T
E
L
K
O
M
NIKA
T
elec
o
mm
un
ica
t
io
n,
Co
m
pu
t
ing
,
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
18
,
No
.
2
,
A
p
r
il 2
0
2
0
,
p
p
.
8
3
0
~
8
3
8
I
SS
N:
1
6
9
3
-
6
9
3
0
,
ac
cr
ed
ited
First Gr
ad
e
b
y
Ke
m
e
n
r
is
te
k
d
i
k
ti,
Dec
r
ee
No
: 2
1
/E/KPT
/2
0
1
8
DOI
: 1
0
.
1
2
9
2
8
/
T
E
L
KOM
NI
KA
.
v
1
8
i2
.
1
4
8
3
8
830
J
o
ur
na
l ho
m
ep
a
g
e
:
h
tt
p
:
//jo
u
r
n
a
l.u
a
d
.
a
c.
id
/in
d
ex
.
p
h
p
/TELK
OM
N
I
K
A
Sing
le objec
t
de
te
ction to sup
po
rt
r
equire
m
en
ts
m
o
deling
using
f
a
ster R
-
CN
N
Na
t
ha
na
el
G
ilb
er
t
,
Andre
R
us
li
De
p
a
rtme
n
t
o
f
In
f
o
rm
a
ti
c
s,
Un
iv
e
rsitas
M
u
lt
im
e
d
ia Nu
sa
n
tara
,
In
d
o
n
e
sia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l 5
,
2
0
1
9
R
ev
i
s
ed
J
an
7
,
2
0
2
0
A
cc
ep
ted
Feb
1
9
,
2
0
2
0
Re
q
u
irem
e
n
ts
e
n
g
in
e
e
rin
g
(RE)
is
o
n
e
o
f
th
e
m
o
st
i
m
p
o
rtan
t
p
h
a
se
s
o
f
a
so
f
twa
re
e
n
g
in
e
e
rin
g
p
ro
jec
t
i
n
w
h
ich
th
e
f
o
u
n
d
a
ti
o
n
o
f
a
so
f
tw
a
re
p
ro
d
u
c
t
is
laid
,
o
b
jec
ti
v
e
s
a
n
d
a
ss
u
m
p
ti
o
n
s,
f
u
n
c
ti
o
n
a
l
a
n
d
n
o
n
-
f
u
n
c
ti
o
n
a
l
n
e
e
d
s
a
re
a
n
a
ly
z
e
d
a
n
d
c
o
n
so
li
d
a
ted
.
M
a
n
y
m
o
d
e
li
n
g
n
o
tatio
n
s a
n
d
t
o
o
ls are
d
e
v
e
lo
p
e
d
to
m
o
d
e
l
t
h
e
i
n
f
o
rm
a
ti
o
n
g
a
th
e
re
d
i
n
t
h
e
RE
p
ro
c
e
ss
,
o
n
e
p
o
p
u
lar
f
ra
m
e
w
o
r
k
is
th
e
iS
tar
2
.
0
.
De
sp
it
e
t
h
e
f
ra
m
e
w
o
rk
s
a
n
d
n
o
tatio
n
s
th
a
t
a
re
in
tro
d
u
c
e
d
,
m
a
n
y
e
n
g
in
e
e
rs
stil
l
f
in
d
th
a
t
d
ra
w
in
g
th
e
d
iag
ra
m
s
is
e
a
sie
r
d
o
n
e
m
a
n
u
a
ll
y
b
y
h
a
n
d
.
P
r
o
b
lem
a
rise
s
w
h
e
n
th
e
c
o
rre
sp
o
n
d
in
g
d
iag
ra
m
n
e
e
d
s
to
b
e
u
p
d
a
te
d
a
s
re
q
u
irem
e
n
ts
e
v
o
lv
e
.
T
h
is
re
se
a
rc
h
a
i
m
s
to
k
ick
sta
rt
th
e
d
e
v
e
lo
p
m
e
n
t
o
f
a
m
o
d
e
li
n
g
to
o
l
u
si
n
g
F
a
ste
r
Re
g
io
n
-
b
a
se
d
C
o
n
v
o
lu
ti
o
n
a
l
Ne
u
ra
l
Ne
t
w
o
rk
f
o
r
sin
g
le
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
re
c
o
g
n
it
i
o
n
o
f
h
a
n
d
-
d
ra
w
n
iS
tar
2
.
0
o
b
jec
ts,
G
lea
m
g
ra
y
sc
a
le,
a
n
d
S
a
lt
a
n
d
P
e
p
p
e
r
n
o
ise
to
d
ig
it
a
li
z
e
h
a
n
d
-
d
ra
w
n
d
iag
ra
m
s.
T
h
e
sin
g
le
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
re
c
o
g
n
it
io
n
t
o
o
l
is
e
v
a
lu
a
ted
a
n
d
d
isp
lay
s
p
ro
m
isin
g
re
su
lt
s
o
f
a
n
o
v
e
ra
ll
a
c
c
u
ra
c
y
a
n
d
p
re
c
isio
n
o
f
9
5
%
,
1
0
0
%
f
o
r
re
c
a
ll
,
a
n
d
9
7
.
2
%
f
o
r
th
e
F
-
1
sc
o
re
.
K
ey
w
o
r
d
s
:
Fas
ter
R
-
C
NN
iSt
ar
2
.
0
Ob
j
ec
t d
etec
tio
n
an
d
r
ec
o
g
n
itio
n
R
eq
u
ir
e
m
en
t
s
m
o
d
eli
n
g
to
o
l
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r
th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
An
d
r
e
R
u
s
li
,
Dep
ar
t
m
en
t o
f
I
n
f
o
r
m
atic
s
,
Un
i
v
er
s
ita
s
Mu
l
ti
m
ed
ia
N
u
s
a
n
tar
a
,
Ka
m
p
u
s
UM
N,
Scie
n
tia
Gar
d
en
,
J
l.
B
o
u
lev
ar
d
Gad
in
g
Ser
p
o
n
g
,
T
an
g
er
an
g
,
B
an
te
n
,
1
5
8
1
0
,
I
n
d
o
n
esia
.
E
m
ail:
a
n
d
r
e.
r
u
s
li
@
u
m
n
.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
B
r
o
ad
ly
s
p
ea
k
i
n
g
,
s
o
f
t
w
ar
e
s
y
s
te
m
s
r
eq
u
ir
e
m
e
n
ts
en
g
i
n
ee
r
in
g
(
R
E
)
is
t
h
e
p
r
o
ce
s
s
o
f
d
is
co
v
er
in
g
th
a
t
p
u
r
p
o
s
e,
b
y
id
en
ti
f
y
i
n
g
s
ta
k
e
h
o
ld
er
s
an
d
t
h
eir
n
ee
d
s
a
n
d
d
o
cu
m
en
tin
g
th
e
s
e
i
n
a
f
o
r
m
th
at
is
a
m
e
n
ab
le
to
an
al
y
s
i
s
,
co
m
m
u
n
icatio
n
,
an
d
s
u
b
s
eq
u
e
n
t
i
m
p
le
m
e
n
ta
ti
o
n
[
1
]
.
T
h
e
i
m
p
o
r
tan
ce
o
f
R
E
is
e
m
p
h
asized
to
d
ev
elo
p
ef
f
ec
ti
v
e
s
o
f
t
w
ar
e
an
d
r
ed
u
ce
s
o
f
t
w
ar
e
m
i
s
tak
e
s
in
th
e
ea
r
l
y
s
ta
g
e
o
f
s
o
f
t
w
ar
e
d
ev
elo
p
m
e
n
t
[
2
]
.
R
eq
u
ir
e
m
en
t
s
m
o
d
eli
n
g
u
s
es
a
co
m
b
in
atio
n
o
f
te
x
t
a
n
d
d
iag
r
a
m
m
atic
f
o
r
m
s
to
d
ep
ict
r
eq
u
ir
e
m
e
n
ts
in
a
w
a
y
th
at
is
r
elati
v
el
y
ea
s
y
to
u
n
d
er
s
tan
d
,
an
d
m
o
r
e
i
m
p
o
r
t
an
t,
s
tr
aig
h
t
f
o
r
w
ar
d
to
r
ev
iew
f
o
r
co
r
r
ec
tn
ess
,
co
m
p
lete
n
e
s
s
,
a
n
d
co
n
s
i
s
ten
c
y
[
3
]
.
I
n
an
al
y
zi
n
g
s
o
f
t
w
ar
e
r
eq
u
ir
e
m
e
n
ts
,
a
f
ter
t
h
e
d
o
m
ai
n
is
u
n
d
er
s
to
o
d
an
d
elicited
,
r
eq
u
ir
e
m
e
n
t
s
ar
e
e
v
alu
ated
a
n
d
n
e
g
o
ti
ated
,
t
h
e
n
th
e
co
n
s
o
lid
ated
r
eq
u
ir
e
m
e
n
ts
ar
e
s
p
ec
if
icat
io
n
s
p
ec
if
ied
a
n
d
d
o
cu
m
e
n
ted
[
4
]
.
T
h
is
r
eq
u
ir
e
m
en
t
s
s
p
ec
i
f
ic
atio
n
an
d
d
o
cu
m
e
n
tatio
n
is
w
h
er
e
r
eq
u
ir
e
m
e
n
t
s
m
o
d
eli
n
g
co
m
m
o
n
l
y
o
cc
u
r
s
.
T
h
r
o
u
g
h
o
u
t
r
eq
u
ir
e
m
e
n
ts
m
o
d
elin
g
,
t
h
e
p
r
im
ar
y
f
o
c
u
s
is
o
n
w
h
at,
n
o
t
h
o
w
,
o
n
iSt
ar
2
.
0
’
s
s
tr
ate
g
ic
d
ep
en
d
en
c
y
m
o
d
el,
th
e
f
o
cu
s
i
s
o
n
d
escr
ib
in
g
th
e
d
ep
en
d
e
n
c
y
r
elatio
n
s
h
ip
b
et
w
ee
n
ea
ch
ac
to
r
in
th
e
s
y
s
te
m
,
alo
n
g
w
it
h
t
h
e
in
te
n
tio
n
al
ele
m
e
n
ts
.
I
n
t
h
e
r
eq
u
ir
e
m
e
n
ts
e
n
g
i
n
ee
r
in
g
co
m
m
u
n
i
t
y
,
iSt
ar
2
.
0
is
g
ain
i
n
g
tr
ac
tio
n
b
o
th
in
t
h
e
ac
ad
e
m
ical
a
n
d
in
d
u
s
tr
ial
f
ie
ld
s
an
d
is
u
s
ed
b
y
m
a
n
y
p
la
y
er
s
i
n
th
e
co
m
m
u
n
it
y
[
5
]
.
T
h
e
f
r
am
e
w
o
r
k
i
s
ap
p
lied
an
d
i
m
p
l
e
m
en
ted
in
v
ar
io
u
s
s
ec
to
r
s
,
s
u
c
h
as
h
ea
l
th
ca
r
e,
s
ec
u
r
it
y
an
al
y
s
is
,
an
d
e
C
o
m
m
er
ce
[
6
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
S
in
g
le
o
b
ject
d
e
tectio
n
to
s
u
p
p
o
r
t req
u
ir
eme
n
ts
…
(
N
a
th
a
n
a
el
Gilb
ert
)
831
W
h
en
m
o
d
eli
n
g
r
eq
u
ir
e
m
en
t
s
an
d
d
esig
n
in
g
s
o
f
t
w
ar
e
p
r
o
d
u
cts,
m
an
y
en
g
i
n
ee
r
s
s
til
l
r
eso
r
t
to
d
r
a
w
in
g
th
e
d
iag
r
a
m
s
m
a
n
u
all
y
b
y
h
an
d
i
n
s
tead
o
f
u
s
i
n
g
s
o
f
t
w
ar
e
to
o
ls
.
O
n
e
r
ea
s
o
n
co
u
ld
b
e
t
h
at
h
an
d
-
d
r
a
w
in
g
t
h
e
d
iag
r
a
m
s
co
u
ld
lead
to
m
o
r
e
f
o
cu
s
ed
w
o
r
k
a
n
d
le
s
s
d
is
tr
ac
t
io
n
[
7
]
.
Ho
w
e
v
er
,
in
a
s
u
s
tai
n
ab
le
p
r
o
j
ec
t
w
i
th
co
n
ti
n
u
o
u
s
r
ev
i
s
io
n
s
ca
u
s
ed
b
y
r
eq
u
ir
e
m
e
n
ts
ev
o
l
u
tio
n
,
i
t
g
r
ad
u
al
l
y
b
ec
a
m
e
ap
p
ar
en
t
th
at
t
h
e
d
ig
italizatio
n
o
f
t
h
e
h
a
n
d
-
d
r
a
w
n
d
ia
g
r
a
m
i
s
ess
e
n
tia
l
i
n
an
e
v
er
-
ev
o
lv
in
g
r
eq
u
ir
e
m
e
n
ts
en
g
i
n
ee
r
i
n
g
ac
ti
v
it
ies.
On
e
o
f
th
e
f
ir
s
t
s
tep
s
in
d
ia
g
r
a
m
d
ig
italiza
t
io
n
is
o
b
j
ec
t
d
etec
tio
n
an
d
r
ec
o
g
n
i
tio
n
.
Ob
j
ec
t
d
etec
tio
n
an
d
r
ec
o
g
n
i
ti
o
n
ai
m
to
d
etec
t
a
n
d
r
ec
o
g
n
iz
e
ev
er
y
o
b
j
ec
t
b
elo
n
g
i
n
g
to
a
k
n
o
w
n
cla
s
s
i
n
a
n
i
m
a
g
e
[
8
]
.
Sev
er
al
p
iece
s
o
f
r
esear
ch
h
a
v
e
s
h
o
w
n
th
e
ab
ili
t
y
o
f
t
h
e
ad
v
a
n
ce
d
n
e
u
r
al
n
et
wo
r
k
s
in
i
m
a
g
e/o
b
j
ec
t
r
ec
o
g
n
itio
n
[
9
,
1
0
,
1
1
]
;
h
en
ce
f
o
r
th
,
t
h
is
r
esear
c
h
m
ea
n
t
to
u
tili
ze
n
e
u
r
al
n
et
w
o
r
k
ar
ch
ite
ctu
r
e
to
i
m
p
le
m
e
n
t
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es
to
d
etec
t
an
d
r
ec
o
g
n
ize
o
b
j
ec
ts
in
t
h
e
r
eq
u
ir
e
m
e
n
ts
d
iag
r
a
m
.
I
n
th
e
m
ac
h
i
n
e
lear
n
in
g
f
ield
,
th
e
R
eg
io
n
-
b
as
ed
C
o
n
v
o
lu
t
io
n
al
Ne
u
r
al
Netw
o
r
k
(
R
-
C
NN)
ar
ch
itect
u
r
e
is
a
p
o
p
u
lar
m
eth
o
d
w
it
h
p
r
o
m
is
i
n
g
p
er
f
o
r
m
a
n
ce
.
T
h
e
r
ap
id
g
r
o
w
t
h
h
as
p
r
o
p
o
s
ed
th
e
cu
r
r
en
tl
y
k
n
o
w
n
Fas
ter
R
-
C
N
N
(
f
r
o
m
it
s
p
r
ed
ec
ess
o
r
s
,
th
e
R
-
C
NN,
a
n
d
t
h
e
Fa
s
t
R
-
C
NN)
w
ith
b
etter
ac
cu
r
ac
y
a
n
d
p
r
o
ce
s
s
in
g
[
1
2
]
.
Oth
e
r
r
esear
ch
al
s
o
d
is
p
la
y
s
t
h
e
p
o
ten
tial o
f
Fas
ter
R
-
C
N
N
t
o
d
etec
t a
n
o
b
j
ec
t in
an
i
m
ag
e
w
it
h
h
i
g
h
ac
c
u
r
ac
y
w
it
h
t
h
e
co
r
r
ec
t d
ataset
[
1
3
]
.
Fu
r
t
h
er
m
o
r
e,
i
m
a
g
e
p
r
e
-
p
r
o
ce
s
s
i
n
g
al
s
o
h
o
ld
s
a
v
ital
r
o
le
in
p
r
o
ce
s
s
in
g
d
atasets
in
o
b
j
ec
t
d
etec
tio
n
[
1
4
]
.
On
e
s
tan
d
a
r
d
p
r
o
ce
s
s
is
th
e
co
lo
r
-
to
-
g
r
a
y
s
c
ale
tech
n
iq
u
e.
Gr
a
y
s
ca
le
i
m
a
g
es
ar
e
i
m
ag
e
s
w
it
h
o
n
l
y
h
a
v
e
a
s
i
n
g
le
v
al
u
e
f
o
r
it
s
ev
er
y
p
i
x
el,
r
es
u
lti
n
g
i
n
a
g
r
e
y
i
m
a
g
e,
w
h
ic
h
te
n
d
s
to
b
e
b
lack
o
n
p
ix
e
ls
w
it
h
w
ea
k
in
te
n
s
it
y
a
n
d
w
h
ite
o
n
p
ix
els
w
it
h
h
i
g
h
in
te
n
s
i
t
y
[
1
5
]
.
T
h
is
r
esear
ch
u
s
es
Glea
m
as
th
e
g
r
e
y
s
ca
lin
g
m
et
h
o
d
,
as
it
is
ar
g
u
ed
th
at
co
m
p
ar
ed
to
o
th
er
tech
n
iq
u
es,
Glea
m
p
er
f
o
r
m
s
b
etter
[
1
6
]
.
Fu
r
t
h
er
m
o
r
e,
to
p
er
f
o
r
m
u
p
s
a
m
p
lin
g
o
f
t
h
e
d
ataset
to
w
ar
d
s
a
h
ig
h
-
p
er
f
o
r
m
i
n
g
m
o
d
el,
Sa
lt
an
d
P
ep
p
er
n
o
is
e
is
u
ti
lized
f
o
r
its
a
b
ilit
y
to
r
ep
licate
i
m
ag
e
d
ata
w
it
h
d
i
f
f
er
en
ce
s
b
y
in
s
er
tin
g
w
r
o
n
g
b
it
tr
an
s
m
i
s
s
io
n
an
d
an
alo
g
to
d
ig
i
tal
co
n
v
er
s
i
o
n
[
1
7
]
.
T
h
is
p
ap
er
r
ep
o
r
ts
th
e
r
es
u
lt
o
f
th
e
ea
r
l
y
s
t
u
d
y
w
h
i
ch
ai
m
s
to
i
m
p
le
m
e
n
t
a
n
d
ev
alu
a
te
th
e
p
er
f
o
r
m
a
n
ce
o
f
Fas
ter
R
-
C
NN,
Glea
m
,
an
d
Salt
a
n
d
P
ep
p
e
r
tech
n
iq
u
e
f
o
r
s
in
g
le
o
b
j
ec
t
d
etec
tio
n
an
d
r
ec
o
g
n
itio
n
i
n
a
h
an
d
-
d
r
a
w
n
iSt
ar
2
.
0
s
tr
ateg
ic
d
ep
e
n
d
en
c
y
m
o
d
el
f
o
r
r
eq
u
ir
e
m
e
n
ts
m
o
d
eli
n
g
.
T
h
e
m
o
d
el
’
s
p
er
f
o
r
m
a
n
ce
is
m
ea
s
u
r
ed
b
y
ca
lcu
lati
n
g
t
h
e
p
r
ec
is
io
n
,
ac
c
u
r
ac
y
,
r
ec
all,
a
n
d
F
-
m
ea
s
u
r
e
w
h
e
n
cl
ass
i
f
y
in
g
t
h
e
n
o
tatio
n
o
f
iSt
a
r
2
.
0
s
y
m
b
o
ls
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
I
n
co
n
d
u
ct
in
g
t
h
e
r
esear
ch
to
i
m
p
le
m
en
t
an
d
e
v
al
u
ate
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
Fas
ter
R
-
C
N
N,
Glea
m
,
an
d
Salt
an
d
P
ep
p
er
tech
n
iq
u
e
to
f
o
r
s
in
g
le
o
b
j
ec
t
d
etec
tio
n
an
d
r
ec
o
g
n
i
tio
n
i
n
a
h
a
n
d
-
d
r
a
w
n
i
Star
2
.
0
s
tr
ateg
ic
d
ep
en
d
en
c
y
m
o
d
el
f
o
r
r
eq
u
ir
em
e
n
t
s
m
o
d
eli
n
g
,
t
h
e
r
esear
ch
m
e
th
o
d
o
lo
g
ies ar
e
as
f
o
llo
w
s
.
-
L
iter
at
u
r
e
r
ev
ie
w
a
n
d
r
eq
u
ir
em
en
ts
a
n
al
y
s
is
,
-
E
x
p
er
i
m
e
n
t a
n
d
s
y
s
te
m
d
esi
g
n
,
-
S
y
s
te
m
co
n
s
tr
u
ctio
n
a
n
d
co
d
in
g
,
-
T
esti
n
g
an
d
e
v
alu
a
tio
n
,
an
d
-
R
esear
ch
d
o
cu
m
e
n
tatio
n
.
Firstl
y
,
liter
at
u
r
e
r
e
v
ie
w
a
n
d
r
eq
u
ir
e
m
en
ts
an
al
y
s
is
ac
tiv
iti
es
ar
e
co
n
d
u
cted
to
d
e
f
i
n
e
t
h
e
p
r
o
b
lem
,
th
en
p
r
o
p
o
s
e
a
s
o
lu
tio
n
,
in
th
i
s
ca
s
e,
d
ec
id
in
g
th
e
m
o
s
t
s
u
ita
b
le
m
et
h
o
d
s
an
d
p
r
ac
tices.
Seco
n
d
l
y
,
af
ter
w
o
r
k
s
o
f
liter
at
u
r
e
ar
e
r
ev
ie
w
ed
,
a
n
d
p
r
o
b
lem
s
ar
e
d
ef
in
ed
t
h
e
ar
ch
itect
u
r
e
an
d
s
y
s
te
m
d
es
ig
n
is
d
o
n
e
u
s
i
n
g
f
lo
w
ch
ar
t
s
to
d
esi
g
n
th
e
f
lo
w
o
f
th
e
s
tep
s
co
n
d
u
cted
i
n
t
h
e
o
b
j
ec
t
d
etec
tio
n
an
d
r
ec
o
g
n
iti
o
n
p
r
o
g
r
a
m
a
n
d
UI
m
o
c
k
u
p
s
f
o
r
tes
tin
g
p
u
r
p
o
s
es.
T
h
en
t
h
e
d
esi
g
n
ed
s
y
s
te
m
is
co
n
s
tr
u
cted
,
a
n
d
te
s
ti
n
g
is
co
n
d
u
cted
to
e
v
al
u
at
e
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
m
a
ch
in
e
lear
n
i
n
g
m
o
d
el.
L
astl
y
,
all
th
e
ac
ti
v
itie
s
co
n
d
u
ct
ed
in
th
e
r
esear
ch
is
d
o
cu
m
e
n
ted
.
2
.
1
.
iSta
r
2
.
0
T
h
e
i
*
la
n
g
u
a
g
e
w
a
s
p
r
ese
n
te
d
in
t
h
e
m
id
-
n
in
et
ies
[
18
]
as
a
g
o
al
-
a
n
d
ac
to
r
-
o
r
ien
ted
m
o
d
elin
g
an
d
r
ea
s
o
n
i
n
g
f
r
a
m
e
w
o
r
k
.
I
t
co
n
s
is
t
s
o
f
a
m
o
d
eli
n
g
lan
g
u
ag
e
alo
n
g
w
it
h
r
ea
s
o
n
in
g
tec
h
n
i
q
u
es
f
o
r
an
a
l
y
zi
n
g
cr
ea
ted
m
o
d
els.
i
*
w
as
q
u
i
ck
l
y
ad
o
p
ted
b
y
t
h
e
r
esear
ch
co
m
m
u
n
it
y
in
f
ie
ld
s
s
u
c
h
as
r
eq
u
ir
e
m
e
n
ts
en
g
i
n
ee
r
i
n
g
a
n
d
b
u
s
i
n
ess
m
o
d
-
elin
g
.
B
en
ef
i
tin
g
f
r
o
m
i
ts
in
te
n
tio
n
al
l
y
o
p
en
n
at
u
r
e,
m
u
ltip
le
ex
ten
s
io
n
s
o
f
th
e
i
*
la
n
g
u
a
g
e
h
av
e
b
ee
n
p
r
o
p
o
s
ed
(
s
ee
[
1
9
,
2
0
]
f
o
r
u
s
ef
u
l
r
ev
ie
w
s
)
,
eit
h
er
b
y
s
li
g
h
tl
y
r
ed
ef
i
n
i
n
g
s
o
m
e
ex
is
t
in
g
co
n
s
tr
u
ct
s
,
b
y
d
etail
in
g
s
o
m
e
s
e
m
an
t
ic
is
s
u
e
s
n
o
t
co
m
p
letel
y
d
ef
in
ed
i
n
t
h
e
s
e
m
i
n
al
p
r
o
p
o
s
al,
o
r
b
y
p
r
o
p
o
s
in
g
n
e
w
co
n
s
tr
u
ct
s
f
o
r
s
p
ec
i
f
ic
d
o
m
ai
n
s
.
A
s
a
r
e
s
p
o
n
s
e
to
th
e
n
ee
d
o
f
b
ala
n
cin
g
th
e
f
r
a
m
e
w
o
r
k
’
s
o
p
en
n
atu
r
e
an
d
a
p
o
s
s
ib
le
s
o
lu
tio
n
to
th
e
af
o
r
e
m
en
t
io
n
e
d
ad
o
p
tio
n
p
r
o
b
lem
s
,
th
e
i
*
r
esear
ch
co
m
m
u
n
it
y
s
tar
ted
an
i
n
it
iati
v
e
to
id
en
t
if
y
a
w
id
el
y
a
g
r
ee
d
u
p
o
n
s
et
o
f
co
r
e
co
n
ce
p
ts
i
n
t
h
e
i
*
la
n
g
u
ag
e.
T
h
e
m
ai
n
g
o
al
is
to
k
ee
p
o
p
en
th
e
ab
ilit
y
to
t
ailo
r
th
e
f
r
a
m
e
w
o
r
k
w
h
ile
a
g
r
ee
in
g
o
n
t
h
e
f
u
n
d
a
m
e
n
tal
co
n
s
tr
u
ct
s
,
th
u
s
b
eg
a
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
18
,
No
.
2
,
A
p
r
il 2
0
2
0
:
8
3
0
-
8
3
8
832
th
e
w
o
r
k
to
p
r
o
p
o
s
e
an
u
p
d
ate
to
th
e
f
r
a
m
e
w
o
r
k
,
a
n
d
t
o
clea
r
ly
d
is
t
in
g
u
is
h
th
is
co
r
e
lan
g
u
ag
e
f
r
o
m
it
s
p
r
ed
ec
ess
o
r
s
,
it is
n
a
m
ed
iS
tar
2
.
0
.
2
.
1
.
1
.
iSta
r
2
.
0
ele
m
ent
s
A
cto
r
s
ar
e
ce
n
tr
al
to
th
e
s
o
cia
l
m
o
d
elin
g
n
at
u
r
e
o
f
th
e
la
n
g
u
ag
e
[
6
]
.
A
cto
r
s
ar
e
ac
tiv
e,
a
u
to
n
o
m
o
u
s
en
titi
e
s
th
at
ai
m
a
t
ac
h
ie
v
in
g
th
eir
g
o
als
b
y
ex
er
cis
in
g
th
e
ir
k
n
o
w
-
h
o
w
,
i
n
co
llab
o
r
atio
n
w
ith
o
t
h
er
ac
to
r
s
.
W
h
en
e
v
er
d
is
tin
g
u
is
h
i
n
g
t
h
e
ty
p
e
o
f
ac
to
r
is
n
o
t
r
elev
an
t,
eit
h
er
b
ec
au
s
e
o
f
th
e
s
ce
n
ar
io
-
at
-
h
a
n
d
o
r
th
e
m
o
d
elin
g
s
ta
g
e,
t
h
e
n
o
ti
o
n
o
f
g
en
er
ic
ac
to
r
-
w
it
h
o
u
t
s
p
ec
ializatio
n
-
ca
n
b
e
u
s
ed
i
n
t
h
e
m
o
d
el
.
A
c
to
r
s
ar
e
r
ep
r
esen
ted
g
r
ap
h
icall
y
as c
ir
c
les.
I
n
ten
t
io
n
al
ele
m
e
n
ts
ar
e
t
h
e
t
h
in
g
s
ac
to
r
s
w
an
t.
A
s
s
u
ch
,
t
h
e
y
m
o
d
el
d
i
f
f
er
en
t
k
i
n
d
s
o
f
r
eq
u
ir
e
m
e
n
t
s
an
d
ar
e
ce
n
tr
al
to
t
h
e
i
Star
2
.
0
lan
g
u
a
g
e.
An
i
n
te
n
tio
n
al
el
e
m
en
t
ap
p
ea
r
in
g
i
n
s
id
e
th
e
b
o
u
n
d
ar
y
o
f
a
n
ac
to
r
d
en
o
tes
s
o
m
et
h
i
n
g
t
h
a
t
is
d
esire
d
o
r
w
a
n
ted
b
y
t
h
at
a
cto
r
.
T
h
e
f
o
llo
w
i
n
g
ele
m
e
n
t
s
ar
e
in
clu
d
ed
in
th
e
lan
g
u
ag
e
[
6
]
,
w
i
th
e
x
a
m
p
l
es sh
o
w
n
i
n
Fi
g
u
r
e
1
:
-
Go
al:
a
s
tate
o
f
af
f
air
s
t
h
at
t
h
e
ac
to
r
w
a
n
t
s
to
ac
h
ie
v
e
an
d
th
at
h
as c
lear
-
cu
t c
r
iter
ia
o
f
ac
h
i
ev
e
m
e
n
t.
-
Qu
alit
y
:
a
n
attr
ib
u
te
f
o
r
w
h
ic
h
an
ac
to
r
d
esire
s
s
o
m
e
lev
el
o
f
ac
h
ie
v
e
m
e
n
t.
Q
u
ali
ties
ca
n
g
u
id
e
t
h
e
s
ea
r
c
h
f
o
r
w
a
y
s
o
f
ac
h
ie
v
i
n
g
g
o
als,
a
n
d
also
s
er
v
e
as c
r
iter
ia
f
o
r
ev
alu
ati
n
g
al
ter
n
ati
v
e
w
a
y
s
o
f
ac
h
iev
in
g
g
o
al
s
.
-
T
ask
:
r
ep
r
esen
ts
ac
tio
n
s
th
at
an
ac
to
r
w
an
ts
to
b
e
ex
ec
u
ted
,
u
s
u
all
y
w
it
h
t
h
e
p
u
r
p
o
s
e
o
f
ac
h
iev
in
g
s
o
m
e
g
o
al.
-
R
eso
u
r
ce
:
A
p
h
y
s
ical
o
r
in
f
o
r
m
atio
n
a
l e
n
ti
t
y
t
h
at
t
h
e
ac
to
r
r
eq
u
ir
es in
o
r
d
er
to
p
er
f
o
r
m
a
t
ask
.
Fig
u
r
e
1
.
iSt
ar
2
.
0
in
ten
tio
n
al
ele
m
e
n
ts
[
6
]
2
.
2
.
F
a
s
t
er
R
-
CNN
,
G
lea
m
,
a
nd
Sa
lt
a
nd
P
epper
No
is
e
Fas
ter
R
e
g
io
n
-
b
ased
C
o
n
v
o
l
u
t
io
n
al
Neu
r
al
Ne
t
w
o
r
k
is
a
n
u
p
g
r
ad
ed
v
er
s
io
n
o
f
R
-
C
NN
w
it
h
a
b
etter
p
er
f
o
r
m
a
n
ce
f
o
r
o
b
j
ec
t
d
et
ec
tio
n
.
Fi
g
u
r
e
2
s
h
o
w
s
t
h
e
ar
ch
itect
u
r
e
o
f
Fas
ter
R
-
C
NN,
w
it
h
s
tep
s
as f
o
llo
w
s
[
2
1
]
.
-
R
eg
io
n
P
r
o
p
o
s
al
Net
w
o
r
k
:
T
h
e
v
er
y
f
ast
tas
k
is
to
s
ea
r
ch
i
n
th
e
g
iv
e
n
i
n
p
u
t
i
m
a
g
e
t
h
e
s
p
a
ce
s
w
h
er
e
t
h
er
e
is
a
p
r
o
b
ab
ilit
y
o
f
lo
ca
tio
n
o
f
o
b
j
ec
t.
T
h
e
p
o
s
itio
n
o
f
t
h
e
o
b
j
ec
t
in
an
i
m
a
g
e
ca
n
b
e
lo
ca
ted
[
2
2
]
.
T
h
ese
r
eg
io
n
s
w
h
er
e
t
h
er
e
is
p
o
s
s
ib
ilit
y
o
f
o
b
j
ec
t
is
b
o
u
n
d
ed
b
y
a
r
eg
io
n
k
n
o
w
n
as
r
eg
io
n
o
f
in
ter
est(
R
O
I
).
-
C
las
s
i
f
icatio
n
:
T
h
e
s
tag
e
is
to
class
if
y
t
h
e
r
eg
io
n
s
o
f
in
ter
est
id
en
ti
f
ied
in
th
e
a
b
o
v
e
s
tep
s
in
to
co
r
r
esp
o
n
d
in
g
clas
s
es.T
h
e
tech
n
iq
u
e
d
ep
lo
y
ed
h
er
e
is
C
o
n
v
o
lu
tio
n
Ne
u
r
al
Net
w
o
r
k
s
(
C
N
N)
.
I
n
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
th
er
e
is
r
ig
r
o
u
s
p
r
o
ce
s
s
o
f
i
d
en
tify
i
n
g
all
s
p
ac
es
o
f
o
b
j
ec
t
lo
ca
tio
n
in
i
m
a
g
e.
Ho
w
e
v
er
i
f
n
o
r
e
g
io
n
s
ar
e
id
en
tif
ied
i
n
t
h
e
f
ir
s
t
s
ta
g
e
o
f
alg
o
r
it
h
m
th
e
n
t
h
er
e
is
n
o
n
ee
d
to
f
u
r
th
e
r
g
o
to
th
e
s
ec
o
n
d
s
tep
o
f
ap
p
r
o
ac
h
[
2
3
]
.
Fig
u
r
e
2
.
Fas
ter
R
-
C
N
N
A
r
ch
i
tectu
r
e
[
2
1
]
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
S
in
g
le
o
b
ject
d
e
tectio
n
to
s
u
p
p
o
r
t req
u
ir
eme
n
ts
…
(
N
a
th
a
n
a
el
Gilb
ert
)
833
C
o
lo
r
-
to
-
g
r
a
y
s
ca
le
is
t
h
e
tr
an
s
f
o
r
m
atio
n
o
f
R
GB
ch
a
n
n
el
to
an
g
r
a
y
s
ca
led
i
m
a
g
e.
Gr
a
y
s
ca
le
is
th
e
co
n
d
itio
n
i
n
w
h
ic
h
an
i
m
a
g
e
co
n
s
i
s
t
o
n
l
y
a
s
i
n
g
le
v
al
u
e
f
o
r
ea
ch
o
f
its
p
ix
el.
Gr
a
y
s
ca
l
ed
im
a
g
e
g
e
n
er
all
y
co
n
s
is
ts
o
f
g
r
e
y
,
b
lack
(
in
p
ix
els
w
it
h
w
ea
k
i
n
te
n
s
it
y
)
,
a
n
d
w
h
ite
(
i
n
p
ix
el
s
w
i
th
s
tr
o
n
g
i
n
ten
s
it
y
)
[
1
5
]
.
Fo
r
m
u
la
(
1
)
is
t
h
e
f
o
r
m
u
la
to
co
n
v
er
t
t
h
e
R
GB
c
h
an
n
el
i
n
a
p
ix
el
in
to
a
s
in
g
le
v
al
u
e
r
an
g
i
n
g
f
r
o
m
0
-
2
5
5
(
g
r
a
y
s
ca
le)
[
1
6
]
,
w
h
er
e
th
e
R
’
,
G’
,
a
n
d
B
’
ar
e
g
et
f
r
o
m
th
e
R
GB
ch
a
n
n
els
w
h
ich
ar
e
g
a
m
m
a
co
r
r
ec
ted
u
s
in
g
Fo
r
m
u
la
(
2
)
.
Fig
u
r
e
3
s
h
o
w
s
t
h
e
r
esu
l
t o
f
a
g
r
a
y
s
ca
li
n
g
p
r
o
ce
s
s
u
s
i
n
g
Glea
m
.
=
1
3
(
′
+
′
+
′
)
(
1
)
(
2
)
Fig
u
r
e
3
.
E
x
a
m
p
le
o
f
g
r
a
y
s
ca
l
in
g
u
s
in
g
Glea
m
Salt
an
d
P
ep
p
er
n
o
is
e
is
u
s
ed
f
o
r
r
e
p
licatin
g
i
m
a
g
es
in
t
h
e
d
ataset
f
o
r
tr
ain
i
n
g
t
h
e
m
o
d
el
b
y
ap
p
ly
i
n
g
n
o
is
e
i
n
th
e
o
r
i
g
in
a
l
i
m
a
g
e.
I
t
d
o
es
s
o
b
y
ch
a
n
g
i
n
g
p
i
x
el
v
al
u
e
in
to
t
h
e
m
in
i
m
u
m
o
r
m
a
x
i
m
u
m
v
al
u
e
ac
ce
p
ted
[
2
4
,
2
5
]
.
Fig
u
r
e
4
b
elo
w
s
h
o
w
s
t
h
e
r
esu
lt o
f
w
h
en
w
e
ap
p
l
y
t
h
e
n
o
i
s
e
in
to
a
n
i
m
a
g
e.
Fig
u
r
e
4
.
A
p
p
licatio
n
o
f
Salt a
n
d
P
ep
p
er
No
is
e
o
n
a
Han
d
-
D
r
a
w
n
T
ask
Ob
j
ec
t in
iSt
ar
2
.
0
s
2
.
3
.
Require
m
ent
s
m
o
deli
ng
t
o
o
ls
Sev
er
al
r
e
s
ea
r
ch
es
h
av
e
alr
e
ad
y
e
m
p
h
as
ized
th
e
i
m
p
o
r
tan
ce
i
*
f
r
a
m
e
w
o
r
k
[
1
8
]
f
o
r
m
o
d
eli
n
g
an
d
d
o
cu
m
en
t
in
g
r
eq
u
ir
e
m
e
n
ts
,
in
cl
u
d
i
n
g
th
e
n
e
w
l
y
-
s
ta
n
d
ar
d
ized
iSt
ar
2
.
0
[
6
,
1
9
,
2
0
]
.
On
p
r
ev
io
u
s
r
esear
ch
es,
t
h
e
p
r
o
p
o
s
al
o
f
in
t
eg
r
atin
g
s
e
v
er
al
r
eq
u
ir
e
m
en
t
s
m
o
d
eli
n
g
f
r
a
m
e
w
o
r
k
a
n
d
n
o
tatio
n
,
i
n
clu
d
i
n
g
t
h
e
ea
r
l
y
i
*
f
r
a
m
e
w
o
r
k
is
co
n
d
u
cted
an
d
s
h
o
w
ed
th
e
p
o
ten
ti
al
o
f
u
s
i
n
g
i
*
a
s
a
to
o
l
to
m
o
d
el
s
tak
e
h
o
ld
er
d
ep
en
d
en
c
y
i
n
a
n
al
y
zin
g
ea
r
l
y
-
p
h
ase
r
eq
u
ir
e
m
e
n
ts
[
2
6
,
2
7
]
.
An
o
th
er
r
e
s
ea
r
ch
r
ec
o
g
n
ized
th
e
n
ee
d
o
f
a
to
o
l
f
o
r
d
r
a
w
i
n
g
a
n
d
ed
itin
g
iSt
ar
2
.
0
d
iag
r
am
s
,
th
e
n
d
ev
elo
p
ed
th
e
p
iSt
ar
to
o
l
f
o
r
s
u
p
p
o
r
tin
g
th
e
cr
ea
tio
n
o
f
t
h
e
r
eq
u
ir
e
m
e
n
ts
m
o
d
el
[
2
8
]
.
Oth
er
r
esear
ch
es
p
r
o
p
o
s
ed
ex
ten
s
io
n
s
to
t
h
e
iS
tar
2
.
0
[
2
9
]
an
d
p
r
o
to
ty
p
e
f
o
r
g
en
er
ati
n
g
m
ea
n
i
n
g
f
u
l
la
y
o
u
t
[
3
0
]
.
Ho
w
e
v
er
,
th
e
to
p
ic
o
n
d
ig
italizatio
n
an
d
th
e
u
s
e
o
f
m
ac
h
i
n
e
lear
n
in
g
ar
ch
itect
u
r
e
f
o
r
o
b
j
ec
t
d
etec
tio
n
o
n
iSt
ar
d
iag
r
a
m
s
i
s
s
til
l
r
ar
e
to
b
e
f
o
u
n
d
.
T
h
is
r
esear
ch
ai
m
s
to
ad
d
r
ess
th
e
m
is
s
i
n
g
to
p
ic
b
y
r
ev
ie
w
in
g
it
s
i
m
p
o
r
tan
ce
an
d
k
ick
s
tar
tin
g
t
h
e
d
ev
elo
p
m
e
n
t o
f
s
u
c
h
to
o
l.
2
.
4
.
Sin
g
le
o
bje
ct
det
ec
t
io
n
a
nd
re
co
g
nitio
n f
o
r
iSta
r
2
.
0
Usi
n
g
t
h
e
ar
ch
itec
tu
r
e
p
r
o
v
id
ed
b
y
th
e
Fas
ter
R
-
C
NN
tech
n
iq
u
e,
g
r
a
y
s
ca
li
n
g
u
s
i
n
g
G
lea
m
,
an
d
u
p
s
a
m
p
l
in
g
th
e
d
atase
t
b
y
r
ep
licati
n
g
th
e
i
m
a
g
e
u
s
in
g
Salt
an
d
P
ep
p
er
,
th
e
p
r
o
g
r
a
m
is
t
h
en
d
esi
g
n
ed
.
Fig
u
r
e
5
s
h
o
w
s
t
h
e
f
lo
w
in
wh
ich
t
h
e
tr
ai
n
i
n
g
ac
ti
v
it
y
i
s
d
o
n
e
to
b
u
ilt
th
e
m
ac
h
i
n
e
lear
n
i
n
g
m
o
d
el
w
h
ich
w
ill
b
e
u
s
ed
to
d
etec
t
a
n
d
r
ec
o
g
n
ize
o
b
j
ec
ts
.
At
t
h
e
b
eg
i
n
n
in
g
,
6
0
0
i
m
a
g
e
d
ata
ar
e
co
llec
ted
as
t
h
e
d
ataset,
co
n
s
is
tin
g
o
f
th
e
d
r
a
w
i
n
g
s
o
f
5
o
b
j
ec
ts
in
th
e
i
Star
2
.
0
n
o
tatio
n
,
g
o
al,
q
u
alit
y
,
ac
to
r
,
t
ask
,
a
n
d
r
eso
u
r
ce
.
Af
ter
th
e
d
ata
s
et
is
co
llected
,
lab
ellin
g
i
s
d
o
n
e
f
o
r
ea
ch
o
b
j
ec
t,
r
esu
ltin
g
i
n
an
XM
L
f
ile
co
n
tai
n
i
n
g
al
l
th
e
i
m
ag
e
s
an
d
t
h
eir
lab
els
.
T
h
e
g
e
n
er
ated
XM
L
is
t
h
en
co
n
v
er
ted
to
a
C
S
V
f
i
lle
w
h
ich
th
en
is
u
s
ed
to
tr
ain
th
e
m
ac
h
in
e
lear
n
i
n
g
m
o
d
el
b
y
r
u
n
n
in
g
t
h
e
e
x
p
o
r
t
in
f
er
e
n
ce
g
r
ap
h
.
T
h
ese
ac
tio
n
s
ar
e
d
o
n
e
b
y
u
tili
z
in
g
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
18
,
No
.
2
,
A
p
r
il 2
0
2
0
:
8
3
0
-
8
3
8
834
th
e
T
en
s
o
r
Flo
w
Ob
j
ec
t
Dete
ctio
n
A
P
I
.
T
h
e
r
esu
lti
n
g
m
o
d
el
is
th
en
te
s
ted
an
d
m
e
asu
r
ed
to
f
i
n
d
o
u
t
th
e
p
er
f
o
r
m
a
n
ce
,
b
ased
o
n
a
co
n
f
u
s
io
n
m
atr
i
x
to
ca
lc
u
lat
e
ac
cu
r
ac
y
,
r
ec
all,
p
r
ec
is
io
n
an
d
th
e
F
-
1
s
co
r
e
o
f
t
h
e
m
o
d
el.
Se
v
er
al
co
n
f
i
g
u
r
atio
n
s
ar
e
test
ed
to
f
in
d
t
h
e
b
est
s
ce
n
ar
io
.
T
h
e
r
es
u
lt
s
ar
e
d
escr
ib
ed
in
th
e
n
e
x
t
s
ec
tio
n
o
f
t
h
i
s
p
ap
er
.
Mo
r
eo
v
er
,
b
esid
es
d
esig
n
i
n
g
th
e
tr
ai
n
i
n
g
ac
ti
v
it
y
alo
n
g
w
i
th
all
i
ts
p
r
o
ce
s
s
es,
th
e
f
lo
w
o
f
th
e
p
r
o
g
r
a
m
f
ea
t
u
r
e
w
h
ic
h
w
il
l
d
etec
t
an
d
r
ec
o
g
n
ize
s
u
b
m
itted
an
d
u
n
lab
eled
i
m
a
g
e
d
ata
is
also
d
esig
n
ed
,
as
s
h
o
w
n
i
n
Fig
u
r
e
6
.
T
h
e
f
lo
w
s
tar
t
s
b
y
g
e
tti
n
g
th
e
s
u
b
m
it
ted
i
m
a
g
e
f
ile
o
f
a
h
an
d
-
d
r
a
w
n
i
Star
2
.
0
o
b
j
ec
t,
th
en
its
p
i
x
els
ar
e
co
n
v
er
ted
an
d
g
r
a
y
s
ca
led
(
u
s
in
g
G
lea
m
)
.
Fu
r
th
er
m
o
r
e,
p
ath
i
n
itializa
t
io
n
is
d
o
n
e
s
o
th
at
th
e
d
ev
elo
p
ed
o
b
j
ec
t
d
etec
tio
n
A
P
I
k
n
o
w
s
t
h
e
e
x
ac
t
p
ath
o
f
th
e
f
ile.
NU
M_
C
L
A
S
SES
d
escr
ib
es
th
e
n
u
m
b
er
o
f
ex
i
s
ti
n
g
c
lass
e
s
i
n
w
h
ich
an
o
b
j
ec
t
w
ill
b
e
clas
s
i
f
ied
to
.
An
o
b
j
ec
t
is
co
n
s
id
er
ed
b
elo
n
g
in
g
to
a
cla
s
s
w
h
e
n
it
ac
h
ie
v
es
a
s
co
r
e
b
ig
g
er
th
a
n
0
.
9
,
if
th
er
e
ar
e
m
o
r
e
t
h
a
n
o
n
e
class
t
h
at
ac
h
ie
v
es
0
.
9
,
th
e
n
th
e
f
ir
s
t
id
en
ti
f
ied
class
is
co
n
s
id
er
ed
as th
e
co
r
r
ec
t c
lass
.
Fig
u
r
e
5
.
Flo
w
c
h
ar
t tr
ain
i
n
g
m
o
d
el
Fig
u
r
e
6
.
Flo
w
c
h
ar
t iStar
2
.
0
o
b
j
ec
t d
etec
tio
n
an
d
r
ec
o
g
n
iti
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
S
in
g
le
o
b
ject
d
e
tectio
n
to
s
u
p
p
o
r
t req
u
ir
eme
n
ts
…
(
N
a
th
a
n
a
el
Gilb
ert
)
835
3.
RE
SU
L
T
S
A
ND
AN
AL
Y
SI
S
I
n
o
r
d
er
to
m
ea
s
u
r
e
t
h
e
m
o
d
e
l’
s
p
er
f
o
r
m
an
ce
a
n
d
ev
al
u
ate
its
p
o
ten
tial
to
b
e
f
u
r
th
er
d
ev
elo
p
ed
as
a
s
u
p
p
o
r
t
to
o
l
f
o
r
r
eq
u
ir
em
e
n
t
s
m
o
d
elin
g
,
s
e
v
en
te
s
t
s
ce
n
ar
i
o
s
w
er
e
d
esig
n
ed
an
d
ex
p
er
i
m
en
ts
ar
e
co
n
d
u
cte
d
to
f
in
d
t
h
e
b
est co
n
d
itio
n
f
r
o
m
all
s
ce
n
ar
io
s
to
b
u
ild
a
h
i
g
h
-
p
er
f
o
r
m
i
n
g
m
o
d
el.
3
.
1
.
T
est
Scena
rio
s
Var
io
u
s
te
s
t
s
ce
n
ar
io
s
ar
e
p
r
ep
ar
ed
b
y
u
s
i
n
g
v
ar
io
u
s
lear
n
in
g
r
ate,
f
ea
tu
r
e
e
x
tr
ac
to
r
,
in
itia
l
cr
o
p
s
ize,
m
ax
p
o
o
l
k
er
n
el,
an
d
m
a
x
p
o
o
l
s
tr
id
e.
T
h
e
d
etails
o
f
t
h
e
s
e
v
en
test
s
ce
n
ar
io
s
e
x
p
er
i
m
e
n
t
ed
ca
n
b
e
s
ee
n
i
n
T
ab
le
1
.
Fro
m
t
h
o
s
e
s
e
v
en
s
c
en
ar
io
s
,
t
h
e
m
o
d
el’
s
ab
ilit
y
t
o
d
etec
t
an
d
r
ec
o
g
n
ize
o
b
j
ec
t
s
is
t
h
en
m
ea
s
u
r
ed
b
ased
o
n
th
e
co
n
f
u
s
io
n
m
atr
ix
r
esu
lt,
ca
lcu
lati
n
g
t
h
eir
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
,
an
d
F
-
1
Sco
r
e.
T
ab
le
1
.
T
est s
ce
n
ar
io
s
S
c
e
n
a
r
i
o
L
e
a
r
n
i
n
g
Rate
F
e
a
t
u
r
e
Ex
t
r
a
c
t
o
r
I
n
i
t
i
a
l
c
r
o
p
si
z
e
M
a
x
p
o
o
l
K
e
r
n
e
l
M
a
x
p
o
o
l
S
t
r
i
d
e
A
n
c
h
o
r
G
e
n
e
r
a
t
o
r
Ty
p
e
F
i
r
st
st
a
g
e
f
e
a
t
u
r
e
s
st
r
i
d
e
H
e
i
g
h
t
S
t
r
i
d
e
W
i
d
t
h
S
t
r
i
d
e
1
0
.
0
0
0
3
f
a
st
e
r
_
r
c
n
n
_
i
n
c
e
p
t
i
o
n
_
r
e
sn
e
t
_
v
2
8
17
1
1
8
8
2
0
.
0
0
0
2
f
a
st
e
r
_
r
c
n
n
_
i
n
c
e
p
t
i
o
n
_
v
2
16
14
2
2
16
16
3
0
.
0
0
0
2
f
a
st
e
r
_
r
c
n
n
_
i
n
c
e
p
t
i
o
n
_
r
e
sn
e
t
_
v
2
16
14
2
2
16
16
4
0
.
0
0
0
1
f
a
st
e
r
_
r
c
n
n
_
i
n
c
e
p
t
i
o
n
_
v
2
16
14
2
2
16
16
5
0
.
0
0
0
3
f
a
st
e
r
_
r
c
n
n
_
i
n
c
e
p
t
i
o
n
_
v
2
8
17
1
1
8
8
6
0
.
0
0
0
3
f
a
st
e
r
_
r
c
n
n
_
i
n
c
e
p
t
i
o
n
_
r
e
sn
e
t
_
v
2
8
14
2
2
8
8
7
0
.
0
0
0
2
f
a
st
e
r
_
r
c
n
n
_
i
n
c
e
p
t
i
o
n
_
v
2
16
17
1
1
16
16
3
.
2
.
Resul
t
Af
ter
e
x
p
er
i
m
e
n
ts
ar
e
co
n
d
u
cted
b
ased
o
n
th
e
v
ar
io
u
s
c
o
n
f
i
g
u
r
atio
n
d
escr
ib
ed
in
t
h
e
p
r
ev
io
u
s
s
ec
tio
n
,
te
s
t
r
esu
lts
a
r
e
as
d
escr
ib
ed
in
T
ab
le
2
.
T
h
e
h
i
g
h
est
p
er
f
o
r
m
in
g
s
ce
n
ar
io
is
f
o
u
n
d
o
n
t
h
e
f
o
u
r
th
s
ce
n
ar
io
,
u
s
i
n
g
lear
n
i
n
g
r
ate
0
.
0
0
0
1
,
f
ea
tu
r
e
ex
tr
ac
to
r
t
y
p
e
f
aster
_
r
cn
n
_
i
n
ce
p
tio
n
_
v
2
w
it
h
1
6
f
ir
s
t
s
ta
g
e
f
ea
t
u
r
es
s
tr
id
e,
1
4
in
itial
cr
o
p
s
ize,
2
m
a
x
p
o
o
l
k
er
n
el
an
d
s
t
r
id
e,
an
d
1
6
h
eig
h
t
an
d
w
id
th
s
tr
id
e,
r
esu
ltin
g
i
n
an
av
er
a
g
e
o
f
9
4
% a
cc
u
r
ac
y
,
9
5
% p
r
ec
is
io
n
,
1
0
0
% r
ec
all,
an
d
9
7
.
2
% F1
-
Sco
r
e
f
o
r
ea
ch
cla
s
s
.
T
ab
le
2
.
T
est r
esu
lts
S
c
e
n
a
r
i
o
A
v
e
r
a
g
e
A
c
c
u
r
a
c
y
P
r
e
c
i
si
o
n
R
e
c
a
l
l
F1
-
sco
r
e
1
5
7
%
6
3
%
9
7
%
7
5
,
0
5
%
2
9
4
%
9
4
%
1
0
0
%
9
6
,
8
7
%
3
4
2
%
5
2
%
9
5
%
6
5
,
5
1
%
4
9
5
%
9
5
%
1
0
0
%
9
7
,
2
0
%
5
9
4
%
9
5
%
9
7
%
9
5
,
9
1
%
6
3
9
%
4
8
%
9
6
%
6
0
,
9
2
%
7
8
8
%
9
1
%
9
9
%
9
4
,
3
6
%
Fro
m
th
e
r
es
u
lts
,
it
ca
n
b
e
s
ee
n
th
at
t
h
e
r
o
le
o
f
f
ea
tu
r
e
ex
tr
ac
to
r
,
esp
ec
ially
i
f
we
ex
a
m
i
n
e
Scen
ar
io
1
,
2
,
an
d
3
,
w
h
er
e
f
ea
t
u
r
e
ex
tr
ac
to
r
o
f
t
y
p
e
f
as
t
er
_
r
cn
n
_
ex
tr
ac
to
r
_
v
2
p
er
f
o
r
m
s
m
u
c
h
b
etter
th
a
n
th
e
o
th
er
.
Fu
r
t
h
er
m
o
r
e,
in
itia
l
cr
o
p
s
ize
also
p
r
o
v
es
to
b
e
q
u
ite
i
m
p
ac
t
f
u
l
lo
o
k
in
g
at
Scen
ar
io
2
an
d
7
.
T
h
e
lear
n
in
g
r
ate
ca
n
al
s
o
b
e
s
ee
n
to
b
e
a
d
eter
m
i
n
i
n
g
f
ac
to
r
ev
en
t
h
o
u
g
h
it
m
ig
h
t
n
o
t
r
esu
lt
i
n
a
b
i
g
g
ap
,
b
et
w
ee
n
Scen
ar
io
4
an
d
2
,
f
o
r
ex
a
m
p
le.
I
n
ad
d
itio
n
to
th
e
m
ac
h
in
e
lea
r
n
in
g
m
o
d
el
as
d
escr
ib
ed
ab
o
v
e,
o
u
r
r
esear
ch
also
d
ev
elo
p
ed
a
s
i
m
p
le
w
eb
-
b
ased
to
o
l
as
t
h
e
i
n
ter
f
ac
e
f
o
r
u
s
er
s
to
d
e
m
o
n
s
tr
ate
t
h
e
m
o
d
el
’
s
ab
ilit
y
to
d
etec
t
a
n
d
r
ec
o
g
n
ize
u
p
lo
ad
ed
h
an
d
-
d
r
a
w
n
iS
tar
2
.
0
n
o
tio
n
s
.
Fig
u
r
e
7
s
h
o
w
s
t
h
e
s
a
m
p
le
i
m
ag
e
th
at
w
i
ll b
e
d
etec
ted
an
d
r
ec
o
g
n
ized
.
T
h
e
n
o
tio
n
d
ep
icted
i
n
Fig
u
r
e
7
is
t
h
e
T
ask
in
t
h
e
i
Star
2
.
0
s
tr
ateg
ic
d
ep
en
d
en
c
y
m
o
d
el.
Fig
u
r
e
8
d
is
p
la
y
s
t
h
e
h
o
m
e
p
ag
e
in
w
h
ic
h
t
h
e
C
h
o
o
s
e
Fil
e
f
ea
tu
r
e
ca
n
b
e
click
ed
to
u
p
lo
a
d
th
e
s
a
m
p
le
i
m
a
g
e,
th
en
w
h
e
n
t
h
e
C
h
ec
k
R
es
u
lt
b
u
tto
n
is
cl
ick
ed
,
t
h
e
s
o
f
t
w
ar
e
w
il
l
p
r
e
-
p
r
o
ce
s
s
th
e
i
m
a
g
e,
d
etec
t
th
e
o
b
j
ec
t,
an
d
d
eter
m
i
n
e
i
f
it
i
s
T
ask
,
R
eso
u
r
ce
,
Q
u
ali
t
y
,
G
o
al,
o
r
A
cto
r
s
,
a
lo
n
g
w
it
h
its
m
a
tch
r
ate.
Fig
u
r
e
9
s
h
o
ws
th
e
r
esu
lt,
w
h
er
e
th
e
s
o
f
t
w
a
r
e
is
s
ee
n
to
b
e
ab
le
to
g
u
ess
co
r
r
ec
tl
y
w
h
at
n
o
tio
n
t
h
e
u
p
lo
ad
ed
im
a
g
e
d
ep
icted
,
T
ask
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
18
,
No
.
2
,
A
p
r
il 2
0
2
0
:
8
3
0
-
8
3
8
836
Fig
u
r
e
7
.
Sa
m
p
le
I
m
a
g
e
f
o
r
O
b
j
ec
t D
etec
tio
n
Fig
u
r
e
8
.
User
i
n
ter
f
ac
e
o
f
th
e
w
eb
-
b
a
s
ed
test
i
n
g
ap
p
licatio
n
Fig
u
r
e
9
.
User
in
ter
f
ac
e
w
h
e
n
th
e
ap
p
licatio
n
d
is
p
la
y
s
t
h
e
o
b
j
ec
t d
etec
tio
n
r
esu
lt
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
S
in
g
le
o
b
ject
d
e
tectio
n
to
s
u
p
p
o
r
t req
u
ir
eme
n
ts
…
(
N
a
th
a
n
a
el
Gilb
ert
)
837
4.
CO
NCLU
SI
O
N
T
h
is
r
esear
ch
u
t
ilized
Fa
s
ter
R
-
C
NN
u
s
i
n
g
a
d
ataset
co
m
p
r
is
in
g
o
f
h
a
n
d
-
d
r
a
w
n
i
Star
2
.
0
o
b
j
ec
ts
s
u
c
h
as
Ge
n
er
ic
A
cto
r
,
T
ask
,
R
eso
u
r
ce
,
Q
u
alit
y
,
an
d
Go
a
l.
I
m
ag
e
s
ar
e
f
ir
s
t
p
r
e
-
p
r
o
ce
s
s
ed
an
d
r
ep
licated
u
s
i
n
g
Glea
m
f
o
r
it
s
co
lo
r
-
to
-
g
r
a
y
s
ca
le
tec
h
n
iq
u
e
a
n
d
Sal
t
an
d
P
ep
p
er
n
o
is
e
to
g
i
v
e
n
o
is
e
to
t
h
e
o
r
ig
i
n
al
d
ataset
an
d
d
u
p
licate
th
e
n
u
m
b
er
o
f
i
m
a
g
es
i
n
th
e
d
ataset
.
T
h
e
r
esu
lti
n
g
p
r
o
g
r
a
m
i
s
b
est
p
er
f
o
r
m
i
n
g
u
s
i
n
g
0
.
0
0
0
1
lear
n
in
g
r
ate,
f
ea
t
u
r
e
ex
tr
ac
to
r
t
y
p
e
f
aster
_
r
cn
n
_
in
ce
p
tio
n
_
v
2
w
it
h
1
6
f
ir
s
t
s
t
ag
e
f
ea
t
u
r
es
s
tr
id
e,
1
4
in
itial
cr
o
p
s
ize,
2
m
a
x
p
o
o
l
k
er
n
el
a
n
d
s
tr
id
e,
an
d
1
6
h
eig
h
t
a
n
d
w
id
t
h
s
tr
id
e,
r
es
u
l
tin
g
i
n
an
a
v
er
ag
e
o
f
9
4
%
ac
cu
r
ac
y
,
9
5
%
p
r
ec
is
io
n
,
1
0
0
%
r
ec
all,
an
d
9
7
.
2
%
F1
-
Sco
r
e
f
o
r
ea
ch
c
lass
.
T
h
e
co
n
d
u
cted
r
esear
ch
d
is
p
la
y
s
t
h
e
p
o
ten
tial
o
f
Fa
s
ter
-
R
C
N
N,
Glea
m
,
an
d
Sal
t
an
d
P
ep
p
er
t
o
b
u
ild
a
m
o
d
el
f
o
r
d
etec
tin
g
an
d
r
ec
o
g
n
izi
n
g
o
b
j
ec
ts
d
r
aw
n
u
s
i
n
g
th
e
i
Star
2
.
0
to
en
ab
le
th
e
d
ig
ital
izatio
n
r
eq
u
ir
e
m
e
n
ts
d
iag
r
a
m
to
s
u
p
p
o
r
t
th
e
r
eq
u
ir
e
m
e
n
ts
m
o
d
eli
n
g
ac
t
iv
it
y
i
n
s
o
f
t
w
ar
e
d
e
v
elo
p
m
e
n
t.
Fu
t
u
r
e
w
o
r
k
s
i
n
cl
u
d
e
i
m
p
r
o
v
i
n
g
th
e
d
ataset
a
n
d
m
ac
h
in
e
lear
n
i
n
g
m
o
d
el
to
b
e
ab
le
to
d
ig
italize
a
w
h
o
l
e
iSt
a
r
2
.
0
d
iag
r
a
m
,
e
n
ab
li
n
g
t
h
e
m
u
lti
-
o
b
j
ec
t
d
etec
tio
n
,
an
d
d
ev
elo
p
in
g
to
o
ls
f
o
r
ed
iti
n
g
a
n
d
cr
ea
ti
n
g
t
h
e
w
h
o
le
d
ia
g
r
a
m
u
s
i
n
g
t
h
e
iSt
ar
2
.
0
an
d
o
t
h
er
n
o
tatio
n
f
o
r
r
eq
u
ir
e
m
e
n
ts
m
o
d
eli
n
g
.
Op
tical
c
h
ar
ac
ter
r
ec
o
g
n
itio
n
tech
n
iq
u
es
ca
n
also
b
e
in
te
g
r
ated
to
b
e
ab
l
e
to
r
ea
d
tex
ts
in
s
id
e
t
h
e
d
r
a
w
n
o
b
j
ec
ts
.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
is
r
esear
ch
w
as
s
u
p
p
o
r
ted
b
y
t
h
e
Mo
b
ile
De
v
elo
p
m
e
n
t
L
ab
o
r
ato
r
y
i
n
U
n
i
v
er
s
ita
s
Mu
lti
m
ed
ia
Nu
s
a
n
tar
a.
W
e
also
th
an
k
o
u
r
co
lleag
u
es
f
r
o
m
t
h
e
Fac
u
lt
y
o
f
E
n
g
in
ee
r
i
n
g
an
d
I
n
f
o
r
m
atics
w
h
o
p
r
o
v
id
ed
in
s
i
g
h
t
a
n
d
e
x
p
er
tis
e
t
h
at
g
r
ea
tl
y
a
s
s
i
s
ted
t
h
e
r
esea
r
ch
,
alt
h
o
u
g
h
th
e
y
m
a
y
n
o
t
ag
r
ee
w
it
h
a
ll
o
f
th
e
i
n
ter
p
r
etatio
n
s
/co
n
cl
u
s
i
o
n
s
o
f
t
h
is
p
ap
er
.
RE
F
E
R
E
NC
E
S
[1
]
Nu
se
ib
e
h
,
B.
,
Eas
terb
r
o
o
k
,
S
.
,
“
Re
q
u
irem
e
n
ts
En
g
in
e
e
rin
g
:
A
R
o
a
d
m
a
p
,
”
Pro
c
e
e
d
in
g
s
o
f
th
e
Co
n
fer
e
n
c
e
o
n
th
e
Fu
tu
re
o
f
S
o
ft
w
a
re
En
g
in
e
e
rin
g
.
L
ime
ric
k
,
p
p
.
3
5
-
4
6
,
2
0
0
0
.
[2
]
Ch
a
k
ra
b
o
rty
,
A
.
,
Ba
o
w
a
l
y
,
M
.
K.,
A
re
f
in
,
A
.
,
Ba
h
a
r,
A
.
N.
,
“
T
h
e
Ro
le
o
f
Re
q
u
irem
e
n
ts
En
g
in
e
e
ri
n
g
in
S
o
f
tw
a
r
e
De
v
e
lo
p
m
e
n
t
L
i
f
e
C
y
c
le
,”
J
o
u
rn
a
l
o
f
Eme
rg
in
g
T
re
n
d
s
in
C
o
m
p
u
ti
n
g
a
n
d
in
f
o
rm
a
t
io
n
S
c
ien
c
e
s
,
v
o
l
3
,
n
o
.
5
,
p
p
.
7
2
3
-
7
2
9
.
2
0
1
2
[3
]
P
re
ss
m
a
n
,
R.
S
.
,
M
a
x
i
m
,
B.
,
“
S
o
f
tw
a
r
e
En
g
in
e
e
rin
g
:
A
P
ra
c
ti
ti
o
n
e
r'
s
A
p
p
ro
a
c
h
,
”
8
th
Ed
i
ti
o
n
.
Ne
w
Yo
rk
:
M
c
G
ra
w
-
Hill
Ed
u
c
a
ti
o
n
.
2
0
1
4
.
[4
]
V
a
n
L
a
m
s
we
e
rd
e
,
A
.
,
“
Re
q
u
ire
m
e
n
ts
En
g
in
e
e
rin
g
:
F
r
o
m
S
y
ste
m
G
o
a
ls
to
UML
M
o
d
e
l
s
to
S
o
f
tw
a
re
S
p
e
c
if
ica
ti
o
n
,
”
Jo
h
n
W
il
e
y
&
S
o
n
s L
td
:
Ch
ich
e
ste
r.
2
0
0
9
.
[5
]
X
.
F
ra
n
c
h
,
A
.
M
a
té,
J.
C.
T
ru
ji
ll
o
a
n
d
C.
Ca
re
s,
"
On
th
e
jo
in
t
u
se
o
f
i
∗
w
it
h
o
th
e
r
m
o
d
e
ll
in
g
f
ra
m
e
w
o
rk
s:
A
v
isio
n
p
a
p
e
r,
"
2
0
1
1
IEE
E
1
9
th
I
n
ter
n
a
ti
o
n
a
l
Req
u
ire
me
n
ts E
n
g
i
n
e
e
rin
g
C
o
n
f
e
re
n
c
e
,
p
p
.
1
3
3
-
1
4
2
,
T
re
n
to
,
2
0
1
1
.
[6
]
Da
lp
iaz
,
F
.
,
F
ra
n
c
h
,
X.,
Ho
rk
o
f
f
,
J.
,
“
iS
tar 2
.
0
L
a
n
g
u
a
g
e
G
u
id
e
,
”
,
a
rX
iv
p
re
p
rin
t
a
rX
iv
:1
6
0
5
.
0
7
7
6
7
v
3
,
2
0
1
6
.
[7
]
M
e
lt
z
e
r,
L
.
,
“
Ex
e
c
u
ti
v
e
F
u
n
c
ti
o
n
in
E
d
u
c
a
ti
o
n
:
F
ro
m
T
h
e
o
r
y
to
P
ra
c
ti
c
e
,
”
Ne
w
Yo
rk
:
T
h
e
G
u
il
f
o
rd
P
re
ss
,
2
0
0
7
.
[8
]
Am
it
Y.,
F
e
lze
n
sz
w
a
lb
P
.
,
“
Ob
jec
t
De
tec
ti
o
n
”
.
I
n
:
Ik
e
u
c
h
i
K.
(e
d
s)
Co
m
p
u
ter V
isi
o
n
.
S
p
rin
g
e
r,
Bo
st
o
n
,
M
A
,
2
0
1
4
.
[9
]
Ra
h
m
a
t,
R.
F
.
,
De
n
n
is,
S
it
o
m
p
u
l
,
O.S
.
,
P
u
r
n
a
m
a
w
a
ti
,
S
.
,
Bu
d
iarto
,
R.
“
A
d
v
e
rti
se
m
e
n
t
b
il
lb
o
a
rd
d
e
tec
ti
o
n
a
n
d
g
e
o
tag
g
in
g
s
y
ste
m
w
it
h
in
d
u
c
ti
v
e
tran
sf
e
r
le
a
rn
in
g
”
T
EL
KOM
NIKA
T
e
lec
o
mm
u
n
Co
m
p
u
t
El
Co
n
tro
l
,
v
o
l.
1
7
,
n
o
.
5
,
p
p
.
2
6
5
9
-
2
6
6
6
,
2
0
1
9
.
[1
0
]
S
u
d
iatm
ik
a
,
I.
B.
K,
Ra
h
m
a
n
,
F
.
,
T
risn
o
,
S
u
y
o
to
,
“
I
m
a
g
e
f
o
rg
e
r
y
d
e
tec
ti
o
n
u
sin
g
e
rro
r
l
e
v
e
l
a
n
a
l
y
sis
a
n
d
d
e
e
p
lea
rn
in
g
,
”
T
EL
KOM
NIKA
,
v
o
l
.
1
7
,
n
o
.
2
,
p
p
.
6
5
3
-
6
5
9
,
2
0
1
9
.
[1
1
]
S
u
g
iarti,
Yu
h
a
n
d
r
i,
Na
’a
m
,
J.,
In
d
ra
,
D.,
S
a
n
t
o
n
y
,
J.
,
“
A
n
a
rti
f
i
c
ial
n
e
u
ra
l
n
e
tw
o
rk
a
p
p
ro
a
c
h
f
o
r
d
e
tec
ti
n
g
sk
in
c
a
n
c
e
r,
”
T
EL
KOM
NIKA,
v
o
l.
1
7
,
n
o
.
2
,
p
p
.
7
8
8
-
7
9
3
,
2
0
1
9
.
[1
2
]
Jia
n
g
,
H.,
Lea
rn
e
d
-
M
il
ler,
E.
,
“
F
a
c
e
D
e
tec
ti
o
n
w
it
h
th
e
F
a
ste
r
R
-
CNN
,
”
1
2
th
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Au
to
m
a
ti
c
F
a
c
e
&
Ge
stu
re
Rec
o
g
n
it
io
n
(
FG
2
0
1
7
)
,
p
p
.
6
5
0
-
6
5
7
,
W
a
sh
in
g
to
n
,
2
0
1
7
.
[1
3
]
Ka
fe
d
z
isk
i,
V
.
,
P
e
c
o
v
,
S
.
,
T
a
n
e
v
sk
i,
D.
,
“
De
tec
ti
o
n
a
n
d
Clas
sif
ic
a
ti
o
n
o
f
L
a
n
d
M
in
e
s
f
ro
m
G
ro
u
n
d
P
e
n
e
tratin
g
Ra
d
a
r
Da
ta Us
in
g
F
a
ste
r
R
-
CNN
,
”
2
6
t
h
T
e
lec
o
mm
u
n
ica
ti
o
n
s
Fo
ru
m (
T
EL
FOR)
,
Be
lg
ra
d
e
.
2
0
1
8
.
[1
4
]
P
a
l,
K.K.
,
S
u
d
e
e
p
,
S
.
,
“
P
re
p
r
o
c
e
ss
in
g
f
o
r
i
m
a
g
e
c
las
si
f
ica
ti
o
n
b
y
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s,”
IEE
E
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
Rec
e
n
t
T
re
n
d
s
in
El
e
c
tro
n
ics
,
I
n
f
o
rm
a
ti
o
n
&
Co
mm
u
n
ica
ti
o
n
T
e
c
h
n
o
l
o
g
y
(
RT
EICT
)
,
p
p
.
1
7
7
8
-
1
7
8
1
,
Ba
n
g
a
lo
re
.
2
0
1
6
.
[1
5
]
F
a
tt
a
,
H.A
.
“
Ko
n
v
e
rsi
F
o
rm
a
t
C
it
ra
Rg
b
Ke
F
o
rm
a
t
G
ra
y
sc
a
le
M
e
n
g
g
u
n
a
k
a
n
V
isu
a
l
Ba
sic
,
”
S
e
min
a
r
Na
si
o
n
a
l
T
e
k
n
o
lo
g
,
Y
o
g
y
a
k
a
rt
a
2
0
0
7
.
[1
6
]
Ka
n
a
n
C,
Co
tt
re
ll
G
W
.
,
“
Co
lo
r
-
to
-
G
ra
y
sc
a
le:
Do
e
s th
e
M
e
th
o
d
M
a
tt
e
r
in
Im
a
g
e
Re
c
o
g
n
it
io
n
?
,”
P
L
o
S
ON
E,
2
0
1
2
.
[1
7
]
Na
z
a
ré
T
.
S
.
,
d
a
Co
sta
G
.
B.
P
.
,
C
o
n
tato
W
.
A
.
,
P
o
n
ti
M
.
,
“
De
e
p
Co
n
v
o
lu
ti
o
n
a
l
N
e
u
ra
l
Ne
tw
o
rk
s
a
n
d
No
isy
I
m
a
g
e
s,”
In
:
M
e
n
d
o
z
a
M
.
,
V
e
las
tí
n
S
.
(e
d
s)
,
“
P
ro
g
re
ss
in
P
a
tt
e
rn
Re
c
o
g
n
it
io
n
,
Im
a
g
e
A
n
a
l
y
sis,
”
Co
m
p
u
ter
V
isio
n
,
a
n
d
A
p
p
li
c
a
ti
o
n
s.
CIA
RP
2
0
1
7
.
L
e
c
tu
re
No
tes
in
C
o
m
p
u
ter S
c
ien
c
e
,
v
o
l
1
0
6
5
7
.
S
p
ri
n
g
e
r,
Ch
a
m
.
2
0
1
8
.
[1
8
]
Yu
,
E.
S
.
K.
,
“
M
o
d
e
ll
in
g
stra
teg
ic rela
ti
o
n
sh
i
p
s f
o
r
p
r
o
c
e
ss
re
e
n
g
in
e
e
rin
g
,
”
P
h
D
t
h
e
sis
,
Un
iv
e
rsity
o
f
T
o
ro
n
to
.
1
9
9
6
.
[1
9
]
J.
Ho
rk
o
ff
e
t
a
l.
,
"
T
a
k
in
g
g
o
a
l
m
o
d
e
ls
d
o
w
n
stre
a
m
:
A
s
y
ste
m
a
ti
c
r
o
a
d
m
a
p
,
"
IEE
E
Ei
g
h
th
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Res
e
a
rc
h
C
h
a
ll
e
n
g
e
s i
n
I
n
fo
r
ma
ti
o
n
S
c
ien
c
e
(
RCIS
)
,
p
p
.
1
-
1
2
,
M
a
rra
k
e
c
h
,
2
0
1
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6930
T
E
L
KOM
NI
K
A
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
18
,
No
.
2
,
A
p
r
il 2
0
2
0
:
8
3
0
-
8
3
8
838
[2
0
]
Ho
rk
o
ff
,
J.,
Yu
,
E.
,
“
Co
m
p
a
riso
n
a
n
d
e
v
a
lu
a
ti
o
n
o
f
g
o
a
l
-
o
rien
te
d
s
a
ti
sf
a
c
ti
o
n
a
n
a
l
y
sis
tec
h
n
iq
u
e
s,”
Req
u
ire
me
n
ts
En
g
i
n
e
e
rin
g
,
v
o
l.
1
8
,
n
o
.
3
,
p
p
.
1
9
9
-
2
2
2
,
2
0
1
3
.
[2
1
]
A
b
b
a
s,
S
.
M
.
,
S
in
g
h
,
S
.
N.
,
“
R
e
g
io
n
-
b
a
se
d
Ob
jec
t
De
tec
ti
o
n
a
n
d
Cl
a
ss
if
ica
ti
o
n
u
sin
g
F
a
ste
r
R
-
CNN
,
”
4
th
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
t
a
ti
o
n
a
l
In
telli
g
e
n
c
e
&
Co
mm
u
n
ica
ti
o
n
T
e
c
h
n
o
lo
g
y
(
CICT
)
,
G
h
a
z
iab
a
d
,
2
0
1
8
.
[2
2
]
Re
n
,
S
.
,
He
,
K.,
G
irsh
ick
,
R.
,
Z
h
a
n
g
,
X
.
,
a
n
d
S
u
n
,
J.,
“
Ob
jec
t
d
e
t
e
c
ti
o
n
n
e
tw
o
rk
s
o
n
c
o
n
v
o
lu
t
io
n
a
l
f
e
a
tu
re
m
a
p
s,”
Co
ro
n
e
ll
Un
ive
rs
ty,
a
rXiv:1
5
0
4
.
0
6
0
6
6
,
2
0
1
6
.
[2
3
]
S
z
e
g
e
d
y
,
C.
,
A
.
T
o
sh
e
v
,
a
n
d
D
.
Erh
a
n
,
“
De
e
p
n
e
u
ra
l
n
e
tw
o
rk
s
f
o
r
o
b
jec
t
d
e
tec
ti
o
n
,
”
i
n
Ne
u
r
a
l
In
f
o
rm
a
ti
o
n
Pro
c
e
ss
in
g
S
y
ste
ms
(
NIPS
)
,
2
0
1
3
.
[2
4
]
Esa
k
k
iraj
a
n
,
S
.
,
V
e
e
ra
k
u
m
a
r,
T
.
,
S
u
b
ra
m
a
n
y
a
m
,
A
.
N.,
P
re
m
c
h
a
n
d
,
C.
H.,
“
Re
m
o
v
a
l
o
f
Hi
g
h
-
De
n
sity
S
a
lt
a
n
d
P
e
p
p
e
r
No
ise
T
h
ro
u
g
h
M
o
d
if
ied
De
c
isio
n
Ba
se
d
Un
sy
m
m
e
tri
c
T
r
imm
e
d
M
e
d
ian
F
il
ter,”
IE
EE
S
ig
n
a
l
Pro
c
e
ss
in
g
L
e
tt
e
rs
,
v
o
l.
1
8
,
n
o
.
5
,
p
p
.
2
8
7
-
2
9
0
,
2
0
1
1
.
[2
5
]
Ch
a
n
,
R.
H.,
H
o
,
C.
,
Nik
o
lo
v
a
,
M
.
,
“
S
a
lt
-
a
n
d
-
P
e
p
p
e
r
No
ise
Re
m
o
v
a
l
b
y
M
e
d
ian
-
Ty
p
e
No
ise
De
tec
t
o
rs
a
n
d
D
e
tail
-
P
re
se
rv
in
g
Re
g
u
lariz
a
ti
o
n
,
”
IE
EE
T
ra
n
s
a
c
ti
o
n
s o
n
Ima
g
e
Pro
c
e
ss
in
g
,
v
o
l.
1
4
,
n
o
.
1
0
,
p
p
.
1
4
7
9
–
1
4
8
5
,
2
0
0
5
.
[2
6
]
Ru
sli,
A
.
,
S
h
ig
o
,
O.
,
“
A
n
In
teg
ra
ted
T
o
o
l
t
o
S
u
p
p
o
rt
Early
-
P
h
a
se
Re
q
u
irem
e
n
ts
A
n
a
l
y
sis
,”
4
t
h
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Ne
w
M
e
d
ia S
tu
d
ies
(CON
M
EDIA
2
0
1
7
)
,
Yo
g
y
a
k
a
r
ta
,
2
0
1
7
.
[2
7
]
Ru
sli,
A
.
S
h
ig
o
,
O.
,
“
In
teg
ra
ted
F
ra
m
e
w
o
rk
f
o
r
S
o
f
t
w
a
re
R
e
q
u
irem
e
n
ts
A
n
a
l
y
sis
a
n
d
Its
S
u
p
p
o
rt
T
o
o
l
,”
In
:
Re
q
u
irem
e
n
ts
En
g
in
e
e
rin
g
T
o
wa
rd
S
u
sta
in
a
b
le
W
o
rld
:
T
h
ird
A
sia
-
P
a
c
if
ic
S
y
m
p
o
siu
m
,
A
P
RES
2
0
1
6
,
Na
g
o
y
a
,
2016.
[2
8
]
P
im
e
n
tel,
J.,
Ca
stro
,
J.
,
“
p
iS
tar
T
o
o
l
–
A
P
lu
g
g
a
b
le
On
li
n
e
T
o
o
l
f
o
r
G
o
a
l
M
o
d
e
li
n
g
,”
IEE
E
2
6
t
h
In
tern
a
ti
o
n
a
l
Re
q
u
irem
e
n
ts E
n
g
in
e
e
rin
g
Co
n
f
e
re
n
c
e
(RE)
,
Ba
n
f
f
,
2
0
1
8
.
[2
9
]
G
o
n
c
a
lv
e
s,
E.
,
A
ra
u
jo
,
J.,
Ca
stro
,
J.
T
o
w
a
rd
s E
x
ten
sio
n
M
e
c
h
a
n
ism
s in
iS
tar 2
.
0
.
i
S
T
A
R@C
A
iS
E.
Talli
n
n
.
2
0
1
8
[3
0
]
W
a
n
g
Y.,
L
i
T
.
,
Zh
a
n
g
H.,
S
u
n
J.
,
Ni
Y.,
G
e
n
g
C.
A
P
ro
t
o
ty
p
e
f
o
r
G
e
n
e
r
a
ti
n
g
M
e
a
n
in
g
f
u
l
L
a
y
o
u
t
o
f
iS
tar
M
o
d
e
ls.
In
:
W
o
o
C.
,
L
u
J.,
L
i
Z.
,
L
in
g
T
.
,
L
i
G
.
,
L
e
e
M
.
(e
d
s)
A
d
v
a
n
c
e
s
in
Co
n
c
e
p
t
u
a
l
M
o
d
e
li
n
g
.
ER
2
0
1
8
.
L
e
c
tu
re
No
tes
in
Co
m
p
u
ter S
c
ie
n
c
e
,
v
o
l.
1
1
1
5
8
.
S
p
ri
n
g
e
r,
Ch
a
m
.
2
0
1
8
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Na
th
a
n
a
e
l
G
il
b
e
r
t
g
ra
d
u
a
ted
f
r
o
m
th
e
D
e
p
a
rt
m
e
n
t
o
f
In
f
o
r
m
a
ti
c
s
in
Un
iv
e
rsitas
M
u
lt
ime
d
ia
Nu
sa
n
tara
,
In
d
o
n
e
sia
,
in
No
v
e
m
b
e
r
2
0
1
9
.
He
h
a
s
a
b
a
c
k
g
ro
u
n
d
i
n
re
se
a
rc
h
in
a
p
p
li
e
d
m
a
c
h
in
e
lea
rn
in
g
f
o
r
i
m
a
g
e
p
ro
c
e
ss
in
g
a
n
d
h
o
l
d
s
k
e
e
n
in
tere
sts
in
th
e
a
re
a
o
f
A
n
d
ro
id
-
b
a
se
d
m
o
b
il
e
a
p
p
li
c
a
ti
o
n
d
e
v
e
lo
p
m
e
n
t.
And
r
e
R
u
sli
re
c
e
iv
e
d
h
is
M
a
ste
r’s
o
f
S
c
ien
c
e
d
e
g
r
e
e
in
In
f
o
r
m
a
ti
o
n
E
n
v
iro
n
m
e
n
t
f
ro
m
T
o
k
y
o
De
n
k
i
Un
iv
e
rsit
y
,
Ja
p
a
n
,
in
2
0
1
7
,
f
o
c
u
sin
g
in
th
e
S
o
f
twa
re
R
e
q
u
irem
e
n
ts
En
g
in
e
e
rin
g
f
ield
.
He
is cu
rre
n
tl
y
a
lec
tu
re
r
a
n
d
re
se
a
rc
h
e
r
in
Un
iv
e
rsitas
M
u
lt
im
e
d
ia
Nu
sa
n
tara
a
n
d
a
lso
se
rv
in
g
a
s
th
e
h
e
a
d
c
o
o
r
d
i
n
a
to
r
o
f
th
e
M
o
b
il
e
De
v
e
lo
p
m
e
n
t
L
a
b
o
ra
to
ry
.
His
re
se
a
rc
h
in
tere
sts
in
c
l
u
d
e
re
q
u
irem
e
n
ts
e
n
g
in
e
e
rin
g
in
so
f
t
w
a
r
e
a
p
p
li
c
a
ti
o
n
d
e
v
e
lo
p
m
e
n
t,
n
a
tu
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
,
a
n
d
h
u
m
a
n
c
o
m
p
u
ter i
n
tera
c
ti
o
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.