i
o
n
f
r
o
m
r
e
sn
e
t
t
o
mo
b
i
l
e
n
e
t
f
o
r
a
c
c
u
r
a
t
e
o
n
-
d
e
v
i
c
e
f
a
c
e
r
e
c
o
g
n
i
t
i
o
n
,
”
AI
PA’s
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
:
Br
i
d
g
i
n
g
T
e
c
h
n
o
l
o
g
y
,
S
o
c
i
e
t
y
a
n
d
P
o
l
i
c
y
,
v
o
l
.
1
,
n
o
.
2
,
p
p
.
1
–
1
5
,
2
0
2
5
,
d
o
i
:
1
0
.
5
2
8
1
/
z
e
n
o
d
o
.
1
5
5
0
2
0
9
1
.
[
2
2
]
B
.
J
a
c
o
b
e
t
a
l
.
,
“
Q
u
a
n
t
i
z
a
t
i
o
n
a
n
d
t
r
a
i
n
i
n
g
o
f
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
e
f
f
i
c
i
e
n
t
i
n
t
e
g
e
r
-
a
r
i
t
h
me
t
i
c
-
o
n
l
y
i
n
f
e
r
e
n
c
e
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EE
E
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
I
EEE,
J
u
n
.
2
0
1
8
,
p
p
.
2
7
0
4
–
2
7
1
3
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
8
.
0
0
2
8
6
.
[
2
3
]
M
.
D
o
u
z
e
e
t
a
l
.
,
“
T
h
e
f
a
i
ss l
i
b
r
a
r
y
,
”
I
EEE
T
r
a
n
sa
c
t
i
o
n
s
o
n
B
i
g
D
a
t
a
,
O
c
t
.
2
0
2
5
,
d
o
i
:
1
0
.
1
1
0
9
/
T
B
D
A
TA
.
2
0
2
5
.
3
6
1
8
4
7
4
.
[
2
4
]
J.
Jo
h
n
so
n
,
M
.
D
o
u
z
e
,
a
n
d
H
.
Je
g
o
u
,
“
B
i
l
l
i
o
n
-
sca
l
e
si
mi
l
a
r
i
t
y
se
a
r
c
h
w
i
t
h
G
P
U
s,”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
Bi
g
D
a
t
a
,
v
o
l
.
7
,
n
o
.
3
,
p
p
.
5
3
5
–
5
4
7
,
J
u
l
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
T
B
D
A
TA
.
2
0
1
9
.
2
9
2
1
5
7
2
.
[
2
5
]
H
.
Z
h
i
b
i
n
g
,
“
Q
u
i
c
k
a
n
d
e
f
f
i
c
i
e
n
t
l
a
r
g
e
sca
l
e
a
p
p
r
o
x
i
mat
e
n
e
a
r
e
st
n
e
i
g
h
b
o
r
se
a
r
c
h
o
n
h
i
g
h
-
d
i
me
n
s
i
o
n
a
l
d
a
t
a
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
m
p
u
t
e
r
C
o
n
f
e
r
e
n
c
e
o
n
Wa
v
e
l
e
t
Ac
t
i
v
e
Me
d
i
a
T
e
c
h
n
o
l
o
g
y
a
n
d
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ssi
n
g
,
I
C
C
WA
MTI
P
,
I
EEE,
D
e
c
.
2
0
2
4
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
W
A
M
T
I
P
6
4
8
1
2
.
2
0
2
4
.
1
0
8
7
3
6
9
3
.
[
2
6
]
K
.
B
.
R
a
j
a
,
R
.
R
a
g
h
a
v
e
n
d
r
a
,
V
.
K
.
V
e
mu
r
i
,
a
n
d
C
.
B
u
s
c
h
,
“
S
mart
p
h
o
n
e
b
a
se
d
v
i
si
b
l
e
i
r
i
s
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
d
e
e
p
sp
a
r
se
f
i
l
t
e
r
i
n
g
,
”
Pa
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
L
e
t
t
e
rs
,
v
o
l
.
5
7
,
p
p
.
3
3
–
4
2
,
M
a
y
2
0
1
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
r
e
c
.
2
0
1
4
.
0
9
.
0
0
6
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
S
o
n
Ng
u
y
e
n
o
b
tain
e
d
B.
S
c
.
i
n
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
T
e
x
a
s
A&
M
Un
iv
e
rsit
y
,
Co
rp
u
s
C
h
risti
,
T
e
x
a
s,
US
A
.
He
is
p
u
rsin
g
m
a
ste
r’s
d
e
g
r
e
e
in
c
o
m
p
u
ter
sc
ien
c
e
a
t
Ch
u
lalo
n
g
k
o
rn
Un
iv
e
rsity
,
Ba
n
g
k
o
k
,
T
h
a
il
a
n
d
.
His
re
se
a
r
c
h
f
o
c
u
e
s
o
n
se
c
u
rit
y
a
n
d
so
f
t
w
a
re
d
e
v
e
lo
p
m
e
n
t.
He
c
a
n
b
e
c
o
n
tac
te
d
a
t
e
m
a
il
:
6
7
7
8
0
1
4
1
2
3
@s
tu
d
e
n
t.
c
h
u
la.ac
.
th
P
r
o
fe
ss
o
r
Dr
.
A
r
th
o
r
n
Lu
a
ng
s
o
d
s
a
i
o
b
tain
e
d
B
.
S
c
.
i
n
Co
m
p
u
ter
S
c
ien
c
e
f
r
o
m
T
h
a
m
m
a
sa
t
Un
iv
e
rist
y
,
T
h
a
il
a
n
d
;
M
.
S
c
.
A
n
a
ly
sis,
De
sig
n
,
a
n
d
M
a
n
a
g
e
m
e
n
t
o
f
In
f
o
rm
a
ti
o
n
S
y
st
e
m
s
f
ro
m
L
o
n
d
o
n
S
c
h
o
o
l
o
f
Eco
n
o
m
ics
,
U.K.;
M
.
S
c
.
a
n
d
P
h
.
D.
i
n
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Un
iv
e
rsit
y
o
f
Esse
x
,
U.K.
H
e
h
o
ld
s
t
h
e
p
o
si
ti
o
n
o
f
A
ss
istan
t
P
r
o
f
e
ss
o
r
in
th
e
De
p
a
rtm
e
n
t
o
f
M
a
th
e
m
a
ti
c
s
a
n
d
Co
m
p
u
ter
S
c
ien
c
e
a
t
th
e
F
a
c
u
lt
y
o
f
S
c
ien
c
e
,
Ch
u
lalo
n
g
k
o
rn
U
n
iv
e
rsity
,
T
h
a
il
a
n
d
.
His
re
se
a
rc
h
f
o
c
u
se
s
o
n
so
f
tw
a
re
e
n
g
in
e
e
rin
g
,
o
u
tl
ier
d
e
tec
ti
o
n
a
n
d
i
n
f
o
rm
a
ti
o
n
s
y
ste
m
s
m
a
n
a
g
e
m
e
n
t.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
rth
o
rn
.
l@c
h
u
la.ac
.
th
Pro
f
e
ss
o
r
D
r
.
Pa
tta
r
a
si
n
e
e
B
h
a
t
ta
r
a
k
o
so
l
o
b
tai
n
e
d
a
B.
S
c
.
in
m
a
th
e
m
a
ti
c
s
f
ro
m
Ch
u
lalo
n
g
k
o
rn
U
n
iv
e
rsity
,
Ba
n
g
k
o
k
,
T
h
a
il
a
n
d
;
a
n
M
.
S
c
.
in
a
p
p
l
ied
sta
ti
stics
f
ro
m
th
e
Na
ti
o
n
a
l
In
stit
u
te
o
f
De
v
e
lo
p
m
e
n
t
A
d
m
in
istratio
n
,
Ba
n
g
k
o
k
,
T
h
a
il
a
n
d
;
a
n
d
a
P
h
.
D.
in
c
o
m
p
u
ter
sc
ien
c
e
f
ro
m
W
o
ll
o
n
g
o
n
g
Un
iv
e
rsity
,
Au
str
a
li
a
.
S
h
e
h
o
ld
s
th
e
p
o
sit
io
n
o
f
a
ss
o
c
iate
p
ro
f
e
ss
o
r
in
th
e
De
p
a
rtme
n
t
o
f
M
a
th
e
m
a
ti
c
s
a
n
d
Co
m
p
u
ter
S
c
ien
c
e
a
t
th
e
F
a
c
u
lt
y
o
f
S
c
ien
c
e
,
Ch
u
lalo
n
g
k
o
rn
Un
iv
e
rsit
y
,
T
h
a
il
a
n
d
.
He
r
re
se
a
r
c
h
f
o
c
u
se
s
o
n
c
o
m
p
u
ter
n
e
tw
o
rk
s,
se
c
u
rit
y
,
a
n
d
so
f
t
w
a
re
a
rc
h
it
e
c
tu
re
.
Sh
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
p
a
tt
a
ra
sin
e
e
.
b
@c
h
u
la.ac
.
th
Evaluation Warning : The document was created with Spire.PDF for Python.