T
E
L
KO
M
N
I
KA
T
e
lec
om
m
u
n
icat
ion
,
Com
p
u
t
i
n
g,
E
lec
t
r
on
ics
an
d
Cont
r
ol
Vol.
18
,
No.
3
,
J
une
2020
,
pp.
13
89
~
13
96
I
S
S
N:
1693
-
6930,
a
c
c
r
e
dit
e
d
F
ir
s
t
G
r
a
de
by
Ke
me
nr
is
tekdikti
,
De
c
r
e
e
No:
21/E
/KP
T
/2018
DO
I
:
10.
12928/
T
E
L
KO
M
NI
KA
.
v18i3.
14790
1389
Jou
r
n
al
h
omepage
:
ht
tp:
//
jour
nal.
uad
.
ac
.
id/
index
.
php/T
E
L
K
OM
N
I
K
A
V
is
io
n
:
a w
e
b
se
r
vi
c
e
f
or
f
ac
e
r
e
c
og
n
ition
u
si
n
g c
o
n
vol
u
t
io
n
al
n
e
t
w
or
k
Aki
n
o
Ar
c
h
il
les
,
Ar
ya
W
icaks
an
a
D
ep
ar
t
men
t
o
f
In
f
o
rmat
i
cs
,
U
n
i
v
ers
i
t
a
s
Mu
l
t
i
me
d
i
a
N
u
s
an
t
ara,
In
d
o
n
es
i
a
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
J
ul
21
,
2019
R
e
vis
e
d
J
a
n
20
,
2020
Ac
c
e
pted
F
e
b
21
,
2020
T
h
i
s
p
ap
er
p
r
o
p
o
s
e
s
a
face
reco
g
n
i
t
i
o
n
mo
d
u
l
e
b
u
i
l
t
as
a
w
eb
s
erv
i
ce.
W
e
i
n
t
r
o
d
u
ce
a
n
o
v
e
l
d
es
i
g
n
an
d
mech
a
n
i
s
m
fo
r
face
reco
g
n
i
t
i
o
n
o
n
a
w
eb
p
l
a
t
fo
rm
a
n
d
t
o
mem
o
ri
ze
m
o
s
t
rece
n
t
u
s
er
s
fo
r
t
h
e
u
s
e
r.
T
h
i
s
w
e
b
s
e
rv
i
ce
i
s
cal
l
e
d
V
i
s
i
o
n
an
d
d
e
v
el
o
p
e
d
u
s
i
n
g
t
h
e
Fl
as
k
an
d
T
en
s
o
rFl
o
w
d
eep
l
ear
n
i
n
g
framew
o
r
k
.
T
h
e
face
rec
o
g
n
i
t
i
o
n
p
r
o
ces
s
i
s
p
o
w
ere
d
b
y
Face
N
et
d
eep
co
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
rk
m
o
d
e
l
.
T
h
e
face
rec
o
g
n
i
t
i
o
n
p
r
o
ces
s
d
o
n
e
b
y
V
i
s
i
o
n
co
u
l
d
al
s
o
b
e
u
t
i
l
i
ze
d
fo
r
u
s
er
au
t
h
e
n
t
i
cat
i
o
n
an
d
u
s
er
memo
ri
za
t
i
o
n
,
b
o
t
h
d
o
n
e
i
n
o
n
a
w
eb
p
l
a
t
fo
rm.
A
s
a
d
emo
n
s
t
rat
i
o
n
o
f
co
n
ce
p
t
an
d
v
i
ab
i
l
i
t
y
,
i
n
t
h
i
s
s
t
u
d
y
,
V
i
s
i
o
n
i
s
i
n
t
eg
ra
t
ed
i
n
t
o
a
w
eb
-
b
a
s
ed
v
o
i
c
e
ch
at
b
o
t
.
T
h
e
t
e
s
t
i
n
g
an
d
ev
a
l
u
a
t
i
o
n
o
f
V
i
s
i
o
n
’s
face
reco
g
n
i
t
i
o
n
p
ro
ce
s
s
s
h
o
w
an
o
v
era
l
l
F
-
s
c
o
re
o
f
o
n
e
f
o
r
al
l
t
es
t
s
cen
ar
i
o
s
.
K
e
y
w
o
r
d
s
:
C
omput
e
r
vis
ion
C
onvolut
ional
ne
ur
a
l
ne
twor
k
F
a
c
e
r
e
c
ognit
ion
W
e
b
s
e
r
vice
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e
.
C
or
r
e
s
pon
din
g
A
u
th
or
:
Ar
ya
W
ica
ks
a
na
,
De
pa
r
tm
e
nt
of
I
nf
or
mat
ics
,
Unive
r
s
it
a
s
M
ult
im
e
dia
Nus
a
ntar
a
,
S
c
ientia
B
ouleva
r
d
S
t.
,
Ga
ding
S
e
r
pong
,
T
a
nge
r
a
ng
-
15810,
B
a
nten,
I
ndone
s
ia.
E
mail:
a
r
ya
.
wic
a
ks
a
na
@umn.
a
c
.
id
1.
I
NT
RODU
C
T
I
ON
A
c
ha
tbot
is
a
c
onve
r
s
a
ti
ona
l
a
ge
nt
whe
r
e
a
c
ompu
ter
pr
ogr
a
m
is
de
s
igned
to
c
onduc
t
a
c
onve
r
s
a
ti
on
(
textua
l
or
a
udit
o
r
y)
[
1
]
.
I
n
the
de
ve
lopm
e
nt
of
c
ha
tbot
s
,
na
tur
a
l
langua
ge
p
r
oc
e
s
s
ing
a
nd
de
e
p
lea
r
ning
a
r
e
the
two
main
tec
hnologi
e
s
of
a
r
ti
f
icia
l
int
e
ll
ig
e
nc
e
that
a
ll
ows
the
a
dva
nc
e
ment
[
2]
.
No
wa
da
ys
,
f
a
c
e
identif
ica
ti
on
a
nd
r
e
c
ognit
ion
a
r
e
a
ls
o
a
l
r
e
a
dy
a
n
e
s
tablis
he
d
a
ppli
c
a
ti
on
o
f
c
omput
e
r
v
is
ion
[
3
-
9
]
.
Although
it
s
pe
r
f
or
manc
e
is
s
ti
ll
not
a
s
good
a
s
the
human
e
ye
s
[
10
]
,
f
a
c
e
identif
ica
ti
on
ha
s
a
lr
e
a
dy
be
e
n
us
e
d
wide
ly
a
s
a
non
-
int
r
us
ive
biom
e
tr
ic
tec
hnique
[
1
1
]
.
T
his
is
due
to
it
s
c
onve
nient
na
t
ur
e
f
or
a
uthentica
ti
ng
us
e
r
s
without
r
e
quir
ing
a
ny
phys
ica
l
c
ontac
t
wi
th
the
de
vice
[
1
2
]
.
I
n
thi
s
s
tudy,
we
p
r
opos
e
a
nove
l
wa
y
of
doing
f
a
c
e
r
e
c
ognit
ion
buil
t
a
s
a
we
b
s
e
r
vice
.
T
he
a
va
il
a
bil
it
y
o
f
a
f
a
c
e
r
e
c
ognit
ion
mo
dule
a
s
a
we
b
s
e
r
vice
would
be
ne
f
it
many
we
bs
it
e
s
that
would
wa
nt
to
e
xplor
e
thi
s
tec
hnology
.
W
e
b
s
e
r
vice
s
a
ll
ow
the
a
ppli
c
a
ti
on
to
be
platf
o
r
m
a
nd
tec
hnology
in
de
pe
nde
nt.
I
n
a
ddit
ion
to
that,
the
p
r
oc
e
s
s
ing
of
the
f
a
c
e
im
a
ge
would
a
ls
o
incr
e
a
s
e
us
e
r
e
xpe
r
ienc
e
.
T
hu
s
,
a
s
pr
oo
f
o
f
c
onc
e
pt,
the
we
b
s
e
r
vice
is
de
ve
l
ope
d
a
nd
c
a
ll
e
d
V
is
ion,
a
nd
it
is
de
s
igned
a
nd
im
pleme
nted
on
J
a
c
ob.
J
a
c
ob
[
1
3
]
is
a
we
b
-
ba
s
e
d
voice
c
ha
tbot
that
is
powe
r
e
d
by
W
it
.
AI
a
n
d
pr
og
r
a
mm
e
d
to
pr
ovide
inf
or
mation
a
bout
Unive
r
s
it
a
s
M
ult
im
e
dia
Nu
s
a
ntar
a
joi
nt
-
de
gr
e
e
I
n
f
or
matics
p
r
ogr
a
m
inf
o
r
mation.
J
a
c
ob
wor
ks
with
s
ound
(
a
udio)
input
a
nd
tr
a
ns
lat
e
s
it
int
o
text
us
ing
the
w
e
b
s
pe
e
c
h
API
.
I
t
is
the
n
s
e
nt
to
W
it
.
AI
to
obtain
the
int
e
nt
(
the
goa
l
of
the
us
e
r
is
c
omi
ng
to
the
c
ha
tbot
)
a
nd
e
nti
ti
e
s
(
im
por
tant
va
r
iable
in
int
e
nt
that
he
lps
a
dd
r
e
leva
nc
e
to
a
n
int
e
nt
)
o
f
the
text.
T
he
obtaine
d
int
e
nt
a
nd
e
nti
ti
e
s
a
r
e
c
he
c
ke
d
a
nd
c
ompar
e
d
with
the
knowle
dge
da
taba
s
e
to
r
e
tu
r
n
with
the
a
pp
r
opr
iate
r
e
s
pons
e
.
T
he
r
e
s
pons
e
whic
h
is
in
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
13
89
-
13
96
1390
text
f
or
mat
is
tr
a
ns
late
d
int
o
a
voice
us
ing
S
pe
e
c
h
S
ynthes
is
of
the
w
e
b
s
pe
e
c
h
API
.
J
a
c
ob
r
e
c
ognize
s
thr
e
e
us
e
r
r
oles
:
a
dmi
nis
tr
a
tor
,
s
upe
r
a
dmi
nis
tr
a
to
r
,
a
nd
ba
s
ic
us
e
r
.
T
he
r
ole
o
f
a
n
a
dmi
nis
tr
a
to
r
a
n
d
s
upe
r
a
dmi
nis
tr
a
tor
is
to
upda
te
the
knowle
dge
da
taba
s
e
with
the
late
s
t
inf
or
mation
to
e
ns
ur
e
that
the
c
ontent
is
up
-
to
-
da
te.
T
he
a
uthentica
ti
on
pr
oc
e
s
s
is
done
u
s
ing
a
pa
s
s
wor
d
[
1
3
]
.
T
he
a
ddit
ion
of
a
f
a
c
e
r
e
c
ognit
ion
modul
e
(
V
is
ion)
to
J
a
c
ob
would
e
nha
nc
e
is
i
nter
a
c
ti
vit
y
a
nd
e
nr
ich
it
s
c
a
pa
bil
it
ies
.
T
he
main
a
lgor
it
hm
us
e
d
f
o
r
V
is
ion
is
the
de
e
p
c
onvolut
ional
ne
twor
k.
As
e
xplaine
d
in
[
1
4
,
1
5
]
,
c
onvolut
ional
ne
ur
a
l
ne
twor
k
(
C
NN
)
pe
r
f
o
r
ms
be
t
ter
than
mul
ti
laye
r
pe
r
c
e
pt
r
on
a
nd
mo
r
e
r
obus
t
in
c
ompl
e
x
pa
tt
e
r
n
r
e
c
o
gnit
ion
s
uc
h
a
s
dis
tor
ti
on,
tr
a
ns
lation,
s
c
a
li
ng,
a
nd
r
otation.
S
im
ple
C
NN
e
xplaine
d
in
[
1
6
]
c
ons
is
ts
of
thr
e
e
main
laye
r
s
a
nd
c
ould
be
opti
mi
z
e
d
a
c
c
or
ding
to
ne
e
ds
.
As
ther
e
a
r
e
va
r
io
us
C
NN
a
r
c
hit
e
c
tur
e
s
,
the
one
that
is
uti
li
z
e
d
by
V
is
ion
is
F
a
c
e
Ne
t.
F
a
c
e
Ne
t
is
a
s
ys
tem
that
dir
e
c
tl
y
lea
r
ns
a
mapping
f
r
om
f
a
c
e
im
a
ge
s
to
a
c
ompac
t
E
uc
li
de
a
n
s
pa
c
e
whe
r
e
dis
tanc
e
s
dir
e
c
tl
y
c
or
r
e
s
pond
to
a
mea
s
ur
e
of
f
a
c
e
s
im
il
a
r
it
y
[
1
7
]
.
F
a
c
e
Ne
t
[
1
7
]
a
c
hieve
d
99
.
63%
a
c
c
ur
a
c
y
on
the
L
a
be
led
F
a
c
e
s
in
the
W
il
d
da
tab
a
s
e
a
nd
95.
12%
on
YouT
ube
F
a
c
e
s
,
whic
h
outper
f
or
ms
s
ome
other
C
NN
models
s
uc
h
a
s
De
e
pI
D,
De
e
p
I
D2,
a
nd
G
a
us
s
ianFac
e
[
1
8
]
.
T
hus
,
F
a
c
e
Ne
t
is
c
hos
e
n
a
s
the
c
or
e
tec
hnology
f
or
the
f
a
c
e
r
e
c
ognit
ion
p
r
oc
e
s
s
in
V
is
ion.
T
he
f
a
c
e
r
e
c
ognit
ion
of
J
a
c
ob
us
e
r
s
is
pe
r
f
or
med
be
twe
e
n
1
a
nd
2
mete
r
s
[
1
9
]
a
nd
the
li
ghti
ng
c
ondit
ions
a
r
e
take
n
int
o
a
c
c
ount
with
int
e
ns
it
y
be
tw
e
e
n
250
a
nd
400
lu
mens
/m
2
[
20
]
.
T
he
im
pleme
ntation
of
a
f
a
c
e
r
e
c
ognit
ion
modul
e
(
V
is
ion)
int
o
a
voice
c
ha
tbot
(
J
a
c
ob)
would
a
ll
ow
the
e
xpa
ns
ion
of
it
s
f
e
a
tur
e
s
,
in
thi
s
wor
k,
s
uc
h
a
s
the
a
uthentica
ti
on
pr
oc
e
s
s
a
nd
us
e
r
memor
iza
ti
on
f
e
a
tur
e
.
T
he
a
uthentica
ti
on
pr
oc
e
s
s
f
or
J
a
c
ob
us
e
r
s
is
c
a
r
r
i
e
d
on
us
ing
f
a
c
e
r
e
c
ognit
ion
done
a
s
a
ba
c
kgr
ound
pr
oc
e
s
s
.
T
he
us
e
r
memo
r
iza
ti
on
f
e
a
tur
e
a
ll
ows
J
a
c
ob
to
gr
e
e
t
r
e
tur
ning
us
e
r
s
,
a
nd
de
ve
lop
a
c
onne
c
ti
on
be
twe
e
n
the
c
ha
tbot
a
nd
the
us
e
r
s
.
T
he
a
uthentica
ti
on
pr
o
c
e
s
s
to
logi
n
int
o
the
s
ys
tem
is
de
s
igned
to
ha
ve
a
f
a
ls
e
pos
it
ive
r
a
te
les
s
than
or
e
qua
l
to
0.
001
%
a
nd
f
a
ls
e
-
ne
ga
ti
ve
r
a
te
les
s
than
or
e
qua
l
to
1%
[
2
1
]
.
T
he
pe
r
f
o
r
manc
e
e
va
luation
o
f
V
is
ion
would
be
mea
s
ur
e
d
a
c
c
or
ding
to
[
2
1
]
,
whe
r
e
F
-
mea
s
ur
e
is
us
e
d
to
mea
s
ur
e
the
F
-
s
c
or
e
[
2
2
]
of
the
a
uthentica
ti
on
(
f
a
c
e
r
e
c
ognit
ion)
pr
oc
e
s
s
f
or
a
dmi
nis
tr
a
tor
a
nd
s
upe
r
a
dmi
nis
tr
a
tor
us
e
r
r
o
les
.
2.
RE
S
E
AR
CH
M
E
T
HO
D
T
his
s
e
c
ti
on
c
ons
is
ts
of
r
e
quir
e
ment
a
na
lys
is
,
d
e
s
ign
,
a
nd
im
p
leme
ntation
o
f
V
is
ion.
T
he
ne
xt
s
e
c
ti
on
dis
c
us
s
e
s
the
r
e
s
ult
a
nd
a
na
lys
is
of
V
is
ion,
including
the
tes
t
s
c
e
na
r
ios
a
nd
pe
r
f
o
r
manc
e
e
va
l
ua
ti
on
of
the
f
a
c
e
r
e
c
ognit
ion
a
nd
a
uthentica
ti
on
pr
oc
e
s
s
.
2
.
1.
Re
q
u
ire
m
e
n
t
s
a
n
alys
is
B
a
s
e
d
on
the
pr
e
li
mi
na
r
y
s
tudi
e
s
on
J
a
c
ob
voice
c
ha
tbot
,
V
is
ion
take
s
input
s
of
us
e
r
de
tails
s
uc
h
a
s
na
me,
r
ole,
a
nd
f
a
c
e
im
a
ge
s
.
T
he
f
a
c
e
im
a
ge
s
a
r
e
us
e
d
f
or
tr
a
ini
ng
the
model
(
c
las
s
if
ier
)
.
T
he
s
e
f
a
c
e
im
a
ge
s
a
r
e
c
a
ptur
e
d
by
V
is
ion
in
the
ba
c
kgr
ound
.
T
he
nu
mber
of
tr
a
ini
ng
im
a
ge
s
is
s
e
t
to
20,
50,
a
nd
100
f
or
tes
ti
ng
pur
pos
e
s
.
J
a
c
ob
a
uthentica
tes
the
a
dmi
nis
tr
a
tor
r
o
le
a
nd
the
s
upe
r
a
dmi
nis
tr
a
tor
r
ole
by
us
ing
the
p
a
s
s
wor
d
dur
ing
logi
n
.
T
hus
,
by
us
ing
V
is
ion,
a
r
e
gis
ter
e
d
a
dmi
nis
tr
a
tor
a
nd
s
upe
r
a
dmi
nis
tr
a
to
r
c
ould
be
a
uthentica
ted
us
ing
f
a
c
e
r
e
c
ognit
ion
by
J
a
c
ob.
T
his
ne
w
a
uthe
nti
c
a
ti
on
pr
oc
e
s
s
is
pr
opos
e
d
to
r
e
plac
e
the
old
-
f
a
s
hioned
pa
s
s
wor
d
-
ba
s
e
d
u
s
e
r
a
uthentica
ti
on
pr
oc
e
s
s
.
Upon
r
e
c
ognizing
the
f
a
c
e
of
a
ba
s
ic
us
e
r
,
V
is
ion
memor
ize
s
by
ke
e
ping
the
f
a
c
e
im
a
ge
s
int
o
the
f
il
e
s
ys
tem.
T
h
is
a
ll
ows
the
memor
iza
ti
on
f
e
a
tur
e
on
J
a
c
ob
a
s
V
is
ion
c
ould
memor
ize
up
to
10
mos
t
r
e
c
e
nt
us
e
r
s
.
C
ons
ider
ing
the
tr
a
ini
ng
pr
oc
e
s
s
of
the
C
NN
m
ode
l
c
ould
take
a
lot
o
f
ti
me,
thus
f
o
r
the
f
a
c
e
r
e
c
ognit
ion
pr
oc
e
s
s
,
V
is
ion
us
e
s
t
he
pr
e
-
tr
a
ine
d
model.
F
ur
ther
mor
e
,
ins
tea
d
of
r
e
-
tr
a
ini
ng
th
e
whole
ne
twor
k,
V
is
ion
only
r
e
-
tr
a
ins
the
c
las
s
if
ier
.
T
he
pr
e
-
tr
a
ined
model
is
tr
a
ined
us
ing
the
VG
GFac
e
2
da
tas
e
t
whic
h
is
ba
s
e
d
on
the
I
nc
e
pti
on
-
R
e
s
Ne
t
-
v1
model
us
e
d
in
the
F
a
c
e
Ne
t
a
s
the
c
las
s
i
f
ier
[
2
3
,
2
4
]
.
P
r
e
-
pr
oc
e
s
s
ing
s
tep
is
done
by
us
ing
F
a
c
e
Ne
t’
s
M
ult
i
-
tas
k
C
a
s
c
a
de
d
C
onvolut
ional
Ne
twor
ks
(
M
T
C
NN
)
to
de
tec
t
a
nd
a
li
gn
f
a
c
e
s
[
2
5
]
.
T
he
c
las
s
if
ier
is
tr
a
ined
us
ing
S
uppor
t
Ve
c
tor
M
a
c
hine
(
S
VM
)
a
s
in
[
2
4
]
.
F
or
the
memo
r
iza
ti
on
f
e
a
tu
r
e
,
a
s
the
number
o
f
b
a
s
ic
us
e
r
s
incr
e
a
s
e
s
,
the
tr
a
ini
ng
ti
me
would
a
ls
o
incr
e
a
s
e
a
c
c
or
dingl
y.
T
his
is
whe
r
e
the
li
mi
tation
of
n
mos
t
r
e
c
e
nt
us
e
r
s
c
omes
to
e
xis
t.
T
he
L
e
a
s
t
R
e
c
e
ntl
y
Us
e
d
a
lgor
it
hm
[
2
6
,
2
7
]
is
us
e
d
f
or
thi
s
f
e
a
tur
e
to
memor
ize
only
n
mos
t
r
e
c
e
nt
us
e
r
s
.
T
he
a
lgo
r
it
h
m
wor
ks
by
r
e
plac
ing
the
oldes
t
memo
r
y
of
the
us
e
r
's
f
a
c
e
with
the
mos
t
r
e
c
e
nt
one
.
T
he
li
mi
t
n
is
s
e
t
to
10
p
e
r
s
ons
f
or
de
mons
tr
a
ti
on
pur
pos
e
s
.
T
he
int
e
gr
a
ti
on
pa
r
t
be
twe
e
n
the
V
is
ion
we
b
s
e
r
vice
with
J
a
c
ob
we
b
a
pp
is
de
s
igned
a
nd
im
pleme
n
ted
us
ing
a
c
li
e
nt
-
s
e
r
ve
r
a
r
c
hit
e
c
tur
e
.
T
he
C
NN
e
ngine
is
r
un
on
the
s
e
r
ve
r
-
s
ide,
thus
e
ns
ur
ing
the
f
a
c
e
r
e
c
ognit
ion
pr
oc
e
s
s
to
ha
ve
a
r
e
li
a
ble
pe
r
f
or
manc
e
with
a
de
qua
te
pr
oc
e
s
s
ing
powe
r
.
I
t
is
then
c
onne
c
ted
with
the
a
ppli
c
a
ti
on
pr
og
r
a
mm
ing
int
e
r
f
a
c
e
(
AP
I
)
that
is
bu
il
t
us
ing
F
las
k
1.
0
.
2
f
r
a
mew
or
k.
On
the
other
ha
nd,
J
a
c
ob’
s
us
e
r
int
e
r
f
a
c
e
r
uns
on
t
he
c
li
e
nt
-
s
ide
a
s
we
ll
a
s
the
V
is
ion
s
ub
-
modul
e
th
a
t
wor
ks
f
or
c
a
ptur
ing
f
a
c
e
im
a
ge
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
V
is
ion:
a
w
e
b
s
e
r
v
ice
for
face
r
e
c
ognit
ion
us
ing
c
onv
olut
ional
ne
tw
or
k
(
A
k
ino
A
r
c
hil
les
)
1391
2
.
2.
De
s
ign
s
T
he
de
s
ign
of
the
we
b
s
e
r
vice
model
f
or
V
is
ion
is
pr
e
s
e
nte
d
in
thi
s
s
e
c
ti
on.
T
he
we
b
s
e
r
vice
model
c
ons
is
t
of
thr
e
e
r
e
s
our
c
e
s
:
AlignT
r
a
in
,
I
de
nti
f
yF
a
c
e
,
a
nd
R
e
gis
ter
F
a
c
e
with
two
r
e
s
our
c
e
pa
ths
:
I
de
n
ti
f
yF
a
c
e
a
nd
R
e
gis
ter
F
a
c
e
us
ing
the
s
a
me
r
e
s
our
c
e
pa
th
.
I
de
nti
f
yF
a
c
e
a
nd
R
e
gis
ter
F
a
c
e
a
r
e
s
e
pa
r
a
ted
in
the
API
by
t
he
pa
r
a
mete
r
r
e
c
e
ived.
Align
T
r
a
in
r
e
s
our
c
e
is
us
e
d
to
pr
e
-
pr
oc
e
s
s
the
r
e
gis
ter
e
d
t
r
a
ini
ng
im
a
ge
s
a
nd
to
t
r
a
in
the
c
las
s
if
ier
in
ba
c
kgr
ound.
T
he
r
e
gis
ter
e
d
us
e
r
s
c
ould
only
be
identi
f
ied
a
f
ter
the
c
las
s
if
ier
ha
s
be
e
n
tr
a
ini
ng
c
ompl
e
tely.
All
o
f
the
r
e
s
our
c
e
s
c
ould
be
a
c
c
e
s
s
e
d
thr
ough
the
given
r
e
s
our
c
e
pa
th
us
ing
the
HT
T
P
pr
otocol.
T
he
r
e
tu
r
n
va
lue
o
f
the
AP
I
is
in
J
S
ON
da
ta
f
o
r
ma
t.
B
e
f
o
r
e
s
tar
ti
ng
V
is
ion
a
s
a
we
b
s
e
r
vice
,
ther
e
’
s
a
ne
e
d
to
c
he
c
k
the
f
il
e
s
ys
tem
f
or
dupli
c
a
ti
on
a
nd
uns
uc
c
e
s
s
f
ul
r
e
gis
tr
a
ti
ons
to
be
r
e
moved.
A
thr
e
a
d
pr
oc
e
s
s
is
then
c
r
e
a
ted
jus
t
be
f
or
e
the
V
is
ion
is
s
tar
ted
a
nd
it
is
s
c
he
duled
to
r
un
e
ve
r
y
two
hou
r
s
(
non
-
blocking)
.
Vi
s
ion
we
b
s
e
r
vice
model
a
s
s
hown
in
F
igur
e
1.
F
igur
e
1.
Vis
ion
we
b
s
e
r
vice
model
I
n
the
r
e
gis
tr
a
ti
on
pr
oc
e
s
s
,
the
r
e
que
s
t
is
c
a
tegor
ize
d
a
s
R
e
gis
ter
F
a
c
e
if
the
pa
r
a
mete
r
c
ontains
:
im
a
ge
,
na
me,
a
c
c
e
s
s
leve
l,
a
nd
pe
ople
lea
ving.
I
t
r
e
tu
r
ns
with
c
ode
,
mes
s
a
ge
,
number
of
i
mage
s
wr
it
ten
,
a
nd
number
of
im
a
ge
s
mus
t
be
take
n.
I
f
the
r
e
gi
s
tr
a
t
ion
pr
oc
e
s
s
is
int
e
r
r
upted,
the
r
e
gis
tr
a
ti
on
p
r
oc
e
s
s
i
s
c
a
nc
e
ll
e
d
a
utom
a
ti
c
a
ll
y.
Upon
a
s
uc
c
e
s
s
f
ul
r
e
gis
tr
a
ti
on,
the
Align
T
r
a
in
r
e
s
our
c
e
is
tr
igger
e
d
to
r
un
to
a
li
gn
the
im
a
ge
s
a
nd
tr
a
in
the
c
las
s
if
ier
.
T
his
pr
oc
e
s
s
is
c
a
r
r
ied
out
in
ba
c
kgr
ound.
All
us
e
r
s
ha
ve
to
pa
s
s
thr
ough
the
r
e
gis
tr
a
ti
on
pr
oc
e
s
s
to
be
s
uc
c
e
s
s
f
ull
y
identif
ied
by
V
is
ion.
R
oles
s
uc
h
a
s
a
dmi
nis
tr
a
tor
a
n
d
s
upe
r
a
dmi
nis
tr
a
tor
a
r
e
r
e
quir
e
d
to
logi
n
int
o
J
a
c
ob
upo
n
a
c
c
e
s
s
ing
the
a
dmi
nis
tr
a
ti
on
s
e
tt
ing
s
.
I
n
the
iden
ti
f
ica
ti
on
pr
oc
e
s
s
,
r
e
que
s
t
w
il
l
be
c
a
tegor
ize
d
a
s
R
e
gis
ter
F
a
c
e
if
the
pa
r
a
mete
r
c
ontains
only
im
a
ge
s
.
I
t
r
e
tu
r
ns
with:
c
ode
,
mes
s
a
ge
,
na
me,
a
c
c
e
s
s
leve
l
a
nd
c
onf
ide
nc
e
leve
l.
Vis
ion
us
e
s
two
thr
e
s
holds
:
0
.
9
a
nd
0.
8
f
or
the
c
onf
idenc
e
leve
l.
C
onf
idenc
e
leve
l
be
low
0
.
8
is
c
ons
ider
e
d
n
ot
r
e
c
ognize
by
V
is
ion.
2
.
3.
I
m
p
lem
e
n
t
at
ion
Vis
ion
UI
is
plac
e
d
in
the
c
e
nter
of
J
a
c
ob
homepa
ge
with
a
c
a
mer
a
int
e
r
f
a
c
e
.
T
he
pr
oc
e
s
s
of
c
a
ptur
ing
im
a
ge
make
s
the
c
a
mer
a
int
e
r
f
a
c
e
b
li
nks
a
s
f
e
e
dba
c
k
to
the
us
e
r
s
.
L
ogin
to
the
a
dmi
nis
tr
a
tor
pa
ge
c
ould
tak
e
plac
e
a
t
thi
s
s
tep.
Af
ter
the
im
a
ge
is
c
a
ptur
e
d
,
a
r
e
que
s
t
is
s
e
nt
to
the
s
e
r
ve
r
with
th
e
im
a
ge
e
nc
ode
d
in
ba
s
e
64
f
or
mat
.
T
he
im
a
ge
is
then
c
onve
r
ted
ba
c
k
to
b
inar
y
a
nd
s
tor
e
d
in
the
f
i
l
e
s
ys
tem.
Vis
ion
pr
oc
e
s
s
e
s
the
im
a
ge
by
r
e
a
ding
it
f
r
om
the
f
il
e
s
ys
t
e
m,
c
onve
r
ti
ng
it
to
R
GB
,
a
nd
r
e
a
ding
the
number
of
de
tec
ted
f
a
c
e
s
.
Vis
ion
will
only
c
onti
nue
to
pr
oc
e
s
s
if
ther
e
is
only
s
ingl
e
f
a
c
e
de
tec
ted,
V
is
ion
doe
s
not
s
uppor
t
identif
ica
ti
on
f
or
mu
lt
ipl
e
f
a
c
e
s
.
T
he
im
a
g
e
is
then
f
li
ppe
d
a
nd
c
r
oppe
d
to
e
xt
r
a
c
t
the
f
e
a
tur
e
s
f
r
om
it
.
T
he
pr
e
whiten
f
e
a
tur
e
e
xtr
a
c
ti
on
is
us
e
d
in
thi
s
s
tep.
T
he
r
e
s
ult
is
then
r
e
tur
ne
d
to
the
c
li
e
nt.
I
f
the
r
e
s
ult
s
hows
that
a
pe
r
s
on
is
identi
f
ied
s
uc
c
e
s
s
f
ull
y,
J
a
c
ob
gr
e
e
ts
the
us
e
r
s
.
T
he
us
e
r
s
c
ould
e
nter
the
r
e
gis
tr
a
ti
on
pr
oc
e
s
s
whe
n
V
i
s
ion
c
ould
not
identif
y
the
f
a
c
e
f
or
two
c
ons
e
c
uti
ve
ti
mes
.
T
he
us
e
r
s
ha
ve
to
r
e
gis
ter
20
im
a
ge
s
int
o
V
i
s
ion.
T
he
inf
or
mation
r
e
quir
e
d
in
the
us
e
r
r
e
gis
tr
a
ti
on
pr
oc
e
s
s
is
the
na
me
of
the
us
e
r
s
.
T
he
r
e
gis
tr
a
ti
on
p
r
oc
e
s
s
s
tar
ts
im
m
e
diate
ly
upon
r
e
c
e
ivi
ng
the
na
me.
I
n
the
e
ve
nt
of
f
a
il
e
d
a
uthentica
ti
on
f
or
a
dmi
nis
tr
a
tor
s
,
logi
n
c
ould
be
d
one
by
r
e
que
s
ti
ng
f
a
c
e
ve
r
if
ica
ti
on
to
J
a
c
ob.
T
his
r
e
plac
e
s
the
old
-
f
a
s
hioned
pa
s
s
wor
d
-
ba
s
e
d
logi
n.
T
he
r
e
gi
s
tr
a
ti
on
pa
ge
c
ons
is
t
of
ba
s
ic
r
e
quir
e
ments
s
uc
h
a
s
na
me,
a
nd
r
ole.
Admini
s
tr
a
tor
ha
s
100
f
a
c
e
im
a
ge
s
c
a
ptur
e
d
f
or
the
r
e
gis
tr
a
ti
on
pr
oc
e
s
s
.
2
.
4.
T
e
s
t
in
g
T
e
s
ti
ng
is
done
unde
r
th
r
e
e
tes
t
s
c
e
na
r
ios
to
s
ho
w
that
the
f
a
c
e
identi
f
ica
ti
on
pr
oc
e
s
s
wor
ks
with
13
us
e
r
s
:
2
s
upe
r
a
dmi
nis
tr
a
tor
r
oles
,
1
a
dmi
nis
tr
a
tor
r
ole
,
a
nd
10
ba
s
ic
us
e
r
r
o
les
.
T
he
s
a
mpl
e
f
a
c
e
im
a
ge
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
13
89
-
13
96
1392
of
the
13
us
e
r
s
a
r
e
dis
playe
d
in
F
igu
r
e
2.
T
he
f
ir
s
t
tes
t
s
c
e
na
r
io
is
r
e
gis
ter
ing
a
ba
s
ic
us
e
r
in
V
is
ion.
T
he
tes
ti
ng
va
r
iable
s
a
r
e
s
hown
in
T
a
ble
1
a
nd
s
a
mpl
e
of
tr
a
ini
ng
im
a
ge
s
hown
in
F
igur
e
3
.
Af
ter
th
e
tr
a
ini
ng
pr
oc
e
s
s
is
done
,
the
Us
e
r
3’
s
f
a
c
e
im
a
ge
a
s
s
hown
in
F
igur
e
4
is
c
a
ptur
e
d
a
nd
s
e
nt
to
the
s
e
r
ve
r
f
or
identif
ica
ti
on.
T
he
identi
f
ica
ti
on
r
e
s
ult
s
h
ows
that
the
c
a
ptur
e
d
im
a
ge
s
of
Us
e
r
3
gives
a
c
onf
idenc
e
leve
l
of
0.
9371.
T
his
r
e
s
ult
pa
s
s
e
d
the
high
thr
e
s
hold
of
0.
9
a
nd
V
is
ion
r
e
c
ognize
s
the
pe
r
s
on
a
s
Us
e
r
3.
T
h
e
s
e
c
ond
tes
t
s
c
e
na
r
io
is
un
r
e
gis
ter
e
d
a
c
c
e
s
s
done
by
a
non
-
r
e
gis
ter
e
d
us
e
r
.
T
he
tes
ti
ng
va
r
iable
s
a
r
e
given
in
T
a
ble
2
a
nd
the
s
a
mpl
e
of
the
tr
a
in
f
a
c
e
im
a
ge
s
of
the
r
e
gis
ter
e
d
us
e
r
s
a
r
e
s
hown
in
F
igur
e
5.
F
igur
e
6
pr
e
s
e
nts
the
f
a
c
e
im
a
ge
s
a
mpl
e
o
f
the
non
-
r
e
gis
ter
e
d
us
e
r
.
He
r
e
the
i
de
nti
f
ica
ti
on
r
e
s
ult
s
hows
that
the
non
-
r
e
gis
ter
e
d
us
e
r
is
not
r
e
c
ognize
by
c
onf
idenc
e
leve
l
o
f
0
.
3281
.
S
ince
the
non
-
r
e
gis
ter
e
d
us
e
r
is
not
r
e
c
ognize
,
s
o
the
us
e
r
is
r
e
gis
ter
e
d
a
utom
a
ti
c
a
ll
y
by
the
V
is
ion
whe
n
the
c
onve
r
s
a
ti
on
be
twe
e
n
the
us
e
r
a
nd
J
a
c
ob
t
a
k
e
pla
c
e
.
Af
ter
the
r
e
gis
tr
a
ti
on
pr
oc
e
s
s
is
f
ini
s
he
d,
thi
s
pr
e
viou
s
ly
non
-
r
e
gis
ter
e
d
us
e
r
is
now
be
c
ome
a
ba
s
ic
us
e
r
a
nd
the
f
a
c
e
im
a
ge
s
a
r
e
s
a
ve
d
in
the
f
il
e
s
ys
tem
a
s
s
hown
in
F
igu
r
e
7.
T
a
ble
1.
F
ir
s
t
tes
t
s
c
e
na
r
io
va
r
iable
s
T
e
s
ti
n
g
V
a
r
ia
b
le
s
C
ondi
ti
on
N
umbe
r
of
t
e
s
ti
ng i
ma
ge
s
100
F
a
c
e
i
ma
ge
i
nput
U
s
e
r
3
F
igur
e
2.
R
e
gis
ter
e
d
us
e
r
s
F
igur
e
3.
F
a
c
e
im
a
ge
s
s
a
mpl
e
of
Us
e
r
3
f
or
tr
a
ini
ng
F
igur
e
4.
F
a
c
e
im
a
ge
s
a
mpl
e
o
f
Us
e
r
3
f
or
identif
ic
a
ti
on
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
V
is
ion:
a
w
e
b
s
e
r
v
ice
for
face
r
e
c
ognit
ion
us
ing
c
onv
olut
ional
ne
tw
or
k
(
A
k
ino
A
r
c
hil
les
)
1393
T
a
ble
2.
S
e
c
ond
tes
t
s
c
e
na
r
io
va
r
iable
s
T
e
s
ti
ng V
a
r
ia
bl
e
s
C
ondi
ti
on
U
s
e
r
s
ta
tu
s
N
o
n
-
r
e
gi
s
te
r
e
d
us
e
r
I
nput
i
ma
ge
N
o
n
-
r
e
gi
s
te
r
e
d us
e
r
N
umbe
r
of
r
e
gi
s
te
r
e
d us
e
r
s
3 pe
r
s
ons
N
umbe
r
of
s
upe
r
a
dmi
ni
s
tr
a
to
r
1 pe
r
s
on
M
a
xi
mum
ba
s
ic
u
s
e
r
s
3
pe
r
s
ons
N
umbe
r
of
r
e
gi
s
te
r
e
d ba
s
ic
us
e
r
s
3 pe
r
s
ons
L
e
a
s
t
r
e
c
e
nt
ly
us
e
d a
lg
or
it
hm
R
unni
ng
(
Us
e
r
4)
(
Us
e
r
5)
(
Us
e
r
6)
F
igur
e
5.
F
a
c
e
im
a
ge
s
a
mpl
e
o
f
r
e
gis
ter
e
d
us
e
r
s
F
igur
e
6.
F
a
c
e
im
a
ge
s
a
mpl
e
o
f
a
non
-
r
e
gis
ter
e
d
us
e
r
F
igur
e
7.
L
is
t
of
r
e
gis
ter
e
d
us
e
r
s
T
he
thi
r
d
tes
t
s
c
e
na
r
io
is
a
S
upe
r
a
dmi
n
is
tr
a
tor
2
that
ha
s
be
e
n
r
e
gis
ter
e
d
in
V
is
ion
us
e
s
J
a
c
ob.
T
he
tes
ti
ng
va
r
iable
s
s
hown
in
T
a
ble
3,
t
r
a
in
im
a
ge
s
a
mpl
e
of
s
upe
r
a
dmi
nis
tr
a
tor
a
nd
a
dmi
nis
tr
a
tor
us
e
r
s
s
hown
in
F
igur
e
8,
a
nd
c
a
ptur
e
d
im
a
ge
of
s
upe
r
a
d
mi
nis
tr
a
tor
2
s
hown
in
F
igur
e
9.
I
de
nti
f
ica
ti
on
r
e
s
ult
s
hows
that
c
a
ptur
e
d
im
a
ge
o
f
S
upe
r
a
dmi
nis
tr
a
tor
2
a
c
h
ieve
c
onf
idenc
e
r
e
s
ult
of
0
.
9287
whic
h
pa
s
s
e
d
t
he
high
thr
e
s
hold
of
0
.
9
a
nd
identif
ied
a
s
S
upe
r
a
dmi
nis
tr
a
t
or
2.
T
a
ble
3.
T
hi
r
d
tes
t
s
c
e
na
r
io
va
r
iable
s
Te
s
ti
ng V
a
r
ia
bl
e
s
C
ondi
ti
on
U
s
e
r
s
ta
tu
s
S
upe
r
a
dmi
ni
s
tr
a
to
r
2 (
r
e
gi
s
te
r
e
d)
I
ma
ge
i
nput
S
upe
r
a
dmi
ni
s
tr
a
to
r
2
N
umbe
r
of
r
e
gi
s
te
r
e
d us
e
r
s
13 pe
r
s
ons
Nu
mbe
r
of
s
upe
r
a
dmi
ni
s
tr
a
to
r
a
nd
a
dmi
ni
s
tr
a
to
r
3 pe
r
s
ons
M
a
xi
mum
ba
s
ic
u
s
e
r
s
10 pe
r
s
ons
N
umbe
r
of
ba
s
ic
us
e
r
s
10 pe
r
s
ons
(
S
upe
r
a
dmi
nis
tr
a
tor
1)
(
S
upe
r
a
dmi
nis
tr
a
tor
2)
(
Admini
s
tator
1)
F
igur
e
8.
Ne
w
li
s
t
o
f
r
e
gis
ter
e
d
us
e
r
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
13
89
-
13
96
1394
F
igur
e
9.
F
a
c
e
im
a
ge
s
a
mpl
e
o
f
S
upe
r
a
dm
ini
s
tr
a
to
r
2
3.
RE
S
UL
T
S
AN
D
AN
AL
YSI
S
T
he
r
e
s
ult
of
identi
f
ica
ti
on
is
r
e
c
or
de
d
a
nd
e
va
luate
d
us
ing
F
-
s
c
or
e
.
T
he
F
-
s
c
or
e
e
va
luations
us
e
s
s
a
mpl
e
s
of
20
,
50
,
a
nd
100
tr
a
ini
ng
im
a
ge
s
f
o
r
c
ompar
is
on.
T
he
e
va
luation
r
e
f
e
r
s
to
two
types
o
f
t
hr
e
s
hold
a
nd
it
is
e
va
luate
d
f
r
om
c
onf
idenc
e
leve
l
of
0.
8
to
1.
T
he
tes
ti
ng
is
done
15
ti
mes
f
o
r
e
a
c
h
nu
mber
o
f
tr
a
ini
ng
im
a
ge
pe
r
a
dmi
nis
tr
a
tor
.
T
a
ble
4
s
hows
the
tes
t
r
e
s
ult
of
S
upe
r
a
dmi
nis
tr
a
tor
1
na
med
S
teve
n.
B
a
s
e
d
on
the
tes
t
r
e
s
ult
s
in
T
a
ble
4
,
the
F
-
s
c
or
e
f
or
S
u
pe
r
a
dmi
nis
tr
a
tor
1
is
one
.
T
a
ble
5
s
hows
the
tes
t
r
e
s
ult
of
S
upe
r
a
dmi
nis
tr
a
to
r
2
na
med
Akino
.
B
a
s
e
d
on
th
e
tes
t
r
e
s
ult
s
in
T
a
ble
5
the
F
-
s
c
or
e
f
or
S
upe
r
a
dmi
nis
tr
a
tor
2
is
one
.
T
a
ble
6
s
hows
the
tes
t
r
e
s
ult
of
Admini
s
tr
a
tor
1
na
med
Oc
ta.
B
a
s
e
d
o
n
the
tes
t
r
e
s
ult
in
T
a
ble
6
,
the
F
-
s
c
or
e
f
or
Admini
s
tr
a
to
r
1
is
one
.
T
a
ble
4.
T
e
s
t
r
e
s
ult
of
s
upe
r
a
dmi
nis
tr
a
tor
1
T
e
s
t
N
o.
S
te
ve
n /
S
upe
r
a
dmi
ni
s
tr
a
to
r
1
20 t
r
a
in
in
g i
ma
ge
s
50 t
r
a
in
in
g i
ma
ge
s
100 tr
a
in
in
g i
ma
ge
s
R
e
s
ul
t
C
onf
id
e
nc
e
R
e
s
ul
t
C
onf
id
e
nc
e
R
e
s
ul
t
C
onf
id
e
nc
e
1
F
a
il
0.4831256707306246
S
te
ve
n
0.8486887517390871
S
te
ve
n
0.8622850226792862
2
F
a
il
0.3968983503374864
F
a
il
0.7912751960836923
S
te
ve
n
0.8963097771048575
3
F
a
il
0.3179188825009534
F
a
il
0.6719356101835616
S
te
ve
n
0.9037249256249601
4
F
a
il
0.3293581751071452
S
te
ve
n
0.8639671038567101
S
te
ve
n
0.8915614756108579
5
F
a
il
0.2915295482988687
S
te
ve
n
0.8393610777619160
S
te
ve
n
0.9189174449524873
6
F
a
il
0.4780330066048949
S
te
ve
n
0.8590177719030167
S
te
ve
n
0.9025752084515857
7
F
a
il
0.6974929480961017
F
a
il
0.7684912796957375
S
te
ve
n
0.902824734897516
8
F
a
il
0.5121738310615238
F
a
il
0.7839105555391056
S
te
ve
n
0.8729555917501845
9
F
a
il
0.3891884170737012
S
te
ve
n
0.8028572961083176
S
te
ve
n
0.8750267206717658
10
F
a
il
0.3088813885614176
S
te
ve
n
0.859101863673292
S
te
ve
n
0.8539075108751068
11
F
a
il
0.3950073892569106
S
te
ve
n
0.8201367392671069
S
te
ve
n
0.9347298547209619
12
F
a
il
0.5725319619748922
S
te
ve
n
0.8491761038671391
S
te
ve
n
0.8986286678563803
13
F
a
il
0.3729178738557185
F
a
il
0.7892751613960386
S
te
ve
n
0.8931622371331799
14
F
a
il
0.5294899956082455
S
te
ve
n
0.8693717639617396
S
te
ve
n
0.8609438571719787
15
F
a
il
0.6741913922990106
S
te
ve
n
0.8193716666671937
S
te
ve
n
0.8821074197564109
T
a
ble
5.
T
e
s
t
r
e
s
ult
of
s
upe
r
a
dmi
nis
tr
a
tor
2
T
e
s
t
N
o.
A
ki
no /
S
upe
r
a
dmi
ni
s
tr
a
to
r
2
20 t
r
a
in
in
g i
ma
ge
s
50 t
r
a
in
in
g i
ma
ge
s
100 tr
a
in
in
g i
ma
ge
s
R
e
s
ul
t
C
onf
id
e
nc
e
R
e
s
ul
t
C
onf
id
e
nc
e
R
e
s
ul
t
C
onf
id
e
nc
e
1
F
a
il
0.6661146706715512
A
ki
no
0.8863968622124053
A
ki
no
0.9206484151507499
2
F
a
il
0.6778212481999762
A
ki
no
0.8609945286854576
A
ki
no
0.9178415651142144
3
F
a
il
0.5895838655488481
F
a
il
0.779757496477541
A
ki
no
0.9038368187576951
4
F
a
il
0.4613156360024367
A
ki
no
0.8807935552948459
A
ki
no
0.9241616389217459
5
F
a
il
0.6045864703957678
A
ki
no
0.8708410150050343
A
ki
no
0.9120598170852047
6
F
a
il
0.5672349222067348
A
ki
no
0.8626607619631554
A
ki
no
0.9206716336519734
7
F
a
il
0.5589413011192211
F
a
il
0.7642900203205136
A
ki
no
0.9343422888699267
8
F
a
il
0.5848885179716995
F
a
il
0.769700825844427
A
ki
no
0.93998698474758
9
F
a
il
0.5244985230576841
A
ki
no
0.8063666092347131
A
ki
no
0.9489666968462456
10
F
a
il
0.6316123637696031
A
ki
no
0.821100171232869
A
ki
no
0.9032190915032803
11
F
a
il
0.6868446788660262
F
a
il
0.7465601333419117
A
ki
no
0.9251503592486126
12
F
a
il
0.6097396524445506
A
ki
no
0.8791109825362003
A
ki
no
0.924164182810119
13
F
a
il
0.6538067013367
A
ki
no
0.8593193087326183
A
ki
no
0.9261999943154746
14
F
a
il
0.6598772619444839
A
ki
no
0.8737677891362153
A
ki
no
0.9182874201435122
15
F
a
il
0.6331449059470585
F
a
il
0.7364376994226135
A
ki
no
0.9063418785474986
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
V
is
ion:
a
w
e
b
s
e
r
v
ice
for
face
r
e
c
ognit
ion
us
ing
c
onv
olut
ional
ne
tw
or
k
(
A
k
ino
A
r
c
hil
les
)
1395
T
a
ble
6.
T
e
s
t
r
e
s
ult
of
a
dmi
nis
tr
a
to
r
1
T
e
s
t
N
o.
O
c
ta
/
A
dmi
ni
s
tr
a
to
r
1
20 t
r
a
in
in
g i
ma
ge
s
50 t
r
a
in
in
g i
ma
ge
s
100 tr
a
in
in
g i
ma
ge
s
R
e
s
ul
t
C
onf
id
e
nc
e
R
e
s
ul
t
C
onf
id
e
nc
e
R
e
s
ul
t
C
onf
id
e
nc
e
1
F
a
il
0.4916995410624423
F
a
il
0.6861709553650175
O
c
ta
0.9032324016883357
2
F
a
il
0.4709441801770915
F
a
il
0.7752983754925702
O
c
ta
0.8599574309961089
3
F
a
il
0.3154287038263585
F
a
il
0.7890147588808165
O
c
ta
0.9092814684895162
4
F
a
il
0.4299015659472068
O
c
ta
0.8597819086345175
O
c
ta
0.8950211711769138
5
F
a
il
0.4018626540552488
O
c
ta
0.8147589534720676
O
c
ta
0.8911998076413719
6
F
a
il
0.388377999272349
F
a
il
0.7499175489710957
O
c
ta
0.8642403121384548
7
F
a
il
0.4605011443261597
F
a
il
0.6935552910483629
O
c
ta
0.8460847480790611
8
F
a
il
0.4079790608866885
O
c
ta
0.8507625601745617
O
c
ta
0.885948001506788
9
F
a
il
0.5120193717264274
F
a
il
0.7826302746185624
O
c
ta
0.8575463693227686
10
F
a
il
0.4017370697546190
F
a
il
0.7819305176010607
O
c
ta
0.9150287792570698
11
F
a
il
0.4150443852165001
F
a
il
0.7498707777169365
O
c
ta
0.9087485643158923
12
F
a
il
0.4158483228853045
O
c
ta
0.8650198573562094
O
c
ta
0.9078024077700945
13
F
a
il
0.3041688954042459
O
c
ta
0.8689157043610945
O
c
ta
0.9039452230759157
14
F
a
il
0.3316467842354488
O
c
ta
0.8109856266988097
O
c
ta
0.8363597838864007
15
F
a
il
0.6101372636222694
F
a
il
0.7947365108563923
O
c
ta
0.8785901336080615
4.
CONC
L
USI
ON
T
he
pr
opos
e
d
f
a
c
e
r
e
c
ognit
ion
mec
ha
nis
m
a
nd
we
b
s
e
r
vice
(
V
is
ion)
ha
ve
be
e
n
s
uc
c
e
s
s
f
ull
y
im
pleme
nted
a
nd
tes
ted
with
the
us
e
of
a
we
b
-
ba
s
e
d
voice
c
ha
tbot
(
J
a
c
ob)
.
T
his
pr
ove
s
that
t
he
de
e
p
c
onvolut
ional
ne
twor
k
c
ould
be
us
e
d
f
or
f
a
c
e
r
e
c
ognit
ion
f
o
r
a
r
e
a
l
-
wor
ld
a
ppli
c
a
ti
on
s
uc
h
a
s
th
e
J
a
c
ob
voice
c
ha
tbot
.
T
he
pe
r
f
o
r
man
c
e
e
va
luation
de
li
ve
r
s
ove
r
a
ll
F
-
s
c
or
e
f
or
S
upe
r
a
dmi
nis
tr
a
t
or
1
a
nd
S
upe
r
a
dmi
nis
tr
a
tor
2,
a
nd
Adm
ini
nis
tr
a
tor
1
o
f
one
.
B
a
s
e
d
on
the
tes
ti
ng
pr
oc
e
s
s
,
it
is
s
hown
that
100
tr
a
ini
ng
im
a
ge
s
give
a
be
tt
e
r
s
uc
c
e
s
s
r
a
te
r
a
ther
than
the
2
0
a
nd
50
tr
a
ini
ng
i
mage
s
.
T
h
e
number
o
f
im
a
ge
s
us
e
d
f
or
tr
a
ini
ng
a
f
f
e
c
ts
the
r
e
c
ognit
ion
c
on
f
idenc
e
r
a
te.
RE
F
E
RE
NC
E
S
[1
]
R.
P.
Mah
ap
at
ra,
et
al
.
,
"
A
d
d
i
n
g
i
n
t
eract
i
v
e
i
n
t
erface
to
E
-
G
o
v
e
r
n
men
t
s
y
s
t
em
s
u
s
i
n
g
A
IML
b
as
e
d
ch
at
t
erb
o
t
s
,
"
2
0
1
2
CS
I
S
i
xt
h
In
t
er
n
a
t
i
o
n
a
l
Co
n
f
e
r
en
ce
on
S
o
f
t
w
a
r
e
E
n
g
i
n
eer
i
n
g
(CO
NS
E
G
)
,
p
p
.
1
-
6,
2
0
1
2
.
[2
]
H
.
N
.
Io
an
d
C.
B.
L
ee,
“
Ch
at
b
o
t
s
an
d
c
o
n
v
ers
a
t
i
o
n
a
l
ag
en
t
s
:
A
b
i
b
l
i
o
me
t
ri
c
a
n
al
y
s
i
s
,
”
i
n
P
r
o
c.
IE
E
E
In
t
er
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
I
n
d
u
s
t
r
i
a
l
E
n
g
i
n
eer
i
n
g
a
n
d
E
n
g
i
n
ee
r
i
n
g
M
a
n
a
g
em
e
n
t
,
p
p
.
2
1
5
-
2
1
9
,
2
0
1
7
.
[3
]
V
.
W
i
l
e
y
an
d
T
.
L
u
cas
,
“Co
mp
u
t
er
V
i
s
i
o
n
an
d
Imag
e
Pro
ces
s
i
n
g
:
A
Pap
er
Rev
i
e
w
,
”
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
A
r
t
i
f
i
c
i
a
l
In
t
e
l
l
i
g
e
n
ce
R
e
s
ea
r
ch
,
v
o
l
.
2
,
n
o
.
1
,
p
p
.
2
9
-
3
6
,
J
u
n
e
2
0
1
8
.
[4
]
E
.
G
.
L
earn
ed
-
Mi
l
l
er
,
“
In
t
r
o
d
u
ct
i
o
n
t
o
Co
m
p
u
t
er
V
i
s
i
o
n
,
”
(2
0
1
1
,
J
an
.
)
,
U
n
i
v
er
s
i
t
y
o
f
Mas
s
ach
u
s
e
t
t
s
.
A
cces
s
e
d
o
n
J
an
u
ary
2
0
,
2
0
1
9
.
[O
n
l
i
n
e].
A
v
ai
l
ab
l
e:
h
t
t
p
s
:
/
/
p
eo
p
l
e.
cs
.
u
mas
s
.
e
d
u
/
~
el
m/
T
eac
h
i
n
g
/
D
o
cs
/
In
t
ro
C
V
_
1
_
1
9
_
1
1
.
p
d
f
[5
]
O
.
H
.
Bab
a
t
u
n
d
e,
et
a
l
.
,
“A
C
o
mp
u
t
er
-
Ba
s
ed
V
i
s
i
o
n
S
y
s
t
ems
fo
r
A
u
t
o
ma
t
i
c
I
d
en
t
i
f
i
cat
i
o
n
o
f
P
l
an
t
Sp
ec
i
es
U
s
i
n
g
K
n
n
an
d
G
en
e
t
i
c
PCA
,
”
Jo
u
r
n
a
l
o
f
A
g
r
i
c
u
l
t
u
r
a
l
In
f
o
r
m
a
t
i
c
s
,
v
o
l
.
6
,
n
o
.
1
,
p
p
.
61
-
71
,
2
0
1
5
.
[6
]
S.
Mat
i
acev
i
ch
,
e
t
a
l
.
,
“Q
u
a
l
i
t
y
Parame
t
ers
o
f
S
i
x
C
u
l
t
i
v
ar
s
o
f
B
l
u
e
b
erry
U
s
i
n
g
C
o
mp
u
t
er
V
i
s
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
F
o
o
d
S
ci
e
n
ce
,
2
0
1
3
.
d
o
i
:
1
0
.
1
1
5
5
/
2
0
1
3
/
4
1
9
5
3
5
[7
]
S.
U
l
l
ma
n
,
et
a
l
.
,
“
A
t
o
ms
o
f
Reco
g
n
i
t
i
o
n
i
n
H
u
man
an
d
C
o
mp
u
t
er
V
i
s
i
o
n
,”
i
n
P
r
o
cee
d
i
n
g
s
o
f
t
h
e
Na
t
i
o
n
a
l
A
ca
d
em
y
o
f
S
c
i
en
ce
s
,
v
o
l
.
1
1
3
,
n
o
.
1
0
,
p
p
.
2
7
4
4
-
2
7
4
9
,
2
0
1
6
.
[8
]
V
.
K
u
rn
i
aw
a
n
,
A
.
W
i
ca
k
s
a
n
a,
an
d
M.
I.
Pras
et
i
y
o
w
a
t
i
,
“
T
h
e
i
m
p
l
em
en
t
at
i
o
n
o
f
ei
g
e
n
face
al
g
o
ri
t
h
m
f
o
r
face
reco
g
n
i
t
i
o
n
i
n
at
t
en
d
an
ce
s
y
s
t
em
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
en
ce
o
n
New
M
ed
i
a
S
t
u
d
i
es
,
p
p
.
1
1
8
-
1
2
4
,
N
o
v
.
2
0
1
7
.
[9
]
K
.
A
l
e
x
an
d
er,
A
.
W
i
ca
k
s
a
n
a,
an
d
N
.
M.
S.
Is
w
ar
i
,
“L
ab
el
i
n
g
A
l
g
o
ri
t
h
m
a
n
d
F
u
l
l
y
C
o
n
n
ect
e
d
N
e
u
ral
N
et
w
o
r
k
fo
r
Aut
o
ma
t
ed
N
u
m
b
er
Pl
at
e
Reco
g
n
i
t
i
o
n
Sy
s
t
em”
i
n
S
t
u
d
i
es
i
n
Co
m
p
u
t
a
t
i
o
n
In
t
e
l
l
i
g
e
n
ce
,
v
o
l
.
8
4
7
,
p
p
1
2
9
-
1
4
5
,
A
u
g
u
s
t
2
0
1
9
.
[1
0
]
F.
Z
h
ao
,
X
.
X
i
e
an
d
M.
Ro
ac
h
,
“Co
m
p
u
t
er
V
i
s
i
o
n
T
e
ch
n
i
q
u
es
f
o
r
T
ra
n
cat
h
et
er
In
t
erv
e
n
t
i
o
n
,
”
IE
E
E
J
o
u
r
n
a
l
o
f
Tr
a
n
s
l
a
t
i
o
n
a
l
E
n
g
i
n
ee
r
i
n
g
i
n
H
e
a
l
t
h
a
n
d
M
ed
i
ci
n
e
,
v
o
l
.
3
,
p
p
.
1
-
3
1
,
J
u
n
e
2
0
1
5
.
[1
1
]
A
.
Maj
u
md
ar
an
d
R.
K
.
W
ard
,
“
Ps
eu
d
o
-
F
i
s
h
erface
Met
h
o
d
fo
r
Si
n
g
l
e
Imag
e
Per
Pers
o
n
Face
Reco
g
n
i
t
i
o
n
,
”
i
n
I
E
E
E
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
A
co
u
s
t
i
c
s
,
S
p
eec
h
a
n
d
S
i
g
n
a
l
P
r
o
ce
s
s
i
n
g
,
p
p
.
9
8
9
-
9
9
2
, 2
0
0
8
.
[1
2
]
C.
Z
h
en
an
d
Y
.
Su
,
“
Res
each
ab
o
u
t
h
u
ma
n
face
reco
g
n
i
t
i
o
n
t
ec
h
n
o
l
o
g
y
,
”
i
n
In
t
e
r
n
a
t
i
o
n
a
l
Co
n
f
er
e
n
ce
o
n
Tes
t
a
n
d
M
ea
s
u
r
em
e
n
t
,
p
p
.
4
2
0
-
4
2
2
,
2
0
0
9
.
[1
3
]
S.
W
i
j
ay
a
an
d
A
.
W
i
ca
k
s
a
n
a,
“J
aco
b
V
o
i
ce
Ch
at
b
o
t
A
p
p
l
i
cat
i
o
n
U
s
i
n
g
W
i
t
.
ai
fo
r
Pro
v
i
d
i
n
g
In
fo
rma
t
i
o
n
i
n
U
M
N
,
”
In
t
e
r
n
a
t
i
o
n
a
l
Jo
u
r
n
a
l
o
f
E
n
g
i
n
ee
r
i
n
g
a
n
d
A
d
v
a
n
ce
d
Tec
h
n
o
l
o
g
y
,
v
o
l
.
8
,
n
o
.
6
S3
,
p
p
.
1
0
5
-
1
0
9
.
2
0
1
9
.
[1
4
]
Y
.
L
ecu
n
,
et
al
.
,
“
G
rad
i
en
t
-
Ba
s
ed
L
earn
i
n
g
A
p
p
l
i
e
d
t
o
D
o
cu
me
n
t
Reco
g
n
i
t
i
o
n
,
”
in
P
r
o
ceed
i
n
g
s
o
f
t
h
e
IE
E
E
,
v
o
l
.
8
6
,
n
o
.
1
1
,
p
p
.
2
2
7
8
-
2
3
2
4
,
N
o
v
em
b
er
1
9
9
8
.
[1
5
]
S.
L
a
w
ren
ce,
et
al
.
,
“Face
reco
g
n
i
t
i
o
n
:
a
co
n
v
o
l
u
t
i
o
n
al
n
eu
ral
-
n
et
w
o
r
k
ap
p
r
o
ach
,
”
IE
E
E
Tr
a
n
s
a
ct
i
o
n
s
o
n
Neu
r
a
l
Net
wo
r
ks
,
v
o
l
.
8
,
n
o
.
1
,
p
p
.
9
8
-
1
1
3
,
J
a
n
u
ar
y
1
9
9
7
.
[1
6
]
K
.
O
’Sh
ea
an
d
R.
N
as
h
,
“A
n
In
t
ro
d
u
c
t
i
o
n
t
o
Co
n
v
o
l
u
t
i
o
n
a
l
N
eu
ral
N
e
t
w
o
rk
s
,
”
arX
i
v
:
1
5
1
1
.
0
8
4
5
8
v
2
[cs
.
N
E
],
D
ec
emb
er
2
0
1
5
.
[1
7
]
F.
Sch
ro
ff,
D
.
K
al
e
n
i
c
h
en
k
o
a
n
d
J
.
Ph
i
l
b
i
n
,
“
FaceN
e
t
:
A
U
n
i
fi
e
d
E
m
b
ed
d
i
n
g
f
o
r
Face
Rec
o
g
n
i
t
i
o
n
a
n
d
C
l
u
s
t
er
i
n
g
,
”
in
IE
E
E
C
o
n
f
er
e
n
ce
o
n
C
o
m
p
u
t
er
V
i
s
i
o
n
a
n
d
P
a
t
t
er
n
R
e
co
g
n
i
t
i
o
n
,
p
p
.
8
1
5
-
8
2
3
,
2
0
1
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
1693
-
6930
T
E
L
KO
M
NI
KA
T
e
lec
omm
un
C
omput
E
l
C
ontr
o
l
,
Vol.
18
,
No
.
3
,
J
une
2020:
13
89
-
13
96
1396
[1
8
]
Y
.
Su
n
,
X
.
W
an
g
,
an
d
X
.
T
a
n
g
,
“
D
ee
p
l
y
l
ear
n
e
d
face
rep
res
e
n
t
a
t
i
o
n
s
are
s
p
ars
e,
s
e
l
ect
i
v
e,
a
n
d
r
o
b
u
s
t
,
”
i
n
I
E
E
E
Co
n
f
er
e
n
ce
o
n
C
o
m
p
u
t
er
V
i
s
i
o
n
a
n
d
P
a
t
t
er
n
R
ec
o
g
n
i
t
i
o
n
,
p
p
.
8
1
5
-
8
2
3
,
2
0
1
5
.
[1
9
]
H.
M.
Mo
o
n
a
n
d
S.
B.
Pan
,
“
T
h
e
PC
A
-
Bas
e
d
L
o
n
g
D
i
s
t
a
n
ce
Face
Reco
g
n
i
t
i
o
n
u
s
i
n
g
Mu
l
t
i
p
l
e
D
i
s
t
an
ce
T
ra
i
n
i
n
g
Imag
es
fo
r
In
t
el
l
i
g
en
t
Su
rv
e
i
l
l
an
ce
S
y
s
t
em
,
”
i
n
1
st
In
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
er
e
n
ce
o
n
In
f
o
r
m
a
t
i
o
n
a
n
d
Co
m
m
u
n
i
c
a
t
i
o
n
Tech
n
o
l
o
g
y
,
p
p
.
5
3
4
-
5
3
9
,
2
0
1
3
.
[2
0
]
T
h
e
N
a
t
i
o
n
a
l
O
p
t
i
cal
A
s
t
ro
n
o
m
y
O
b
s
er
v
at
o
ry
,
“
Reco
m
men
d
e
d
L
i
g
h
t
L
ev
el
s
(Il
l
u
m
i
n
a
n
ce)
fo
r
O
u
t
d
o
o
r
a
n
d
I
n
d
o
o
r
V
en
u
es
,
”
A
meri
ca,
2
0
1
9
.
[O
n
l
i
n
e].
A
v
ai
l
ab
l
e:
h
t
t
p
s
:
/
/
w
w
w
.
n
o
a
o
.
ed
u
/
e
d
u
ca
t
i
o
n
/
Q
L
T
k
i
t
/
A
CT
I
V
IT
Y
_
D
o
c
u
me
n
t
s
/
Safet
y
/
L
i
g
h
t
L
ev
e
l
s
_
o
u
t
d
o
o
r+
i
n
d
o
o
r.
p
d
f.
A
cces
s
ed
:
Feb
r
u
ary
3
,
2
0
1
9
.
[2
1
]
P.
Ch
arl
es
an
d
S.
L
.
Pfl
eeg
er,
“
A
n
al
y
zi
n
g
C
o
mp
u
t
er
S
ecu
ri
t
y
:
A
T
h
reat
/
v
u
l
n
erab
i
l
i
t
y
/
c
o
u
n
t
ermea
s
u
re
A
p
p
ro
a
ch
,
”
2
n
d
e
d
.
U
p
p
er
Sad
d
l
e
Ri
v
er,
N
J
:
Pears
o
n
Pren
t
i
ce
H
al
l
,
2
0
1
2
.
[2
2
]
C.
D
.
Man
n
i
n
g
an
d
H
.
Sch
u
t
ze,
“
Fo
u
n
d
a
t
i
o
n
s
o
f
St
at
i
s
t
i
cal
N
a
t
u
ra
l
L
an
g
u
ag
e
Pr
o
ce
s
s
i
n
g
,
”
2
n
d
ed
.
L
o
n
d
o
n
,
T
h
e
MIT
Pre
s
s
.
1
9
9
9
.
[2
3
]
C.
Szeg
ed
y
,
S.
Io
ffe,
V
.
V
an
h
o
u
ck
e,
A
.
A
.
A
l
em
i
,
“
In
cep
t
i
o
n
-
v
4
,
In
cep
t
i
o
n
-
R
es
N
et
an
d
t
h
e
Imp
ac
t
o
f
Res
i
d
u
a
l
Co
n
n
ect
i
o
n
s
o
n
L
earn
i
n
g
,
”
i
n
A
A
A
I
Co
n
f
e
r
en
ce
o
n
A
r
t
i
f
i
ci
a
l
In
t
e
l
l
i
g
e
n
ce
,
Feb
ru
ar
y
2
0
1
7
.
[2
4
]
D
.
San
d
b
erg
.
“
Face
Reco
g
n
i
t
i
o
n
u
s
i
n
g
T
en
s
o
rf
l
o
w
,
”
G
i
t
H
u
b
.
A
cces
s
ed
o
n
2
3
J
a
n
u
a
ry
2
0
1
8
.
[
O
n
l
i
n
e].
A
v
ai
l
a
b
l
e
:
h
t
t
p
s
:
/
/
g
i
t
h
u
b
.
co
m
/
d
a
v
i
d
s
a
n
d
b
erg
/
facen
e
t
[2
5
]
K
.
Z
h
an
g
,
e
t
al
.
,
“J
o
i
n
t
Face
D
e
t
ect
i
o
n
an
d
A
l
i
g
n
m
en
t
U
s
i
n
g
Mu
l
t
i
t
a
s
k
Ca
s
cad
e
d
Co
n
v
o
l
u
t
i
o
n
al
N
et
w
o
r
k
s
,
”
I
E
E
E
S
i
g
n
a
l
P
r
o
ce
s
s
i
n
g
Let
t
er
s
S
o
c
i
et
y
,
v
o
l
.
2
3
,
n
o
.
1
0
,
p
p
.
1
4
9
9
-
1
5
0
3
,
O
c
t
o
b
er
2
0
1
6
.
[2
6
]
A
.
K
h
o
s
r
o
zad
e
h
,
et
al
.
,
“Pres
e
n
t
i
n
g
a
N
o
v
el
Pa
g
e
Rep
l
acemen
t
A
l
g
o
ri
t
h
m
Ba
s
ed
o
n
L
RU
,
”
Jo
u
r
n
a
l
o
f
B
a
s
i
c
a
n
d
A
p
p
l
i
ed
S
c
i
en
t
i
f
i
c
R
es
e
a
r
c
h
,
v
o
l
.
2
,
n
o
.
1
0
,
p
p
.
1
0
3
7
7
-
1
0
3
8
3
,
2
0
1
2
.
[2
7
]
E
.
J
.
O
’N
ei
l
,
P.
E
.
O
’N
ei
l
an
d
G
.
W
ei
k
u
m,
“A
n
o
p
t
i
mal
i
t
y
p
r
o
o
f
o
f
t
h
e
L
RU
-
K
p
ag
e
re
p
l
aceme
n
t
al
g
o
r
i
t
h
m
,
”
Jo
u
r
n
a
l
o
f
t
h
e
A
CM
,
v
o
l
.
4
6
,
n
o
.
1
,
p
p
.
9
2
-
1
1
2
,
J
an
u
ary
1
9
9
9
.
B
I
OG
RA
P
HI
E
S
OF
AU
T
HO
RS
A
ki
no
A
rchi
l
l
es
i
s
g
rad
u
at
e
d
fro
m
U
n
i
v
er
s
i
t
as
M
u
l
t
i
me
d
i
a
N
u
s
an
t
ara
as
Sar
j
an
a
K
o
mp
u
t
er
(S.
K
o
m.
).
A
n
ew
res
earch
er
en
t
h
u
s
ed
i
n
n
e
u
r
al
n
et
w
o
rk
s
.
Cu
rren
t
l
y
w
o
rk
i
n
g
as
a
s
o
f
t
w
are
en
g
i
n
eer
an
d
A
I
res
earc
h
er
i
n
o
n
e
o
f
t
h
e
b
i
g
g
e
s
t
In
d
o
n
es
i
a
ret
a
i
l
co
m
p
an
y
.
A
ry
a
Wi
c
a
ks
a
na
is
g
rad
u
at
e
d
fr
o
m
U
n
i
v
ers
i
t
i
T
u
n
k
u
A
b
d
u
l
Ra
h
man
i
n
V
L
SI
E
n
g
i
n
eeri
n
g
(M.
E
n
g
.
Sc.
)
an
d
U
n
i
v
er
s
i
t
as
Mu
l
t
i
med
i
a
N
u
s
an
t
ara
i
n
C
o
mp
u
t
er
Sci
e
n
ce
(S.
K
o
m.
).
Res
earch
i
n
t
eres
t
s
a
n
d
w
o
r
k
s
are
:
q
u
an
t
u
m
co
mp
u
t
i
n
g
an
d
co
m
p
u
t
at
i
o
n
a
l
i
n
t
el
l
i
g
en
ce.
Evaluation Warning : The document was created with Spire.PDF for Python.