I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
10
,
No
.
5
,
Octo
b
er
2
0
2
0
,
p
p
.
4
7
1
2
~
4
7
2
0
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
1
0
i
5
.
pp
4
7
1
2
-
4
7
2
0
4712
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m/in
d
ex
.
p
h
p
/I
JE
C
E
Alg
o
rith
m
of d
ete
ction, clas
sifi
ca
tion a
nd gripp
ing
o
f
o
ccluded ob
jec
ts by CN
N t
echni
ques a
nd
H
a
a
r cl
a
ss
ifiers
P
a
ula
Useche
1
,
Ro
bin
s
o
n J
i
m
e
nez
-
M
o
re
no
2
,
J
a
v
ier
M
a
r
t
ínez
B
a
qu
er
o
3
1,
2
De
p
a
rtm
e
n
t
o
f
M
e
c
h
a
tro
n
ic E
n
g
in
e
e
rin
g
,
M
il
it
a
r
N
u
e
v
a
G
r
a
n
a
d
a
M
il
it
a
ry
Un
iv
e
rsit
y
,
Co
lo
m
b
ia
3
De
p
a
rtme
n
t
o
f
El
e
c
tro
n
ic E
n
g
in
e
e
rin
g
,
De
lo
s L
lan
o
s Un
iv
e
rsity
,
Co
lo
m
b
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
4
,
2
0
1
9
R
ev
i
s
ed
Ma
r
4
,
2
0
2
0
A
cc
ep
ted
Ma
r
21
,
2
0
2
0
T
h
e
f
o
l
l
o
w
i
n
g
p
a
p
e
r
p
r
e
s
e
n
t
s
t
h
e
d
e
v
e
l
o
p
m
e
n
t
o
f
a
n
a
lg
o
r
i
t
h
m
,
i
n
c
h
a
rg
e
o
f
d
e
t
e
c
t
i
n
g
,
c
la
s
s
if
y
i
n
g
a
n
d
g
ra
b
b
i
n
g
o
c
c
l
u
d
e
d
o
b
j
e
c
t
s
,
u
s
i
n
g
a
r
t
if
ic
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
te
c
h
n
i
q
u
e
s
,
m
a
c
h
i
n
e
v
i
s
i
o
n
f
o
r
t
h
e
r
e
c
o
g
n
i
t
i
o
n
o
f
t
h
e
e
n
v
i
r
o
n
m
e
n
t
,
a
n
a
n
t
h
r
o
p
o
m
o
r
p
h
i
c
m
a
n
i
p
u
l
a
t
o
r
f
o
r
t
h
e
m
a
n
i
p
u
l
a
t
i
o
n
o
f
t
h
e
e
l
e
m
e
n
t
s
.
5
t
y
p
e
s
o
f
t
o
o
l
s
w
e
re
u
s
e
d
f
o
r
t
h
e
i
r
d
e
te
c
ti
o
n
a
n
d
c
l
a
s
s
if
i
c
a
t
i
o
n
,
w
h
e
r
e
t
h
e
u
s
e
r
se
l
e
c
t
s
o
n
e
o
f
t
h
e
m
,
s
o
t
h
a
t
t
h
e
p
r
o
g
r
a
m
s
e
a
rc
h
e
s
f
o
r
i
t
i
n
t
h
e
w
o
r
k
e
n
v
ir
o
n
m
e
n
t
a
n
d
d
e
l
i
v
e
rs
i
t
i
n
a
s
p
e
c
if
i
c
a
re
a
,
o
v
e
r
c
o
m
i
n
g
d
i
f
f
ic
u
l
t
i
e
s
s
u
c
h
a
s
o
c
c
l
u
s
i
o
n
s
o
f
u
p
t
o
7
0
%
.
T
h
e
se
t
o
o
l
s
w
e
r
e
c
l
a
s
s
if
ie
d
u
s
i
n
g
tw
o
C
N
N
(
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
rk
)
ty
p
e
n
e
tw
o
rk
s
,
a
f
a
s
t
R
-
C
N
N
(
f
a
s
t
r
e
g
i
o
n
-
b
a
s
e
d
C
N
N
)
f
o
r
t
h
e
d
e
t
e
c
t
i
o
n
a
n
d
c
la
s
s
if
ic
a
t
i
o
n
o
f
o
c
c
l
u
s
i
o
n
s
,
a
n
d
a
DA
G
-
C
NN
(
d
i
r
e
c
te
d
a
c
y
c
l
ic
g
ra
p
h
-
C
N
N
)
f
o
r
t
h
e
c
l
a
s
s
if
ic
a
t
i
o
n
t
o
o
l
s
.
F
u
r
t
h
e
r
m
o
r
e
,
a
Ha
a
r
c
l
a
ss
if
i
e
r
w
a
s
tr
a
i
n
e
d
i
n
o
r
d
e
r
t
o
c
o
m
p
a
re
i
ts
a
b
i
l
i
ty
t
o
r
e
c
o
g
n
i
z
e
o
c
c
l
u
s
i
o
n
s
w
i
t
h
r
e
s
p
e
c
t
t
o
t
h
e
f
a
s
t
R
-
C
NN
.
F
a
s
t
R
-
C
NN
a
n
d
D
A
G
-
C
N
N
a
c
h
i
e
v
e
d
7
0
.
9
%
a
n
d
9
6
.
2
%
a
c
c
u
ra
c
y
,
r
e
s
p
e
c
t
iv
e
ly
,
H
a
a
r
c
la
s
s
if
i
e
rs
w
i
t
h
a
b
o
u
t
5
0
%
a
c
c
u
ra
c
y
,
a
n
d
a
n
a
c
c
u
ra
c
y
o
f
g
r
i
p
a
n
d
d
e
l
i
v
e
ry
o
f
o
c
c
l
u
d
e
d
o
b
j
e
c
t
s
o
f
9
0
%
i
n
t
h
e
a
p
p
l
i
c
a
t
i
o
n
,
w
a
s
a
c
h
i
e
v
e
d
.
K
ey
w
o
r
d
s
:
DAG
-
C
NN
Fas
t R
-
C
NN
Haa
r
class
i
f
ier
s
Ma
ch
i
n
e
v
i
s
io
n
Ma
n
ip
u
la
to
r
r
o
b
o
t
Occ
lu
d
ed
o
b
j
ec
ts
VR
M
L
Co
p
y
rig
h
t
©
2
0
2
0
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
R
o
b
in
s
o
n
J
i
m
en
ez
-
Mo
r
en
o
,
Dep
ar
t
m
en
t o
f
Me
ch
atr
o
n
ic
E
n
g
i
n
ee
r
i
n
g
,
Militar
Nu
e
v
a
Gr
an
ad
a
Milita
r
y
U
n
i
v
er
s
it
y
,
C
ar
r
er
a
1
1
#
1
0
1
-
8
0
,
B
o
g
o
tá,
C
o
lo
m
b
ia
.
E
m
ail:
r
o
b
in
s
o
n
.
j
i
m
en
ez
@
u
n
i
m
ilit
ar
.
ed
u
.
co
1.
I
NT
RO
D
UCT
I
O
N
W
o
r
k
in
g
co
n
d
itio
n
s
t
h
at
o
cc
u
r
in
an
u
n
k
n
o
w
n
p
h
y
s
ical
en
v
ir
o
n
m
e
n
t
h
a
v
e
led
r
o
b
o
tics
to
th
e
n
ee
d
to
in
cr
ea
s
e
th
e
in
d
ep
en
d
e
n
ce
o
f
r
o
b
o
ts
to
p
er
f
o
r
m
ta
s
k
s
th
a
t
o
v
er
co
m
e
v
ar
io
u
s
d
i
f
f
icu
lti
es
n
o
t
f
o
r
eseen
in
th
e
en
v
ir
o
n
m
en
t,
s
u
c
h
as
th
e
p
r
esen
ce
o
f
o
cc
lu
s
io
n
s
d
u
r
in
g
t
h
e
tr
ac
k
o
f
a
tr
aj
ec
to
r
y
f
o
r
a
m
an
ip
u
lato
r
r
o
b
o
t
[
1
]
,
o
r
v
ar
io
u
s
o
b
s
tacle
s
th
at
m
o
v
e
ar
o
u
n
d
th
e
e
n
v
ir
o
n
m
e
n
t,
w
h
er
e
t
h
e
m
an
ip
u
lat
o
r
m
u
s
t
r
et
h
i
n
k
it
s
tr
aj
ec
to
r
y
w
h
e
n
s
aid
o
b
s
tacle
is
to
o
clo
s
e
to
i
ts
s
tr
u
c
tu
r
e,
to
av
o
id
a
co
llis
io
n
[
2
]
.
Ho
w
ev
er
,
th
e
p
r
esen
ce
o
f
o
b
s
tacle
s
in
th
e
p
ath
o
f
a
m
a
n
ip
u
lato
r
is
n
o
t
th
e
o
n
l
y
u
n
f
o
r
eseen
ev
e
n
t
th
a
t
m
u
s
t
b
e
o
v
er
co
m
e,
b
u
t
o
cc
lu
s
io
n
s
in
th
e
ele
m
e
n
ts
to
b
e
f
aste
n
e
d
as
m
e
n
tio
n
ed
in
[
3
]
,
w
h
er
e
an
R
GB
-
D
s
en
s
o
r
w
a
s
u
s
e
d
in
th
e
g
r
ip
p
er
o
f
th
e
m
a
n
ip
u
lato
r
to
g
e
n
er
ate
a
3
D
v
o
x
e
l
m
ap
o
f
th
e
en
v
ir
o
n
m
e
n
t,
w
h
ic
h
a
llo
w
s
to
r
ec
o
g
n
ize
t
h
e
o
cc
lu
d
ed
o
b
j
ec
ts
an
d
g
e
n
er
ate
a
tr
aj
ec
to
r
y
f
o
r
its
s
u
b
j
ec
tio
n
w
i
th
o
u
t
co
llis
io
n
s
.
Ki
n
ec
t
s
e
n
s
o
r
s
is
an
R
GB
-
D
ca
m
er
a
u
s
ed
f
o
r
co
n
tr
o
l r
o
b
o
ts
[
4
,
5]
I
n
w
o
r
k
s
s
u
c
h
as
t
h
o
s
e
p
r
es
en
ted
i
n
[
6
]
an
d
[
7
]
,
alg
o
r
it
h
m
s
f
o
c
u
s
ed
o
n
t
h
e
o
cc
lu
s
io
n
d
etec
tio
n
p
r
o
ce
s
s
h
av
e
b
ee
n
d
ev
elo
p
ed
,
to
u
s
e
th
is
in
f
o
r
m
at
io
n
in
t
h
e
m
o
v
e
m
e
n
t
o
f
m
an
ip
u
lato
r
an
d
m
o
b
ile
r
o
b
o
ts
,
w
h
er
e
[
6
]
p
er
f
o
r
m
ed
an
al
g
o
r
ith
m
o
f
o
cc
l
u
s
io
n
ed
g
e
d
e
tectio
n
,
u
s
in
g
C
NN
w
it
h
R
GB
,
R
GB
-
D,
a
n
d
R
GB
-
D
-
UV
i
n
p
u
t
i
m
ag
e
s
an
d
v
id
eo
s
,
w
h
er
e
D
r
ep
r
ese
n
ts
t
h
e
d
ep
th
an
d
UV
h
o
r
izo
n
tal
an
d
v
er
tical
co
m
p
o
n
e
n
t
s
o
f
th
e
o
p
tical
f
lo
w
f
ield
,
th
u
s
allo
w
in
g
d
if
f
er
en
t
iatio
n
b
et
w
ee
n
o
c
clu
s
io
n
ed
g
es,
o
f
th
e
ap
p
ea
r
an
ce
ed
g
es
o
f
t
h
e
d
esire
d
ele
m
en
t
s
,
w
h
ile
i
n
[
7
]
,
an
alg
o
r
ith
m
w
a
s
m
ad
e
f
o
r
th
e
d
etec
tio
n
o
f
m
u
ltip
le
p
eo
p
le
in
a
r
ea
l
-
w
o
r
ld
en
v
ir
o
n
m
e
n
t,
u
s
in
g
t
h
er
m
al
an
d
d
ep
th
in
f
o
r
m
atio
n
,
to
d
eter
m
in
e
t
h
e
p
o
s
itio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
lg
o
r
ith
m
o
f d
etec
tio
n
,
cl
a
s
s
if
ica
tio
n
a
n
d
g
r
ip
p
in
g
o
f
o
cc
lu
d
ed
o
b
jects b
y
C
N
N
tech
n
iq
u
es
... (
P
a
u
la
Usech
e
)
4713
o
f
ea
ch
p
er
s
o
n
f
r
o
m
th
e
ir
t
h
er
m
al
d
etec
tio
n
,
a
n
d
d
ef
i
n
e
t
h
e
p
r
esen
ce
o
f
o
cc
lu
s
io
n
s
ac
co
r
d
in
g
to
t
h
e
r
es
u
lt
s
o
f
th
e
d
ep
th
in
f
o
r
m
a
tio
n
,
to
g
e
n
e
r
ate
a
f
r
ee
p
ath
o
f
o
b
s
tacle
s
f
o
r
a
m
o
b
ile
r
o
b
o
t.
C
N
N'
s
a
r
e
d
es
ig
n
e
d
t
o
p
e
r
f
o
r
m
r
e
c
o
g
n
i
t
i
o
n
o
f
d
e
s
i
r
e
d
p
a
tt
e
r
n
s
a
n
d
c
h
a
r
a
c
t
e
r
is
ti
c
s
i
n
im
a
g
e
s
,
to
c
l
a
s
s
if
y
t
h
em
in
t
o
a
s
p
e
c
if
ic
c
at
eg
o
r
y
,
a
s
m
en
t
i
o
n
e
d
in
[
8
,
9
]
.
T
h
e
C
N
N
is
c
o
m
p
o
s
e
d
o
f
c
o
n
v
o
lu
t
i
o
n
al
f
il
t
e
r
s
w
h
o
s
e
p
a
r
am
et
e
r
s
a
r
e
t
r
a
in
e
d
w
ith
a
d
a
t
a
b
a
s
e
th
a
t
h
a
s
im
ag
es
o
r
r
ef
er
e
n
c
es
o
f
t
h
e
p
a
tt
e
r
n
s
t
o
b
e
c
l
as
s
i
f
i
e
d
,
w
h
ich
a
ll
o
w
t
h
e
n
e
t
w
o
r
k
t
o
le
a
r
n
an
d
ex
t
r
a
c
t
th
e
m
o
s
t
r
e
le
v
a
n
t
ch
a
r
a
c
t
er
i
s
t
ic
s
o
f
an
im
ag
e
,
w
h
o
s
e
i
n
f
o
r
m
a
ti
o
n
is
u
s
e
d
to
g
en
e
r
a
te
a
c
l
ass
if
i
c
at
i
o
n
,
a
s
e
x
p
l
a
i
n
e
d
in
[
1
0
]
,
b
u
t
n
o
t
o
n
ly
im
a
g
es
l
ik
e
i
s
ex
p
o
s
e
d
in
[
1
1
]
.
So
m
e
o
f
th
e
r
ec
en
tl
y
d
e
v
elo
p
ed
C
NN
ap
p
licatio
n
s
ar
e:
1
)
a
34
-
la
y
er
C
NN
led
to
t
h
e
d
e
tectio
n
o
f
a
w
id
e
r
an
g
e
o
f
ca
r
d
iac
ar
r
h
y
th
m
ia
s
f
r
o
m
elec
tr
o
ca
r
d
io
g
r
a
m
s
,
w
h
o
s
e
p
er
f
o
r
m
a
n
ce
e
x
c
ee
d
ed
th
e
av
er
ag
e
r
esu
lt
s
o
f
Me
d
ical
p
r
ed
ictio
n
,
m
ad
e
b
y
a
g
r
o
u
p
o
f
6
ca
r
d
io
lo
g
is
t
s
[
1
2
]
;
2
)
a
tr
ain
ed
ca
s
ca
d
e
o
f
C
NNs
to
d
etec
t
an
d
class
i
f
y
f
ac
es
i
n
a
r
ea
l
en
v
ir
o
n
m
e
n
t,
o
v
er
co
m
i
n
g
d
if
f
i
c
u
lties
s
u
c
h
as
p
o
s
e
ch
an
g
es,
ex
p
r
ess
io
n
,
an
d
lig
h
ti
n
g
,
a
v
o
id
in
g
b
ec
o
m
in
g
co
m
p
u
tatio
n
all
y
e
x
p
en
s
i
v
e
[
1
3
]
;
3
)
e
x
p
an
d
ed
u
s
e
o
f
C
NN
s
to
w
ar
d
s
a
th
r
ee
-
d
i
m
en
s
io
n
al
en
v
ir
o
n
m
e
n
t,
wh
er
e
v
o
l
u
m
es
o
f
m
a
g
n
e
tic
r
eso
n
an
ce
v
o
x
el
o
f
t
h
e
p
r
o
s
t
ate
w
er
e
ev
a
lu
ated
,
g
en
er
ati
n
g
a
s
e
g
m
en
tatio
n
o
f
th
e
e
n
tire
v
o
l
u
m
e,
u
s
i
n
g
o
n
l
y
a
f
r
ac
tio
n
o
f
p
r
o
ce
s
s
i
n
g
ti
m
e
co
m
p
ar
ed
to
o
th
er
m
et
h
o
d
s
p
r
ev
io
u
s
l
y
u
s
ed
[
1
4
]
.
B
esid
es
C
NN,
o
th
er
ar
tif
icia
l
in
telli
g
en
ce
m
et
h
o
d
s
h
av
e
b
ee
n
d
ev
elo
p
ed
f
o
r
th
e
class
i
f
icatio
n
o
f
p
atter
n
s
,
s
u
ch
a
s
t
h
e
f
a
s
t
R
-
C
NN
[
1
5
]
an
d
th
e
D
AG
-
C
N
N
[
1
6
,
1
7
]
,
in
th
e
f
ir
s
t
ca
s
e
i
s
f
o
r
m
er
a
s
tag
e
o
f
ex
tr
ac
tio
n
o
f
a
R
e
g
io
n
s
o
f
I
n
t
er
est
(
R
OI
s
)
th
a
t
is
r
esp
o
n
s
ib
le
f
o
r
o
f
d
etec
tin
g
d
esire
d
ele
m
en
ts
i
n
t
h
e
in
p
u
t
i
m
a
g
e,
ex
tr
ac
ti
n
g
th
e
m
a
n
d
en
ter
in
g
t
h
e
m
in
to
a
C
NN
f
o
r
class
i
f
icatio
n
,
as
e
x
p
lai
n
ed
in
[
1
8
]
,
w
h
i
le
th
e
la
s
t
co
n
s
is
ts
o
f
a
b
r
an
c
h
ed
s
tr
u
ct
u
r
e
w
h
er
e
ea
c
h
b
r
an
ch
co
n
tain
s
a
s
eq
u
en
ce
o
f
co
n
v
o
l
u
tio
n
al
l
a
y
er
s
w
h
o
s
e
f
i
lter
s
v
ar
y
o
f
d
i
m
e
n
s
io
n
,
to
ex
tr
ac
t
ch
ar
ac
ter
is
tic
s
o
f
g
r
ea
ter
an
d
s
m
al
ler
s
ize
o
f
t
h
e
i
n
p
u
t
i
m
a
g
e,
an
d
in
t
h
e
en
d
to
un
i
f
y
t
h
e
r
esu
l
ts
to
g
i
v
e
a
clas
s
if
ica
tio
n
,
as i
n
d
icate
d
in
[
1
9
]
.
So
m
e
e
x
a
m
p
le
s
o
f
ap
p
licatio
n
s
f
o
r
th
e
f
as
t
R
-
C
NN
ar
e
r
ep
o
r
ted
in
[
2
0
]
an
d
[
2
1
]
.
T
h
e
f
ir
s
t
r
ep
o
r
t
ap
p
lied
th
e
Fas
ter
R
-
C
NN
f
o
r
f
ac
e
d
etec
tio
n
an
d
class
if
icatio
n
,
to
ac
h
ie
v
e
a
h
i
g
h
er
p
r
o
ce
s
s
in
g
s
p
ee
d
co
n
ce
r
n
i
n
g
o
th
er
m
et
h
o
d
s
o
f
d
ee
p
lear
n
in
g
,
a
n
d
t
h
e
s
ec
o
n
d
r
ep
o
r
t
a
m
u
lti
-
clas
s
f
r
u
i
t
d
etec
tio
n
u
s
i
n
g
a
r
o
b
o
tic
v
is
io
n
s
y
s
te
m
b
ased
o
n
Fas
ter
R
C
NN.
O
n
t
h
e
o
t
h
er
h
a
n
d
,
th
e
D
A
G
-
C
N
N
h
a
s
b
ee
n
u
s
ed
i
n
ap
p
licatio
n
s
s
u
c
h
as
th
o
s
e
d
escr
ib
ed
in
[
2
2
]
an
d
[
2
3
]
,
w
h
er
e
t
h
e
f
ir
s
t
p
u
b
licati
o
n
u
s
ed
t
h
e
D
A
G
-
C
N
N
f
o
r
th
e
esti
m
a
tio
n
o
f
a
g
e
in
p
eo
p
le
o
f
d
if
f
er
en
t
g
e
n
d
er
s
an
d
eth
n
icitie
s
,
tak
i
n
g
ad
v
an
ta
g
e
o
f
th
e
e
x
tr
ac
tio
n
o
f
ch
ar
ac
t
er
is
tics
at
m
u
l
tip
le
s
ca
les
o
f
s
aid
n
et
w
o
r
k
,
w
it
h
an
ac
cu
r
ac
y
o
f
ar
o
u
n
d
8
0
%,
an
d
th
e
s
ec
o
n
d
p
u
b
licatio
n
u
s
ed
th
e
D
A
G
-
C
NN
f
o
r
th
e
clas
s
i
f
icatio
n
o
f
h
ea
r
tb
ea
ts
f
r
o
m
e
lectr
o
ca
r
d
io
g
r
a
m
i
m
ag
es
(
E
C
G)
,
ac
h
iev
in
g
t
h
e
i
d
en
tific
atio
n
o
f
1
5
d
if
f
er
e
n
t si
g
n
als
w
i
th
9
7
.
1
5
% a
cc
u
r
ac
y
.
A
p
ar
t
f
r
o
m
th
e
C
NNs,
t
h
er
e
ar
e
o
th
er
m
e
th
o
d
s
o
f
r
ec
o
g
n
i
zin
g
ele
m
e
n
t
s
in
i
m
a
g
es
,
s
u
c
h
as
Haa
r
class
i
f
ier
s
,
w
h
ich
tr
ai
n
a
s
er
ies
o
f
w
ea
k
ca
s
ca
d
in
g
clas
s
i
f
i
er
s
,
w
h
ich
r
ec
eiv
e
a
n
in
p
u
t
i
m
ag
e
a
n
d
th
r
o
u
g
h
a
s
lid
in
g
w
i
n
d
o
w
t
h
e
y
d
is
c
ar
d
th
o
s
e
ar
ea
s
t
h
at
d
o
n
o
t
co
n
tain
th
e
d
es
ir
ed
ele
m
e
n
t
u
n
til,
i
n
t
h
e
e
n
d
,
th
e
w
in
d
o
w
s
i
n
d
icatin
g
th
e
o
b
j
ec
t
o
f
in
ter
est,
ar
e
o
b
tain
ed
,
[
2
4
]
.
E
x
a
m
p
les
o
f
ap
p
licatio
n
o
f
Haa
r
class
if
ier
s
ar
e
r
ep
o
r
ted
in
[
2
5
]
an
d
[
2
6
]
,
w
h
er
e
t
h
e
f
ir
s
t o
n
e
ca
r
r
ied
o
u
t
a
p
r
o
ce
s
s
o
f
r
ea
l
-
ti
m
e
d
etec
ti
o
n
o
f
co
w
n
ip
p
les t
o
g
en
er
ate
an
a
u
to
m
at
ic
m
il
k
i
n
g
s
y
s
te
m
,
w
h
ile
t
h
e
s
ec
o
n
d
d
ev
elo
p
e
d
a
m
eth
o
d
o
f
co
u
n
tin
g
an
d
d
etec
tin
g
th
e
n
u
m
b
er
o
f
b
o
o
k
s
s
tack
ed
o
n
s
h
el
v
es,
w
h
er
e
9
6
%
r
ec
o
g
n
itio
n
w
a
s
ac
h
ie
v
ed
w
h
en
tes
tin
g
th
e
s
y
s
te
m
o
n
a
to
tal
o
f
2
0
s
h
elv
e
s
an
d
2
3
3
b
o
o
k
s
.
Fo
r
th
e
f
o
llo
w
i
n
g
ar
ticle,
a
s
y
s
te
m
f
o
r
d
etec
tin
g
,
clas
s
if
y
i
n
g
a
n
d
g
r
ab
b
in
g
o
c
clu
d
ed
o
b
j
ec
ts
w
a
s
d
esig
n
ed
u
s
in
g
a
m
a
n
ip
u
lato
r
r
o
b
o
t
in
a
p
h
y
s
ical
a
n
d
v
i
r
tu
al
e
n
v
ir
o
n
m
e
n
t,
w
h
er
e
ar
t
if
icial
in
telli
g
en
c
e
tech
n
iq
u
es
w
er
e
u
s
ed
f
o
r
th
e
p
r
o
ce
s
s
o
f
d
etec
tio
n
an
d
c
lass
i
f
icatio
n
o
f
ele
m
e
n
ts
i
n
t
h
e
w
o
r
k
ar
ea
,
an
d
a
Kin
ec
t
V1
w
a
s
u
s
ed
as
th
e
m
ac
h
i
n
e
v
is
io
n
s
y
s
te
m
.
T
h
is
w
o
r
k
g
en
er
ate
s
a
f
u
n
d
a
m
en
t
al
co
n
tr
ib
u
tio
n
f
o
r
th
e
i
n
cr
ea
s
e
o
f
t
h
e
a
u
to
m
atio
n
o
f
r
o
b
o
tized
p
r
o
ce
s
s
es
w
it
h
in
u
n
k
n
o
w
n
en
v
ir
o
n
m
e
n
ts
,
wh
er
e
a
s
eq
u
en
ce
o
f
d
etec
tio
n
an
d
eli
m
in
atio
n
o
f
o
cc
lu
s
io
n
s
w
as
estab
li
s
h
ed
s
i
n
ce
it
is
n
o
t
p
o
s
s
ib
l
e
to
d
o
d
g
e
th
e
m
e
m
p
lo
y
i
n
g
o
b
s
tacle
av
o
id
an
ce
alg
o
r
it
h
m
s
,
b
u
t th
e
y
m
u
s
t b
e
r
e
m
o
v
ed
to
b
e
ab
le
to
h
o
ld
th
e
d
esire
d
o
b
ject.
T
h
e
p
r
esen
t a
r
ticle
is
d
iv
id
ed
i
n
to
4
m
ai
n
s
ec
tio
n
s
.
I
n
t
h
e
f
ir
s
t sectio
n
,
a
b
r
ief
i
n
tr
o
d
u
ctio
n
w
a
s
m
ad
e.
I
n
th
e
s
ec
o
n
d
s
ec
t
io
n
,
t
h
e
o
p
er
atio
n
o
f
t
h
e
al
g
o
r
ith
m
p
r
o
p
o
s
ed
f
o
r
th
e
eli
m
i
n
atio
n
o
f
o
cc
l
u
s
io
n
s
i
s
ex
p
lai
n
ed
.
I
n
th
e
th
ir
d
s
ec
t
io
n
,
ca
r
r
ied
o
u
t
o
n
t
h
e
al
g
o
r
ith
m
i
n
a
p
h
y
s
ical
en
v
ir
o
n
m
e
n
t,
to
o
b
tain
t
h
e
q
u
ali
t
y
o
f
r
ec
o
g
n
itio
n
an
d
d
eli
v
er
y
o
f
d
e
s
ir
ed
ele
m
e
n
ts
,
i
n
t
h
e
p
r
esen
c
e
o
f
p
ar
tial
o
cc
lu
s
io
n
s
.
F
in
al
l
y
,
in
t
h
e
last
s
ec
tio
n
,
a
s
er
ies
o
f
co
n
clu
s
io
n
s
w
er
e
estab
lis
h
ed
r
eg
ar
d
in
g
t
h
e
o
p
er
atio
n
o
f
th
e
al
g
o
r
ith
m
,
ac
co
r
d
in
g
to
th
e
r
es
u
lts
o
b
tain
ed
d
u
r
in
g
t
h
e
tes
ts
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
e
f
o
llo
w
i
n
g
is
t
h
e
b
asic
o
p
er
atio
n
o
f
th
e
p
r
o
g
r
a
m
f
o
r
d
etec
tin
g
,
clas
s
if
y
i
n
g
an
d
g
r
ab
b
in
g
o
cc
lu
d
ed
o
b
j
ec
ts
,
u
s
i
n
g
ar
ti
f
i
cial
in
tel
lig
e
n
ce
tec
h
n
iq
u
es,
an
d
ca
p
tu
r
ed
th
r
ee
-
d
i
m
e
n
s
io
n
al
i
n
f
o
r
m
atio
n
b
y
Kin
ec
t
V1
(
s
ee
[
2
7
]
)
.
Fo
r
th
is
,
s
ec
tio
n
2
w
a
s
d
iv
id
ed
in
t
o
2
s
u
b
s
ec
tio
n
s
,
t
h
e
f
ir
s
t
r
aise
s
th
e
p
h
y
s
ical
w
o
r
k
i
n
g
co
n
d
itio
n
s
f
o
r
th
e
i
m
p
le
m
e
n
ta
tio
n
o
f
t
h
e
ap
p
licatio
n
,
th
e
s
ec
o
n
d
d
escr
ib
es th
e
alg
o
r
it
h
m
’
s
f
lo
w
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
4
7
1
2
-
4720
4714
T
h
e
d
ev
elo
p
ed
alg
o
r
ith
m
c
ap
tu
r
es
a
n
R
GB
i
m
a
g
e
o
f
th
e
w
o
r
k
e
n
v
ir
o
n
m
en
t,
t
h
en
ac
q
u
ir
es
th
e
th
r
ee
-
d
i
m
en
s
io
n
al
i
n
f
o
r
m
atio
n
o
f
th
e
e
n
v
ir
o
n
m
e
n
t
to
d
etec
t
th
e
h
eig
h
t
o
f
ea
c
h
e
le
m
e
n
t
co
n
ce
r
n
i
n
g
th
e
tab
le,
an
d
ac
co
r
d
in
g
to
th
e
d
is
p
o
s
itio
n
o
f
t
h
e
f
o
u
n
d
o
b
j
ec
ts
,
a
g
r
ip
alg
o
r
ith
m
is
g
e
n
er
ated
th
at
allo
ws
to
r
e
m
o
v
e
t
h
e
to
o
ls
u
s
in
g
a
m
an
ip
u
lato
r
r
o
b
o
t
an
d
r
elo
ca
te
th
e
m
i
n
t
h
e
d
eli
v
er
y
ar
ea
.
T
h
e
s
eq
u
en
ce
o
f
o
p
er
atio
n
o
f
th
e
alg
o
r
it
h
m
is
s
h
o
w
n
i
n
t
h
e
f
lo
w
ch
ar
t
o
f
Fig
u
r
e
1
,
w
h
er
e
ea
c
h
s
tep
o
f
t
h
e
p
r
o
g
r
a
m
w
a
s
m
ar
k
ed
w
it
h
n
u
m
b
er
s
f
r
o
m
1
to
5
,
w
h
ic
h
ar
e
d
escr
ib
ed
in
th
e
f
o
llo
w
i
n
g
s
u
b
s
ec
tio
n
s
.
Ne
x
t,
th
e
o
p
er
atio
n
o
f
th
e
al
g
o
r
ith
m
is
e
x
p
lai
n
ed
,
th
e
s
tep
s
in
Fi
g
u
r
e
1
ar
e
d
escr
ib
ed
,
an
d
its
ap
p
licatio
n
i
n
th
e
p
h
y
s
ical
an
d
v
ir
tu
a
l
en
v
ir
o
n
m
e
n
t i
s
s
h
o
w
n
.
Fig
u
r
e
1
.
A
l
g
o
r
ith
m
f
lo
w
c
h
ar
t
2
.
1
.
M
e
t
ho
ds
a
nd
m
a
t
er
ia
ls
A
p
h
y
s
ical
w
o
r
k
e
n
v
ir
o
n
m
e
n
t
o
f
3
0
x
2
0
c
m
ar
ea
w
as
e
s
tab
lis
h
ed
,
at
a
h
ei
g
h
t
o
f
6
c
m
f
r
o
m
t
h
e
g
r
o
u
n
d
,
w
it
h
a
b
l
u
e
h
u
e
li
k
e
t
h
e
o
n
e
s
h
o
w
n
i
n
F
ig
u
r
e
2
(
a
)
.
A
Ki
n
ec
t
V1
w
a
s
u
s
ed
to
ca
p
tu
r
e
th
e
w
o
r
k
e
n
v
ir
o
n
m
en
t
an
d
w
a
s
lo
ca
ted
at
a
h
eig
h
t
o
f
9
8
cm
f
r
o
m
t
h
e
g
r
o
u
n
d
,
in
th
e
ce
n
ter
o
f
th
e
w
o
r
k
ar
ea
tab
le
o
r
G
r
ip
Z
o
n
e.
A
Del
iv
er
y
Z
o
n
e
w
as
e
s
tab
l
is
h
ed
,
w
h
er
e
t
w
o
r
e
g
io
n
s
wer
e
d
em
ar
ca
ted
to
o
r
g
an
ize
th
e
to
o
ls
,
o
n
e
f
o
r
th
e
Occ
l
u
s
io
n
s
an
d
th
e
o
t
h
er
f
o
r
th
e
Desire
d
Ob
j
ec
ts
,
as
s
h
o
w
n
i
n
Fi
g
u
r
e
2
(
b
)
.
T
h
e
r
o
b
o
t
u
s
ed
is
s
h
o
w
n
i
n
Fig
u
r
e
2
(
c
)
an
d
h
as
3
DOF
(
Deg
r
ee
s
o
f
Fre
ed
o
m
)
,
all
r
o
t
atio
n
al.
I
ts
s
ch
e
m
atic
r
ep
r
esen
tatio
n
i
s
s
h
o
w
n
i
n
Fig
u
r
e
2
(
d
)
,
w
h
er
e
th
e
a
n
g
les
o
f
r
o
tatio
n
w
er
e
m
ar
k
ed
,
th
e
j
o
in
ts
f
r
o
m
1
to
3
w
er
e
lis
ted
,
an
d
th
e
li
n
k
s
f
r
o
m
e1
to
e3
w
er
e
m
ar
k
ed
.
Fiv
e
t
y
p
e
s
o
f
to
o
ls
w
er
e
s
e
lec
ted
f
o
r
t
h
e
ap
p
licatio
n
:
S
ca
lp
el,
Scis
s
o
r
,
Scr
e
w
d
r
i
v
er
,
Sp
an
n
er
,
an
d
P
lier
s
,
w
h
er
e
ea
c
h
o
n
e
was
d
esi
g
n
ed
w
i
th
a
li
g
h
t
an
d
s
o
f
t
m
ater
ial
(
f
o
a
m
)
t
h
at
f
ac
ili
t
ates
t
h
e
ad
j
u
s
t
m
en
t
o
f
th
e
g
r
ip
p
er
o
n
th
e
ele
m
en
t
s
an
d
r
ed
u
ce
s
th
e
i
n
f
lu
e
n
ce
o
f
th
e
w
ei
g
h
t
o
f
t
h
e
o
b
j
ec
ts
o
n
th
e
d
y
n
a
m
ics
o
f
th
e
m
a
n
ip
u
lato
r
.
E
ac
h
o
b
j
ec
t
w
a
s
m
a
n
u
f
ac
t
u
r
ed
w
i
th
1
c
m
t
h
ick
n
e
s
s
an
d
m
ax
i
m
u
m
d
i
m
e
n
s
io
n
s
o
f
3
.
5
x
1
0
c
m
,
an
d
f
o
r
ea
ch
ca
teg
o
r
y
,
th
e
s
h
a
p
e
an
d
co
lo
r
o
f
th
e
to
o
ls
w
er
e
v
ar
ied
,
in
o
r
d
er
to
teac
h
t
h
e
p
r
o
g
r
a
m
to
r
ec
o
g
n
iz
e
ea
ch
ele
m
e
n
t
w
i
th
to
ler
an
ce
f
o
r
v
ar
iatio
n
s
.
(
a)
(
b
)
©
(
d
)
Fig
u
r
e
2
.
(
a)
W
o
r
k
tab
le,
(
b
)
g
r
ip
an
d
d
eliv
er
y
zo
n
es o
f
ele
m
e
n
ts
,
(
c)
an
t
h
r
o
p
o
m
o
r
p
h
ic
r
o
b
o
t,
an
d
(
d
)
s
ch
e
m
atic
r
ep
r
esen
tati
o
n
o
f
th
e
r
o
b
o
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
lg
o
r
ith
m
o
f d
etec
tio
n
,
cl
a
s
s
if
ica
tio
n
a
n
d
g
r
ip
p
in
g
o
f
o
cc
lu
d
ed
o
b
jects b
y
C
N
N
tech
n
iq
u
es
... (
P
a
u
la
Usech
e
)
4715
2
.
2
.
Alg
o
rit
h
m
f
lo
w
Step
1
p
r
o
g
r
am
i
n
it
ializatio
n
:
T
h
e
f
ir
s
t step
co
n
s
i
s
ts
o
f
th
e
i
n
itializa
tio
n
o
f
th
e
v
ar
iab
les to
b
e
u
s
ed
in
th
e
co
u
r
s
e
o
f
t
h
e
ap
p
licatio
n
,
w
h
er
e
i
m
a
g
D
w
as
d
ef
i
n
ed
a
s
th
e
v
ec
to
r
t
h
at
co
n
ta
in
s
t
h
e
n
u
m
b
er
o
f
p
i
x
els
o
cc
u
p
ied
b
y
t
h
e
Gr
ip
Z
o
n
e
in
t
h
e
R
GB
i
m
a
g
e
o
f
th
e
K
in
ec
t
(
[
1
2
8
1
7
4
]
p
ix
el
s
)
,
an
d
th
e
P
FH
v
ar
iab
le
s
to
r
es
t
w
o
m
atr
ices,
ea
c
h
w
it
h
1
0
r
o
w
s
an
d
3
co
lu
m
n
s
,
wh
er
e
th
e
f
ir
s
t
m
atr
i
x
s
a
v
es
t
h
e
f
in
a
l
p
o
s
itio
n
s
f
o
r
th
e
d
esire
d
o
b
j
ec
ts
,
an
d
th
e
s
ec
o
n
d
s
av
es t
h
e
f
in
a
l p
o
s
itio
n
s
f
o
r
th
e
u
n
w
a
n
ted
ele
m
e
n
t
s
(
o
cc
lu
s
io
n
s
)
.
Step
2
to
o
l
d
etec
tio
n
:
I
n
th
is
s
tep
,
th
e
f
ir
s
t
d
etec
tio
n
o
f
to
o
ls
in
th
e
e
n
v
ir
o
n
m
e
n
t
is
p
er
f
o
r
m
ed
,
w
h
er
e
it
is
d
eter
m
in
ed
w
h
eth
er
t
h
e
p
r
o
g
r
am
is
e
x
ec
u
ted
o
r
ter
m
in
a
ted
ac
co
r
d
in
g
to
w
h
a
t
is
id
en
ti
f
ied
in
th
e
Gr
ip
Z
o
n
e
as
s
h
o
w
n
i
n
Fig
u
r
e
1
.
T
w
o
m
eth
o
d
s
o
f
o
b
j
ec
t
d
etec
tio
n
,
th
e
f
ast
R
-
C
NN
,
an
d
th
e
Haa
r
class
i
f
ier
s
,
w
er
e
tr
ain
ed
i
n
o
r
d
er
to
co
m
p
ar
e
th
e
ac
cu
r
ac
y
a
n
d
p
r
ec
is
io
n
o
f
d
etec
tio
n
o
f
ea
ch
,
an
d
th
u
s
d
ef
i
n
e
w
h
ic
h
o
f
th
e
t
w
o
is
u
s
ed
in
t
h
e
al
g
o
r
ith
m
.
A
d
atab
ase
o
f
4
5
0
p
h
o
to
s
w
as
estab
li
s
h
ed
t
h
at
w
er
e
a
u
g
m
e
n
ted
w
it
h
t
h
e
Data
A
u
g
m
en
tat
io
n
p
r
o
g
r
am
d
e
v
elo
p
ed
in
[
2
8
]
,
r
e
ac
h
in
g
a
to
tal
o
f
3
1
5
0
im
a
g
es
,
w
h
er
e
2
8
0
0
w
er
e
u
s
ed
f
o
r
tr
ain
i
n
g
a
n
d
3
5
0
f
o
r
test
in
g
.
T
h
e
d
etec
tio
n
b
o
x
es
f
o
r
b
o
th
m
et
h
o
d
s
w
er
e
d
ef
in
ed
as
5
5
x
2
8
p
ix
els
s
in
ce
t
h
at
is
t
h
e
m
ax
i
m
u
m
v
a
l
u
e
o
f
t
h
e
s
p
ac
e
o
cc
u
p
ied
b
y
t
h
e
to
o
ls
in
th
e
i
m
a
g
e.
T
h
e
f
ast
R
-
C
NN
i
s
a
C
N
N
-
t
y
p
e
n
e
u
r
al
n
et
w
o
r
k
t
h
at
h
as
an
R
OI
e
x
tr
ac
tio
n
s
ta
g
e
d
e
s
ig
n
ed
f
o
r
th
e
d
etec
tio
n
o
f
ele
m
e
n
ts
in
i
m
a
g
e
s
,
as
d
etailed
i
n
[
1
8
]
.
T
h
e
ar
ch
itect
u
r
e
o
f
t
h
is
n
et
w
o
r
k
is
d
ef
i
n
ed
i
n
th
e
s
a
m
e
w
a
y
as
t
h
at
o
f
a
co
n
v
en
t
io
n
al
C
NN
,
u
s
in
g
co
n
v
o
l
u
tio
n
al
la
y
er
s
(
C
V)
,
r
ec
ti
f
ied
lin
ea
r
u
n
i
ts
(
R
L
U)
,
m
ax
p
o
o
l
(
MP
)
,
av
er
ag
e
p
o
o
l,
b
atch
n
o
r
m
aliza
tio
n
(
B
)
,
f
u
ll
y
co
n
n
ec
ted
(
F
C
)
,
d
r
o
p
o
u
t
(
DO)
,
s
o
f
t
m
a
x
(
S
OFT
)
,
a
m
o
n
g
o
t
h
er
s
,
e
x
p
lain
ed
i
n
[
1
0
]
.
T
h
e
tr
ain
ed
f
ast
R
-
C
NN
h
a
s
t
h
e
ar
ch
i
tectu
r
e
s
h
o
w
n
i
n
T
ab
le
1
,
w
h
er
e
it
w
as
o
b
s
er
v
ed
th
at,
w
h
e
n
u
s
i
n
g
r
ec
tan
g
u
lar
f
il
ter
s
,
4
co
n
v
o
lu
tio
n
la
y
er
s
,
3
0
0
tr
ai
n
in
g
t
i
m
e
s
a
n
d
a
Min
iB
atc
h
Size
less
t
h
an
3
0
,
th
e
h
i
g
h
est
p
e
r
ce
n
tag
e
o
f
ac
c
u
r
ac
y
w
as
ac
h
iev
ed
,
r
ec
eiv
i
n
g
as
in
p
u
t
a
n
i
m
a
g
e
o
f
i
m
a
g
e
d
i
m
en
s
io
n
s
(
[
1
2
8
1
7
4
]
p
ix
e
ls
).
T
ab
le
1
.
A
r
ch
itect
u
r
e
o
f
th
e
f
a
s
t R
-
C
NN
f
o
r
th
e
d
etec
tio
n
an
d
class
if
icatio
n
o
f
o
cc
l
u
d
ed
to
o
ls
L
a
y
e
r
s
C
V
+
B
+
R
L
U
C
V
+
B
+
R
L
U
MP
C
V
+
B
+
R
L
U
C
V
+
B
+
R
L
U
F
C
1
+
R
L
U
+
D
O
F
C
2
+
R
L
U
+
D
O
Fi
l
t
e
r
S
i
z
e
6
x
4
5
x
3
2
x
3
4
x
3
4
x
3
--
--
S
t
r
i
d
e
1
1
2
1
1
--
--
N
u
m
b
e
r
o
f
F
i
l
t
e
r
s
16
1
2
8
--
2
5
6
5
1
2
--
--
T
h
e
r
esu
lts
o
f
th
e
tr
ain
ed
f
ast
R
-
C
NN
ar
e
s
h
o
w
n
in
Fi
g
u
r
e
3
,
w
h
er
e
Fi
g
u
r
e
3
(
a
)
s
h
o
w
s
th
e
co
n
f
u
s
io
n
m
atr
i
x
an
d
Fig
u
r
e
3
(
b
)
s
h
o
w
s
t
h
e
r
ec
all
v
s
p
r
ec
is
io
n
g
r
ap
h
,
w
it
h
1
b
ein
g
th
e
f
r
ee
ca
teg
o
r
y
,
2
b
ein
g
th
e
Occ
l
u
s
io
n
ca
teg
o
r
y
,
a
n
d
3
th
e
B
ac
k
g
r
o
u
n
d
.
T
h
e
r
esu
l
ts
o
b
tain
ed
in
t
h
e
f
ast
R
-
C
N
N
w
er
e
n
o
t
id
ea
l,
h
o
w
ev
er
,
it
w
as
d
ec
id
ed
to
u
s
e
th
is
n
e
t
w
o
r
k
,
s
i
n
ce
th
e
s
tr
u
ct
u
r
e
p
r
o
p
o
s
ed
f
o
r
th
e
alg
o
r
ith
m
allo
w
s
i
n
s
tep
5
to
r
ec
o
g
n
ize
t
h
e
to
o
ls
t
h
at
w
er
e
n
o
t
ca
p
tu
r
ed
in
S
tep
2
,
r
ed
u
cin
g
t
h
e
i
n
f
l
u
en
ce
o
f
th
e
lo
w
a
cc
u
r
ac
y
o
f
th
e
Fas
t
R
-
C
NN
i
n
t
h
e
ap
p
licatio
n
,
as
d
escr
ib
ed
later
in
s
ec
tio
n
3
.
Fo
r
th
e
tr
ain
i
n
g
o
f
t
h
e
Haa
r
cl
ass
i
f
ier
s
,
t
h
e
v
al
u
es
o
f
a
s
er
ie
s
o
f
p
ar
a
m
eter
s
th
a
t
d
eter
m
i
n
e
th
e
b
asic
tr
ain
i
n
g
ch
ar
ac
ter
is
tics
,
e
x
p
lai
n
ed
in
[
2
9
]
,
w
er
e
ad
j
u
s
ted
.
A
Haa
r
class
i
f
ier
w
a
s
tr
ai
n
ed
f
o
r
th
e
r
ec
o
g
n
it
io
n
o
f
th
e
f
r
ee
ca
teg
o
r
y
a
n
d
an
o
th
er
f
o
r
th
e
Occ
lu
s
io
n
ca
te
g
o
r
y
,
wh
o
s
e
ac
c
u
r
ac
y
a
n
d
tr
ain
i
n
g
p
a
r
a
m
eter
s
ar
e
s
h
o
w
n
in
T
ab
le
2
,
w
h
er
e
t
h
e
h
ig
h
e
s
t
ac
cu
r
ac
y
o
b
tai
n
ed
w
as
4
2
%
a
n
d
4
8
%,
r
esp
ec
tiv
el
y
.
As
e
x
p
l
ain
ed
i
n
[
2
2
]
,
Haa
r
class
i
f
ier
s
n
ee
d
lar
g
er
d
atab
as
es
to
o
b
tain
b
etter
d
etec
tio
n
r
esu
lt
s
,
h
o
w
ev
er
,
t
h
e
o
b
j
ec
tiv
e
is
to
co
m
p
ar
e
Haa
r
class
i
f
ier
s
w
i
th
f
ast
R
-
C
N
N
,
th
e
r
ea
s
o
n
w
h
y
t
h
e
d
atab
ase
was n
o
t in
cr
ea
s
ed
an
d
w
a
s
u
s
ed
in
b
o
th
ca
s
es
.
(
a)
(
b
)
Fig
u
r
e
3
.
Fas
t R
-
C
NN
:
(
a)
c
o
n
f
u
s
io
n
m
atr
i
x
an
d
(
b
)
r
ec
all
v
s
p
r
ec
is
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
4
7
1
2
-
4720
4716
T
ab
le
2
.
T
r
ain
in
g
p
ar
a
m
eter
s
o
f
Haa
r
class
i
f
ier
s
P
a
r
a
me
t
e
r
A
c
c
u
r
a
c
y
F
a
l
s
e
A
l
a
r
m
Rate
N
u
mb
e
r
C
a
sc
a
d
e
S
t
a
g
e
s
N
e
g
a
t
i
v
e
S
a
mp
l
e
s
F
a
c
t
o
r
T
r
u
e
P
o
si
t
i
v
e
Rate
T
o
t
a
l
P
o
si
t
i
v
e
s
S
a
mp
l
e
s
T
o
t
a
l
N
e
g
a
t
i
v
e
S
a
mp
l
e
s
N
e
g
a
t
i
v
e
S
a
mp
l
e
s
N
u
mb
e
r
P
o
si
t
i
v
e
S
a
mp
l
e
s
Fr
e
e
4
2
%
0
.
0
1
15
3
6
0
%
2
8
0
0
5
0
0
0
1
2
7
2
4
2
4
Occ
4
8
%
0
.
0
1
1
0
0
3
6
0
%
2
8
0
0
5
0
0
0
2
0
4
68
Step
3
to
o
l
class
i
f
icatio
n
:
Af
t
er
th
e
to
o
l
d
etec
tio
n
p
r
o
ce
s
s
,
th
e
r
ec
o
g
n
ized
ele
m
en
t
s
w
er
e
ex
tr
ac
ted
f
r
o
m
t
h
e
i
m
a
g
e
to
cla
s
s
i
f
y
t
h
e
m
b
y
m
ea
n
s
o
f
a
D
A
G
-
C
NN
i
n
o
n
e
o
f
t
h
e
5
tr
ai
n
ed
ca
teg
o
r
i
es: Scalp
el,
Scis
s
o
r
,
Scr
e
w
d
r
i
v
er
,
Sp
an
n
er
,
Scis
s
o
r
an
d
th
e
r
esu
l
t
w
as
co
m
p
ar
ed
w
it
h
t
h
e
d
esire
d
o
b
j
ec
t
b
y
th
e
u
s
er
,
in
o
r
d
er
to
d
eter
m
in
e
i
f
it
b
elo
n
g
s
to
th
e
g
r
o
u
p
o
f
d
esire
d
o
b
j
ec
ts
o
r
o
c
clu
s
io
n
s
.
A
clas
s
i
f
icatio
n
b
o
x
o
f
7
0
x
7
0
p
ix
els
w
a
s
d
ef
in
ed
,
t
h
e
d
i
m
e
n
s
io
n
s
o
f
wh
ich
w
er
e
tak
e
n
f
r
o
m
t
h
e
u
p
p
er
lef
t
co
r
n
er
o
f
th
e
d
etec
tio
n
b
o
x
g
e
n
er
ated
in
Step
2
o
r
Step
5
,
an
d
7
0
p
ix
e
ls
to
t
h
e
r
i
g
h
t
an
d
7
0
p
ix
el
s
d
o
w
n
w
er
e
ca
p
tu
r
ed
.
I
n
Fi
g
u
r
e
4
(
a
)
,
th
e
s
tr
u
ctu
r
e
u
s
ed
f
o
r
th
e
D
AG
-
C
NN
o
f
t
h
e
ap
p
licatio
n
is
s
h
o
w
n
,
w
h
er
e
a
d
iv
is
io
n
o
f
2
b
r
an
ch
e
s
th
at
r
ec
eiv
e
th
e
i
n
p
u
t
i
m
a
g
e
w
as
g
en
er
ated
an
d
f
i
n
all
y
j
o
in
ed
in
a
F
C
t
o
clas
s
if
y
,
an
d
i
n
Fi
g
u
r
e
4
(
b
)
s
h
o
w
s
a
n
e
x
a
m
p
le
o
f
th
e
d
atab
ase
u
s
ed
f
o
r
n
et
w
o
r
k
tr
ain
i
n
g
.
(
a)
No
in
ter
f
er
e
n
ce
P
lier
s
1
in
ter
f
er
en
ce
Scalp
el
2
in
ter
f
er
en
ce
Scis
s
o
r
3
in
ter
f
er
en
ce
Scr
e
w
d
r
i
v
er
4
o
r
m
o
r
e
in
ter
f
er
e
n
ce
Sp
an
n
er
(
b
)
Fig
u
r
e
4
.
(
a)
Stru
ctu
r
e
o
f
an
d
(
b
)
d
atab
ase
f
o
r
th
e
DA
G
-
C
NN
T
ab
le
3
r
ef
lecte
d
th
e
ar
ch
itectu
r
e
o
f
t
h
e
D
A
G
-
C
NN,
w
h
ic
h
w
a
s
tr
ai
n
ed
f
o
r
9
0
ep
o
ch
s
,
w
it
h
a
d
atab
ase
o
f
3
0
0
0
tr
ain
in
g
i
m
ag
e
s
an
d
3
0
0
test
i
m
a
g
es,
p
er
ca
teg
o
r
y
.
Fig
u
r
e
5
s
h
o
w
s
t
h
e
co
n
f
u
s
io
n
m
atr
i
x
o
f
th
e
D
AG
-
C
NN,
w
h
ich
r
ea
ch
ed
9
6
.
2
%
ac
cu
r
ac
y
,
w
it
h
a
m
in
i
m
u
m
o
f
9
1
%
ac
cu
r
ac
y
p
er
ca
teg
o
r
y
.
T
h
e
n
u
m
b
er
s
co
r
r
esp
o
n
d
to
th
e
c
lass
if
ica
tio
n
ca
teg
o
r
ies
,
w
h
er
e
1
i
s
p
lier
s
,
2
i
s
s
c
alp
el,
3
is
s
c
is
s
o
r
,
4
is
s
cr
e
w
d
r
i
v
er
,
an
d
5
is
s
p
a
n
n
er
.
Fo
r
ex
a
m
p
le
,
t
h
e
c
lass
if
icatio
n
o
f
Sca
lp
el
w
as
o
n
e
o
f
th
e
ca
teg
o
r
ies
t
h
at
p
r
esen
ted
m
o
r
e
d
if
f
icu
lties
i
n
r
ec
o
g
n
itio
n
,
co
n
f
u
s
in
g
its
e
lf
with
t
h
e
P
lier
s
ca
teg
o
r
y
m
ain
l
y
.
T
ab
le
3
.
A
r
ch
itect
u
r
e
o
f
th
e
D
A
G
-
C
NN
(
a)
la
y
er
s
b
r
an
c
h
1
C
V
+
B
+
R
L
U
MP
C
V
+
B
+
R
L
U
MP
C
V
+
B
+
R
L
U
MP
C
V
+
B
+
R
L
U
F
C
1
+
R
L
U
+
D
O
F
C
2
+
R
L
U
+
D
O
F
i
l
t
e
r
S
i
z
e
4
x
4
3
x
3
3
x
3
3
x
3
3
x
3
3
x
3
3
x
3
--
--
S
t
r
i
d
e
1
2
1
2
1
2
1
--
--
N
u
mb
e
r
o
f
F
i
l
t
e
r
s
16
--
1
2
8
--
2
5
6
--
5
1
2
--
--
T
ab
le
3
.
A
r
ch
itect
u
r
e
o
f
th
e
D
A
G
-
C
NN
(
b
)
la
y
er
s
b
r
an
ch
2
C
V
+
B
+
R
L
U
MP
C
V
+
B
+
R
L
U
MP
C
V
+
B
+
R
L
U
MP
F
C
1
+
R
L
U
+
D
O
F
C
2
+
R
L
U
+
D
O
F
i
l
t
e
r
S
i
z
e
6
x
6
3
x
3
4
x
4
3
x
3
3
x
3
2
x
2
--
--
S
t
r
i
d
e
1
2
1
2
1
2
--
--
N
u
mb
e
r
o
f
F
i
l
t
e
r
s
16
--
2
5
6
--
5
1
2
--
--
--
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
lg
o
r
ith
m
o
f d
etec
tio
n
,
cl
a
s
s
if
ica
tio
n
a
n
d
g
r
ip
p
in
g
o
f
o
cc
lu
d
ed
o
b
jects b
y
C
N
N
tech
n
iq
u
es
... (
P
a
u
la
Usech
e
)
4717
Fig
u
r
e
5
.
C
o
n
f
u
s
io
n
m
atr
ix
f
o
r
th
e
D
A
G
-
C
N
N
Step
4
g
r
ip
an
d
d
eliv
er
y
o
f
to
o
ls
:
On
ce
th
e
alg
o
r
it
h
m
d
e
f
in
es
w
h
et
h
er
th
e
to
o
l
is
a
d
esire
d
o
b
j
ec
t
o
r
an
o
cc
lu
s
io
n
,
th
e
m
o
v
e
m
en
t
o
f
th
e
m
a
n
ip
u
lato
r
r
o
b
o
t
f
o
r
th
e
tr
an
s
f
er
o
f
th
e
ele
m
e
n
t
to
th
e
d
eliv
er
y
zo
n
e
i
s
ex
ec
u
ted
.
T
h
e
ce
n
ter
o
f
th
e
d
e
tectio
n
b
o
x
o
f
t
h
e
to
o
l
clas
s
if
i
ed
as
th
e
h
o
ld
in
g
p
o
i
n
t
w
as
s
e
lecte
d
,
an
d
th
e
d
ata
o
f
th
e
P
FH
v
ar
iab
le
w
a
s
u
s
ed
to
d
eter
m
in
e
t
h
e
d
eliv
er
y
co
o
r
d
in
ates
o
f
th
e
ele
m
en
t.
T
o
d
etec
t
th
e
cla
m
p
i
n
g
h
eig
h
t
o
f
ea
ch
to
o
l
co
n
ce
r
n
in
g
th
e
tab
le
(
Z
co
o
r
d
in
ate)
it
w
a
s
n
ec
e
s
s
ar
y
to
ev
al
u
ate
t
h
e
th
r
ee
-
d
i
m
e
n
s
io
n
a
l
in
f
o
r
m
ati
o
n
o
f
ea
ch
d
etec
tio
n
b
o
x
an
d
ca
lc
u
late
its
a
v
er
ag
e.
T
h
e
alg
o
r
ith
m
f
ir
s
t
r
e
m
o
v
es
th
e
e
le
m
e
n
t
s
class
i
f
ied
a
s
f
r
ee
,
an
d
t
h
e
n
t
h
o
s
e
d
ef
i
n
ed
a
s
o
cc
lu
s
io
n
,
c
h
a
n
g
in
g
t
h
e
d
etec
tio
n
b
o
x
o
n
l
y
w
h
en
all
t
h
e
ele
m
e
n
ts
o
f
th
e
to
o
l
s
tack
h
a
v
e
b
ee
n
r
e
m
o
v
ed
,
w
h
ich
m
ea
n
s
th
a
t
th
e
ca
lc
u
lated
a
v
er
ag
e
h
ei
g
h
t
co
r
r
esp
o
n
d
s
to
th
e
h
ei
g
h
t o
f
t
h
e
tab
le.
Step
5
n
e
w
to
o
l
d
etec
tio
n
:
B
ef
o
r
e
ter
m
in
at
in
g
t
h
e
p
r
o
g
r
a
m
,
Step
5
w
as
ad
d
ed
,
to
v
er
if
y
th
e
p
r
esen
ce
o
f
to
o
ls
in
th
e
Gr
ip
Z
o
n
e.
T
h
is
s
tep
is
r
esp
o
n
s
ib
le
f
o
r
r
ep
ea
tin
g
t
h
e
p
r
o
ce
s
s
o
f
d
e
tectio
n
o
f
Step
2
,
th
r
o
u
g
h
w
h
ich
th
e
e
x
is
te
n
ce
o
f
n
e
w
ele
m
e
n
ts
in
th
e
e
n
v
ir
o
n
m
en
t
is
d
ef
i
n
ed
,
an
d
ac
co
r
d
in
g
to
th
e
r
esu
lts
o
b
tain
ed
,
Step
3
an
d
Step
4
ar
e
ex
ec
u
ted
,
o
r
th
e
p
r
o
g
r
a
m
is
ter
m
in
a
ted
,
as sh
o
wn
in
Fig
u
r
e
1
.
3.
RE
SU
L
T
S AN
D
AN
AL
Y
SI
S
T
h
e
f
ir
s
t
co
m
p
ar
is
o
n
b
et
w
ee
n
th
e
m
et
h
o
d
s
o
f
d
etec
tio
n
an
d
class
i
f
icat
io
n
o
f
o
cc
lu
s
io
n
s
m
en
tio
n
ed
in
s
ec
tio
n
2
,
w
h
er
e
th
e
p
er
ce
n
tag
e
s
o
f
ac
cu
r
ac
y
,
w
h
er
e
th
e
f
ast
R
-
C
NN
ex
ce
ed
ed
th
e
Haa
r
class
if
ier
s
b
y
2
0
%
ac
cu
r
ac
y
.
A
d
d
itio
n
all
y
,
a
co
m
p
ar
is
o
n
w
as
m
ad
e
b
et
w
ee
n
b
o
th
m
et
h
o
d
s
ap
p
l
y
i
n
g
t
h
e
m
to
th
e
p
h
y
s
ica
l
w
o
r
k
en
v
ir
o
n
m
e
n
t,
w
it
h
d
if
f
er
en
t
p
er
ce
n
tag
e
s
o
f
o
cc
lu
s
io
n
.
T
h
e
f
ast
R
-
C
NN
m
an
a
g
ed
to
r
ec
o
g
n
ize
a
g
r
ea
ter
n
u
m
b
er
o
f
to
o
ls
in
t
h
e
e
n
v
ir
o
n
m
e
n
t,
a
n
d
ca
teg
o
r
ized
th
e
m
m
o
r
e
ac
c
u
r
atel
y
t
h
an
t
h
e
Haa
r
class
i
f
ier
s
,
i
n
ad
d
itio
n
t
o
p
r
esen
ti
n
g
g
r
ea
te
r
ac
cu
r
ac
y
i
n
t
h
e
d
etec
tio
n
b
o
x
es.
Fo
r
th
e
s
e
r
ea
s
o
n
s
,
it
was
d
ec
id
ed
to
u
s
e
th
e
f
a
s
t
R
-
C
NN
i
n
th
e
al
g
o
r
ith
m
.
T
h
e
alg
o
r
ith
m
w
a
s
ap
p
lied
in
a
p
r
ev
io
u
s
s
i
m
u
latio
n
en
v
i
r
o
n
m
e
n
t
w
as
te
s
ted
b
y
co
m
p
ar
in
g
th
e
n
u
m
b
er
o
f
d
esire
d
to
o
ls
m
o
v
ed
to
t
h
e
d
eli
v
er
y
z
o
n
e
,
w
it
h
r
e
s
p
ec
t
to
t
h
e
to
tal
n
u
m
b
er
o
f
d
esire
d
to
o
ls
p
r
esen
t
i
n
t
h
e
Gr
ip
Z
o
n
e,
t
h
e
r
esu
lt
s
o
f
w
h
ich
ar
e
s
h
o
w
n
i
n
T
ab
le
4
,
w
h
er
e
t
h
e
co
l
u
m
n
“
Del
iv
er
ed
to
o
ls
”
s
h
o
w
s
t
h
e
a
m
o
u
n
t
o
f
Desire
d
o
b
j
ec
ts
d
eliv
er
ed
(
n
u
m
er
at
o
r
)
w
it
h
r
esp
ec
t
to
th
e
to
tal
n
u
m
b
er
o
f
d
es
ir
ed
o
b
j
ec
ts
p
r
esen
t
i
n
t
h
e
e
n
v
ir
o
n
m
e
n
t
(
d
en
o
m
i
n
ato
r
)
f
o
r
tw
o
f
aste
n
i
n
g
test
s
p
er
to
o
l,
an
d
th
e
co
lu
m
n
“
P
er
ce
n
ta
g
e”
s
h
o
w
s
t
h
e
p
er
ce
n
tag
e
s
o
f
s
u
cc
e
s
s
f
u
l
d
eli
v
er
ies
f
o
r
ea
ch
c
ase
ac
co
r
d
in
g
to
th
e
r
es
u
lts
o
f
th
e
co
lu
m
n
“De
li
v
er
ed
to
o
ls
”,
in
w
h
ic
h
t
h
e
n
u
m
b
er
o
f
ele
m
en
ts
,
t
h
eir
p
o
s
itio
n
s
,
t
h
e
o
cc
lu
s
io
n
p
er
ce
n
tag
e
s
,
an
d
th
e
o
b
j
ec
ts
d
esire
d
b
y
th
e
u
s
er
w
as
v
ar
ied
.
A
cc
o
r
d
in
g
to
th
e
r
esu
lts
o
f
T
ab
le
4
,
th
e
al
g
o
r
it
h
m
r
ea
ch
ed
8
0
%
ac
cu
r
ac
y
i
n
t
h
e
p
r
o
ce
s
s
o
f
d
etec
tin
g
,
class
i
f
y
in
g
a
n
d
g
r
ab
b
in
g
o
cc
l
u
d
ed
o
b
j
ec
ts
,
w
ith
in
a
s
i
m
u
lat
ed
en
v
ir
o
n
m
e
n
t,
w
h
ic
h
m
ea
n
s
th
at
a
t
least
4
o
f
5
d
esire
d
o
b
j
ec
ts
w
il
l
b
e
d
eliv
er
ed
co
r
r
ec
tly
.
Du
r
i
n
g
t
h
e
te
s
ts
,
it
w
a
s
p
o
s
s
ib
le
to
o
b
s
er
v
e
t
w
o
f
ac
to
r
s
th
a
t
af
f
ec
ted
th
e
ac
c
u
r
ac
y
o
f
th
e
a
l
g
o
r
ith
m
:
t
h
e
lo
w
ac
cu
r
ac
y
o
f
t
h
e
d
etec
tio
n
b
o
x
e
s
g
e
n
er
ated
b
y
t
h
e
f
ast
R
-
C
NN
,
an
d
th
e
r
es
u
lts
o
f
Step
5
,
w
h
ic
h
d
e
m
o
n
s
tr
ated
in
ab
ili
t
y
to
r
ec
o
g
n
ize
a
n
y
ele
m
e
n
t
in
th
e
s
i
m
u
lated
en
v
ir
o
n
m
e
n
t,
s
i
n
ce
t
h
er
e
ar
e
n
o
v
ar
iatio
n
s
i
n
th
e
i
n
p
u
t i
m
a
g
e
th
at
allo
w
n
e
w
ele
m
e
n
t
s
to
b
e
r
ec
o
g
n
ized
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
4
7
1
2
-
4720
4718
T
ab
le
4
.
Qu
alit
y
o
f
d
eli
v
er
y
o
f
d
e
s
ir
ed
o
b
j
ec
ts
o
f
th
e
s
i
m
u
la
t
ed
alg
o
r
ith
m
D
e
si
r
e
d
o
b
j
e
c
t
s
D
e
l
i
v
e
r
e
d
t
o
o
l
s
P
e
r
c
e
n
t
a
g
e
A
v
e
r
a
g
e
S
c
r
e
w
d
r
i
v
e
r
3
/
4
3
/
3
7
5
%
1
0
0
%
80%
S
c
a
l
p
e
l
3
/
5
2
/
2
6
0
%
1
0
0
%
S
c
i
sso
r
4
/
5
4
/
5
8
0
%
8
0
%
P
l
i
e
r
s
5
/
5
1
/
2
1
0
0
%
5
0
%
S
p
a
n
n
e
r
3
/
4
4
/
5
7
5
%
8
0
%
3
.
1
.
F
un
ct
io
nin
g
in t
he
ph
y
s
ica
l e
nv
iro
n
m
ent
T
o
d
eter
m
i
n
e
t
h
e
p
er
f
o
r
m
a
n
c
e
o
f
th
e
a
lg
o
r
it
h
m
w
it
h
i
n
a
p
h
y
s
ical
e
n
v
ir
o
n
m
en
t,
t
w
o
p
r
e
m
is
e
s
w
er
e
estab
lis
h
ed
,
th
e
f
ir
s
t
f
o
cu
s
ed
o
n
d
eter
m
i
n
in
g
t
h
e
p
er
ce
n
ta
g
e
o
f
to
o
l
s
tack
s
d
etec
ted
w
ith
r
esp
ec
t
to
th
e
to
tal
a
m
o
u
n
t
o
f
to
o
l
to
w
er
s
p
r
ese
n
t
in
t
h
e
Gr
ip
Z
o
n
e
(
Dete
ctio
n
Q
u
alit
y
)
,
an
d
th
e
s
ec
o
n
d
d
ef
i
n
e
d
th
e
p
er
ce
n
tag
e
o
f
to
o
ls
h
eld
co
r
r
ec
tly
,
w
it
h
r
esp
ec
t
to
th
e
to
tal
a
m
o
u
n
t
o
f
ele
m
en
t
s
p
r
esen
t
i
n
th
e
e
n
v
ir
o
n
m
e
n
t
(
Ho
ld
in
g
q
u
alit
y
)
.
T
ab
le
5
r
ef
lecte
d
th
e
r
esu
lt
s
o
b
tain
ed
f
o
r
b
o
th
p
r
em
is
e
s
,
w
h
er
e
th
e
“Stac
k
ed
T
o
o
ls
”
r
o
w
i
n
d
icate
d
th
e
n
u
m
b
er
o
f
to
o
ls
p
r
esen
t
in
a
s
tack
o
f
ele
m
e
n
ts
,
an
d
i
n
th
e
“
R
ei
n
f
o
r
ce
m
e
n
t
Dete
ctio
n
”
r
o
w
th
e
p
er
ce
n
ta
g
e
o
f
r
ec
o
g
n
ized
to
o
l stack
s
w
as
r
ec
o
r
d
e
d
in
th
e
en
v
ir
o
n
m
e
n
t a
f
ter
ex
ec
u
ti
n
g
Step
5
o
n
ce
.
T
ests
w
er
e
ca
r
r
ied
o
u
t
w
i
t
h
s
t
ac
k
s
o
f
ele
m
en
ts
th
a
t
h
a
v
e
b
etw
ee
n
1
an
d
5
to
o
ls
,
w
h
er
e
1
0
test
s
w
er
e
m
ad
e
f
o
r
ea
ch
q
u
a
n
tit
y
o
f
s
ta
ck
ed
ele
m
en
ts
a
n
d
v
ar
ied
f
r
o
m
t
h
eir
p
o
s
i
tio
n
s
to
t
h
eir
o
r
ie
n
tatio
n
s
.
T
h
e
b
est
r
esu
lt
s
o
f
d
etec
tio
n
a
n
d
cla
m
p
in
g
o
f
to
o
ls
w
er
e
o
b
tain
ed
f
o
r
p
iles
o
f
ele
m
e
n
ts
b
et
w
ee
n
1
an
d
3
to
o
ls
,
w
h
er
e
th
eir
av
er
a
g
e
p
r
ec
is
io
n
r
a
n
g
ed
b
et
w
ee
n
8
0
%
an
d
9
0
%
w
h
ile
w
h
e
n
in
cr
ea
s
in
g
to
4
an
d
5
to
o
ls
,
th
e
p
er
ce
n
ta
g
e
d
r
o
p
p
ed
t
o
alm
o
s
t
a
6
0
%.
T
h
e
d
etec
tio
n
f
ail
u
r
es
w
er
e
b
ec
au
s
e,
w
h
e
n
th
e
h
eig
h
t
o
f
t
h
e
s
tack
o
f
o
b
j
ec
ts
in
cr
ea
s
es,
t
h
e
y
g
et
to
o
clo
s
e
to
th
e
Kin
ec
t’
s
ca
m
er
a,
ca
u
s
i
n
g
th
e
m
to
ta
k
e
u
p
a
lo
t
o
f
s
p
ac
e
in
t
h
e
i
m
a
g
e
a
n
d
th
e
d
etec
tio
n
b
o
x
f
a
ils
to
r
ec
o
g
n
ize
t
h
e
m
d
u
e
to
th
eir
d
i
m
en
s
io
n
s
,
w
h
ile
t
h
e
cla
m
p
i
n
g
f
au
lts
w
er
e
d
u
e
to
t
h
e
f
a
c
t
th
a
t
,
w
h
en
c
a
l
cu
la
t
in
g
t
h
e
h
e
ig
h
t
o
f
th
e
e
l
em
en
t
,
th
e
h
e
i
g
h
ts
o
f
th
e
l
o
w
e
r
t
o
o
l
s
o
f
th
e
s
ta
ck
a
r
e
d
e
t
e
ct
e
d
,
c
a
u
s
in
g
t
h
e
av
e
r
ag
e
t
o
b
e
a
lt
e
r
e
d
a
n
d
l
es
s
th
an
th
e
d
es
i
r
e
d
t
o
o
l
.
On
t
h
e
o
th
e
r
h
a
n
d
,
th
e
r
e
in
f
o
r
c
em
en
t
d
e
t
e
ct
i
o
n
p
r
o
v
e
d
t
o
g
en
e
r
a
t
e
a
1
5
.
5
%
im
p
r
o
v
em
en
t
i
n
th
e
d
e
t
e
ct
i
o
n
q
u
a
li
ty
o
f
th
e
p
r
o
g
r
am
,
r
e
a
ch
i
n
g
a
9
0
%
d
e
t
e
ct
i
o
n
o
f
e
l
em
en
ts
w
i
th
a
s
in
g
l
e
ex
e
cu
t
i
o
n
o
f
S
t
e
p
5
,
w
h
i
ch
d
em
o
n
s
t
r
a
t
es
th
e
a
lg
o
r
i
th
m
'
s
a
b
i
l
ity
t
o
r
e
c
o
g
n
iz
e
n
e
w
e
l
em
en
ts
af
t
e
r
th
e
f
i
r
s
t
d
el
iv
e
r
y
o
f
t
o
o
ls
,
an
d
th
e
a
b
i
li
ty
t
o
d
e
liv
e
r
th
em
alm
o
s
t
e
n
t
i
r
e
ly
.
T
ab
le
5
.
Qu
alit
y
o
f
t
h
e
alg
o
r
it
h
m
ap
p
lied
to
a
p
h
y
s
ical
en
v
ir
o
n
m
e
n
t
S
t
a
c
k
e
d
T
o
o
l
s
1
2
3
4
5
1
t
o
3
st
a
c
k
e
d
o
b
j
e
c
t
s
4
t
o
5
st
a
c
k
e
d
o
b
j
e
c
t
s
G
e
n
e
r
a
l
a
v
e
r
a
g
e
D
e
t
e
c
t
i
o
n
Q
u
a
l
i
t
y
(
%)
7
1
.
4
3
1
0
0
7
8
.
5
7
6
4
.
2
9
6
0
.
3
1
8
3
.
3
3
6
2
.
3
7
4
.
9
2
H
o
l
d
i
n
g
Q
u
a
l
i
t
y
(
%)
1
0
0
7
9
.
1
7
8
5
.
1
9
8
8
.
8
9
7
7
.
7
8
8
8
.
1
2
8
3
.
3
4
8
6
.
2
1
R
e
i
n
f
o
r
c
e
me
n
t
D
e
t
e
c
t
i
o
n
(
%)
9
0
.
4
8
4.
CO
NCLU
SI
O
N
T
h
e
alg
o
r
ith
m
o
f
d
etec
tio
n
,
class
i
f
icatio
n
a
n
d
to
o
l
g
r
ip
p
r
o
p
o
s
ed
,
s
u
cc
ess
f
u
l
l
y
d
eli
v
er
ed
8
0
%
o
f
th
e
v
ir
tu
a
l
to
o
ls
,
a
n
d
9
0
%
o
f
to
o
ls
in
th
e
p
h
y
s
ical
e
n
v
ir
o
n
m
en
t,
id
en
t
if
y
i
n
g
t
h
e
p
r
esen
c
e
o
f
o
cc
lu
s
io
n
s
a
n
d
class
i
f
y
in
g
u
p
to
5
d
if
f
er
en
t
t
y
p
e
s
o
f
ele
m
e
n
ts
.
T
h
ese
r
es
u
lt
s
d
e
m
o
n
s
tr
ate
th
e
ab
ilit
y
o
f
t
h
e
alg
o
r
it
h
m
to
h
o
ld
an
d
d
eliv
er
th
e
to
o
ls
d
esire
d
b
y
t
h
e
u
s
er
o
v
er
co
m
in
g
d
if
f
ic
u
lties
s
u
c
h
as
o
cc
lu
s
io
n
s
o
f
o
n
e
an
d
m
o
r
e
ele
m
e
n
ts
.
Step
5
allo
w
ed
to
i
m
p
r
o
v
e
th
e
ab
ilit
y
to
r
ec
o
g
n
i
ze
ele
m
en
ts
i
n
th
e
p
h
y
s
ical
e
n
v
ir
o
n
m
e
n
t
b
y
m
o
r
e
th
an
1
0
%
w
i
th
a
s
in
g
le
e
x
ec
u
tio
n
,
h
o
w
ev
er
,
it
d
id
n
o
t
g
e
n
er
ate
an
y
ef
f
ec
t
o
n
th
e
s
i
m
u
lated
en
v
ir
o
n
m
e
n
t,
b
ec
au
s
e
t
h
e
in
p
u
t
i
m
a
g
e
i
n
St
ep
5
d
o
es
n
o
t
s
h
o
w
lar
g
e
v
ar
i
atio
n
s
w
i
th
r
e
s
p
ec
t
to
th
at
o
f
Step
2
,
w
h
ich
av
o
id
s
a
n
e
w
d
etec
tio
n
o
f
to
o
ls
.
T
h
e
ca
lcu
latio
n
o
f
t
h
e
h
ei
g
h
t
o
f
ea
ch
to
o
l
w
i
th
r
esp
ec
t
to
t
h
e
tab
le
w
as
af
f
ec
ted
b
y
th
e
p
r
esen
ce
o
f
o
th
er
ele
m
e
n
ts
i
n
t
h
e
d
etec
tio
n
b
o
x
,
r
esu
lt
in
g
i
n
a
h
o
ld
in
g
q
u
al
it
y
o
f
8
6
%.
Fin
all
y
,
a
co
m
p
a
r
is
o
n
b
et
w
ee
n
t
w
o
d
et
ec
tio
n
m
et
h
o
d
s
w
as
ac
h
ie
v
ed
,
s
u
c
h
a
s
t
h
e
Haa
r
cla
s
s
i
f
ier
s
an
d
th
e
f
ast
R
-
C
NN
,
w
h
er
e
t
h
e
s
ec
o
n
d
p
r
ese
n
ted
a
2
0
% i
m
p
r
o
v
e
m
e
n
t
in
th
e
d
ete
ctio
n
ac
cu
r
ac
y
w
it
h
r
e
s
p
ec
t to
th
e
f
ir
s
t,
i
n
ad
d
itio
n
to
d
em
o
n
s
tr
ati
n
g
a
b
etter
ab
ilit
y
to
r
ec
o
g
n
ize
ele
m
en
ts
a
n
d
class
i
f
y
t
h
e
m
co
r
r
ec
tl
y
as Fr
ee
o
r
Occ
lu
s
io
n
.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
e
au
th
o
r
s
ar
e
g
r
atef
u
l
to
t
h
e
Un
i
v
er
s
id
ad
Militar
N
u
e
v
a
G
r
an
ad
a,
w
h
ic
h
t
h
r
o
u
g
h
it
’
s
Vi
ce
r
ec
to
r
y
f
o
r
r
esear
ch
s
f
in
a
n
ce
s
t
h
e
p
r
es
en
t p
r
o
j
ec
t w
it
h
co
d
e
I
MP
-
I
NG
-
2290.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
A
lg
o
r
ith
m
o
f d
etec
tio
n
,
cl
a
s
s
if
ica
tio
n
a
n
d
g
r
ip
p
in
g
o
f
o
cc
lu
d
ed
o
b
jects b
y
C
N
N
tech
n
iq
u
es
... (
P
a
u
la
Usech
e
)
4719
RE
F
E
R
E
NC
E
S
[1
]
M
.
Be
n
z
a
o
u
i,
e
t
a
l
.
,
“
T
ra
jec
to
r
y
trac
k
in
g
w
it
h
o
b
sta
c
le
a
v
o
id
a
n
c
e
o
f
re
d
u
n
d
a
n
t
m
a
n
ip
u
lato
r
b
a
se
d
o
n
f
u
z
z
y
in
f
e
re
n
c
e
s
y
ste
m
s
,
”
Ne
u
ro
c
o
mp
u
t
in
g
,
v
o
l.
1
9
6
,
p
p
.
2
3
-
3
0
,
2
0
1
6
.
[2
]
D.
Ha
n
,
e
t
a
l.
,
“
Dy
n
a
m
ic
o
b
sta
c
le
a
v
o
id
a
n
c
e
f
o
r
m
a
n
ip
u
lato
rs
u
si
n
g
d
istan
c
e
c
a
lcu
latio
n
a
n
d
d
isc
r
e
te
d
e
tec
ti
o
n
,
”
Ro
b
o
ti
c
s a
n
d
Co
mp
u
ter
-
In
teg
r
a
te
d
M
a
n
u
fa
c
t
u
rin
g
,
v
o
l
.
4
9
,
p
p
.
9
8
-
1
0
4
,
2
0
1
8
.
[3
]
G
.
Ka
h
n
,
e
t
a
l.
,
“
A
c
ti
v
e
e
x
p
lo
ra
ti
o
n
u
sin
g
traje
c
to
ry
o
p
ti
m
iza
ti
o
n
f
o
r
r
o
b
o
ti
c
g
ra
sp
in
g
in
t
h
e
p
re
se
n
c
e
o
f
o
c
c
lu
sio
n
s
,
”
in
2
0
1
5
IE
EE
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Ro
b
o
t
ics
a
n
d
A
u
to
m
a
ti
o
n
,
ICRA
,
2
0
1
5
,
p
p
.
4
7
8
3
-
4
7
9
0
,
doi
:
1
0
.
1
1
0
9
/ICRA
.
2
0
1
5
.
7
1
3
9
8
6
4
.
[4
]
Q
.
Ch
a
n
g
a
n
d
Z
.
X
io
n
g
,
“
V
isi
o
n
-
a
wa
re
t
a
rg
e
t
re
c
o
g
n
it
io
n
to
w
a
rd
s
a
u
to
n
o
m
o
u
s
ro
b
o
t
b
y
Kin
e
c
t
s
e
n
so
rs,
”
S
ig
n
a
l
Pro
c
e
ss
in
g
:
Ima
g
e
Co
mm
u
n
ica
t
i
o
n
,
2
0
2
0
.
Do
i
:
1
0
.
1
0
1
6
/j
.
im
a
g
e
.
2
0
2
0
.
1
1
5
8
1
0
.
[5
]
X
.
L
i,
“
Hu
m
a
n
–
ro
b
o
t
in
tera
c
ti
o
n
b
a
se
d
o
n
g
e
stu
re
a
n
d
m
o
v
e
m
e
n
t
re
c
o
g
n
it
io
n
,
”
S
i
g
n
a
l
Pr
o
c
e
ss
in
g
:
Ima
g
e
Co
mm
u
n
ica
ti
o
n
,
v
o
l
.
8
1
,
2
0
2
0
.
D
o
i
:
1
0
.
1
0
1
6
/
j.
im
a
g
e
.
2
0
1
9
.
1
1
5
6
8
6
.
[6
]
S
.
S
a
rk
a
r,
e
t
a
l.
,
“
D
e
e
p
lea
rn
in
g
f
o
r
a
u
to
m
a
ted
o
c
c
lu
sio
n
e
d
g
e
d
e
tec
ti
o
n
in
RG
B
-
D
f
r
a
m
e
s
,
”
J
o
u
rn
a
l
o
f
S
i
g
n
a
l
Pro
c
e
ss
in
g
S
y
ste
ms
,
v
o
l
.
8
8
,
n
o
.
2
,
p
p
.
2
0
5
-
2
1
7
,
2
0
1
7
.
[7
]
H.
S
.
Ha
d
i
,
e
t
a
l.
,
“
F
u
si
o
n
o
f
th
e
rm
a
l
a
n
d
d
e
p
th
im
a
g
e
s
f
o
r
o
c
c
lu
sio
n
h
a
n
d
li
n
g
f
o
r
h
u
m
a
n
d
e
tec
ti
o
n
f
ro
m
m
o
b
il
e
ro
b
o
t
,
”
i
n
2
0
1
5
1
0
t
h
Asi
a
n
C
o
n
tr
o
l
Co
n
fer
e
n
c
e
(
AS
CC)
,
p
p
.
1
-
5
,
2
0
1
5
. D
oi
:
1
0
.
1
1
0
9
/A
S
CC.2
0
1
5
.
7
2
4
4
7
2
2
[8
]
A
.
Kriz
h
e
v
s
k
y
,
I.
S
u
tsk
e
v
e
r,
a
n
d
G
.
E.
Hin
to
n
,
“
Im
a
g
e
N
e
t
c
las
si
f
ic
a
ti
o
n
w
it
h
d
e
e
p
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
,
”
Ad
v
a
n
c
e
s i
n
n
e
u
r
a
l
i
n
f
o
rm
a
ti
o
n
p
ro
c
e
ss
in
g
sy
ste
ms
,
p
p
.
1
0
9
7
-
1
1
0
5
,
2
0
1
2
.
[9
]
D
.
Ko
n
sta
n
ti
n
id
is,
e
t
a
l
.
,
“
A
m
o
d
u
lar
CNN
-
b
a
se
d
b
u
il
d
i
n
g
d
e
tec
to
r
f
o
r
re
m
o
te
se
n
sin
g
i
m
a
g
e
s,
”
Co
mp
u
ter
Ne
two
rk
s
,
v
o
l.
1
6
8
,
2
0
2
0
.
D
o
i
:
1
0
.
1
0
1
6
/
j.
c
o
m
n
e
t.
2
0
1
9
.
1
0
7
0
3
4
.
[1
0
]
C.
C.
A
g
g
a
r
w
a
l,
“
Ne
u
ra
l
n
e
tw
o
rk
s a
n
d
d
e
e
p
lea
rn
in
g
,
”
Be
rli
n
,
G
e
rm
a
n
y
:
S
p
rin
g
e
r
,
2
0
1
8
.
[1
1
]
J
.
O
.
P
i
n
z
ó
n
-
A
re
n
a
s
a
n
d
R
.
Ji
m
é
n
e
z
-
M
o
re
n
o
,
“
C
o
m
p
a
riso
n
b
e
tw
e
e
n
h
a
n
d
w
rit
ten
w
o
rd
a
n
d
sp
e
e
c
h
re
c
o
rd
i
n
re
a
l
-
ti
m
e
u
sin
g
CNN
A
rc
h
it
e
c
tu
re
s
,”
In
ter
n
a
t
io
n
a
l
J
o
u
r
n
a
l
o
f
E
lec
trica
l
a
n
d
Co
mp
u
ter
E
n
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
1
0
,
n
o
4
,
p
p
.
4
3
1
3
-
4
3
2
1
,
2
0
2
0
.
[1
2
]
P
.
Ra
jp
u
rk
a
r,
e
t
a
l.
,
“
Ca
rd
io
l
o
g
ist
-
lev
e
l
a
rrh
y
th
m
ia
d
e
tec
ti
o
n
w
it
h
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
,
”
a
rXiv
p
re
p
rin
t
a
rX
iv
:1
7
0
7
.
0
1
8
3
6
,
2
0
1
7
.
[1
3
]
H.
L
i,
e
t
a
l.
,
“
A
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
c
a
sc
a
d
e
f
o
r
f
a
c
e
d
e
t
e
c
ti
o
n
,
”
in
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
IEE
E
c
o
n
fer
e
n
c
e
o
n
c
o
m
p
u
ter
v
isi
o
n
a
n
d
p
a
tt
e
rn
re
c
o
g
n
i
ti
o
n
,
p
p
.
5
3
2
5
-
5
3
3
4
,
2
0
1
5
.
[1
4
]
F
.
M
il
leta
ri,
N.
Na
v
a
b
,
a
n
d
S
.
A
.
A
h
m
a
d
i,
“
V
-
n
e
t:
F
u
ll
y
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
f
o
r
v
o
lu
m
e
tri
c
m
e
d
ica
l
im
a
g
e
s
e
g
m
e
n
tatio
n
,
”
i
n
2
0
1
6
Fo
u
rt
h
I
n
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
3
D
Vi
sio
n
(
3
DV)
,
p
p
.
5
6
5
-
5
7
1
,
2
0
1
6
.
Do
i
:
1
0
.
1
1
0
9
/
3
DV
.
2
0
1
6
.
7
9
.
[1
5
]
J
.
O.
P
in
z
ó
n
-
A
re
n
a
s,
R.
Jim
é
n
e
z
-
M
o
re
n
o
,
a
n
d
Cé
sa
r
G
.
P
a
c
h
ó
n
-
S
u
e
sc
ú
n
,
“
Re
sSeg
:
Re
sid
u
a
l
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
f
o
r
f
o
o
d
se
g
m
e
n
tatio
n
,”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
Co
m
p
u
te
r
En
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
1
0
,
n
o
.
1
,
p
p
.
1
0
1
7
-
1
0
2
6
,
2
0
2
0
.
[1
6
]
J
.
O
.
P
i
n
z
ó
n
-
A
re
n
a
s
,
R
.
J
im
é
n
e
z
-
M
o
r
e
n
o
,
a
n
d
C
é
s
a
r
G
P
a
c
h
ó
n
-
S
u
e
s
c
ú
n
,
“
Of
f
l
i
n
e
s
ig
n
a
t
u
r
e
v
e
r
if
i
c
a
t
io
n
u
s
i
n
g
DA
G
-
C
N
N
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
ic
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
(
I
J
EC
E
)
,
v
o
l
.
9
,
n
o
.
4
,
p
p
.
3
3
1
4
-
3322
,
2
0
1
9
.
[1
7
]
J
.
O.
P
in
z
ó
n
-
A
re
n
a
s
a
n
d
R.
Jim
é
n
e
z
-
M
o
re
n
o
,
“
Ob
jec
t
so
rti
n
g
i
n
a
n
e
x
ten
d
e
d
w
o
rk
a
re
a
u
sin
g
c
o
ll
a
b
o
ra
ti
v
e
ro
b
o
ti
c
s
a
n
d
DA
G
-
CNN
,
”
AR
PN
J
o
u
rn
a
l
o
f
E
n
g
i
n
e
e
rin
g
a
n
d
A
p
p
l
ied
S
c
ien
c
e
s
,
v
o
l.
15
,
n
o
.
2
,
2
0
2
0
.
[1
8
]
R.
G
irsh
ick
,
“
F
a
st
R
-
CNN
,
”
in
Pro
c
e
e
d
in
g
s
o
f
th
e
IE
EE
i
n
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
c
o
m
p
u
te
r
v
isió
n
,
2
0
1
5
,
p
p
.
1
4
4
0
-
1
4
4
8
.
[1
9
]
S
.
Ya
n
g
a
n
d
D.
Ra
m
a
n
a
n
,
“
M
u
lt
i
-
sc
a
le
re
c
o
g
n
it
io
n
w
it
h
DA
G
-
C
NN
s
,
”
in
Pro
c
e
e
d
in
g
s
o
f
th
e
IE
E
E
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
ter
v
isió
n
,
p
p
.
1
2
1
5
-
1
2
2
3
,
2
0
1
5
.
[2
0
]
H.
Jia
n
g
a
n
d
E.
L
e
a
rn
e
d
-
M
il
l
e
r,
“
F
a
c
e
d
e
tec
ti
o
n
w
it
h
th
e
f
a
st
e
r
R
-
CNN
,
”
in
2
0
1
7
1
2
t
h
IEE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Au
to
ma
ti
c
Fa
c
e
&
Ge
stu
re
Rec
o
g
n
it
io
n
(
FG
2
0
1
7
)
,
p
p
.
6
5
0
-
657
,
2
0
1
7
. D
o
i:
1
0
.
1
1
0
9
/
F
G
.
2
0
1
7
.
8
2
[2
1
]
S
.
W
a
n
a
n
d
S
,
G
o
u
d
o
s,
“
F
a
ste
r
R
-
CNN
f
o
r
m
u
lt
i
-
c
las
s
f
ru
it
d
e
te
c
ti
o
n
u
si
n
g
a
ro
b
o
ti
c
v
isio
n
sy
ste
m
,
”
Co
mp
u
ter
Ne
two
rk
s
,
v
o
l.
1
6
8
,
2
0
2
0
.
D
o
i
:
1
0
.
1
0
1
6
/
j.
c
o
m
n
e
t.
2
0
1
9
.
1
0
7
0
3
6
.
[2
2
]
S
.
T
a
h
e
ri
a
n
d
Ö.
T
o
y
g
a
r,
“
On
th
e
u
se
o
f
DA
G
-
C
NN
a
rc
h
it
e
c
tu
re
f
o
r
a
g
e
e
sti
m
a
ti
o
n
w
it
h
m
u
lt
i
-
sta
g
e
f
e
a
tu
re
s
f
u
sio
n
,
”
Ne
u
ro
c
o
m
p
u
t
in
g
,
v
o
l.
3
2
9
,
p
p
.
3
0
0
-
3
1
0
,
2
0
1
9
. D
o
i:
1
0
.
1
0
1
6
/j
.
n
e
u
c
o
m
.
2
0
1
8
.
1
0
.
0
7
1
[2
3
]
Z.
G
o
lri
z
k
h
a
ta
m
i,
S
.
T
a
h
e
ri,
a
n
d
A
.
A
c
a
n
,
“
M
u
lt
i
-
sc
a
le
fe
a
tu
re
s
fo
r
h
e
a
rtb
e
a
t
c
las
sif
ica
ti
o
n
u
sin
g
d
irec
ted
a
c
y
c
li
c
g
ra
p
h
CNN
,
”
Ap
p
li
e
d
Art
if
icia
l
In
telli
g
e
n
c
e
,
v
o
l.
3
2
,
n
o
.
7
-
8
,
p
p
.
6
1
3
-
6
2
8
,
2
0
1
8
.
[2
4
]
M
.
G
.
Krish
n
a
a
n
d
A
.
S
rin
iv
a
su
lu
,
“
F
a
c
e
d
e
tec
ti
o
n
sy
ste
m
o
n
A
d
a
Bo
o
st
a
lg
o
rit
h
m
u
sin
g
Ha
a
r
c
las
si
f
ier
s
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
M
o
d
e
rn
En
g
i
n
e
e
rin
g
Res
e
a
rc
h
,
v
o
l.
2
,
n
o
.
5
,
p
p
.
3
5
5
6
-
3
5
6
0
,
2
0
1
2
.
[2
5
]
A
.
Ra
sto
g
i,
A
.
P
a
l,
a
n
d
B.
S
.
R
y
u
h
,
“
Re
a
l
-
ti
m
e
tea
t
d
e
tec
ti
o
n
u
sin
g
h
a
a
r
c
a
sc
a
d
e
c
l
a
ss
i
f
ier
in
s
m
a
rt
a
u
to
m
a
ti
c
m
il
k
in
g
s
y
ste
m
,
”
in
2
0
1
7
7
t
h
I
EE
E
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Co
n
tro
l
S
y
ste
m,
Co
m
p
u
ti
n
g
a
n
d
E
n
g
i
n
e
e
rin
g
(
ICCS
CE)
,
p
p
.
7
4
-
79
,
2
0
1
7
.
Do
i
:
1
0
.
1
1
0
9
/ICC
S
CE.
2
0
1
7
.
8
2
8
4
3
8
3
.
[2
6
]
A
.
B.
Ka
n
b
u
r
o
g
lu
a
n
d
F
.
B
.
T
e
k
,
“
A
Ha
a
r
Clas
si
f
ier
Ba
se
d
Ca
ll
N
u
m
b
e
r
De
tec
ti
o
n
a
n
d
C
o
u
n
ti
n
g
M
e
th
o
d
f
o
r
L
ib
ra
r
y
Bo
o
k
s,”
2
0
1
8
3
rd
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Co
m
p
u
te
r
S
c
ien
c
e
a
n
d
En
g
in
e
e
rin
g
(
UBM
K)
,
S
a
ra
jev
o
,
p
p
.
5
0
4
-
508
,
2
0
1
8
.
[2
7
]
Z.
Zh
a
n
g
,
“
M
icro
so
f
t
k
in
e
c
t
se
n
s
o
r
a
n
d
it
s e
f
f
e
c
t
,
”
IEE
E
mu
lt
ime
d
i
a
,
v
o
l.
1
9
,
n
o
.
2
,
p
p
.
4
-
1
0
,
2
0
1
2
.
[2
8
]
P
.
C.
U.
M
u
r
il
lo
,
J.
O.
P
.
A
re
n
a
s,
a
n
d
R.
J.
M
o
re
n
o
,
“
Im
p
lem
e
n
tatio
n
o
f
a
Da
ta
A
u
g
m
e
n
tatio
n
A
lg
o
rit
h
m
V
a
li
d
a
ted
b
y
M
e
a
n
s
o
f
th
e
Ac
c
u
ra
c
y
o
f
a
Co
n
v
o
l
u
ti
o
n
a
l
Ne
u
ra
l
Ne
tw
o
rk
,
”
J
o
u
rn
a
l
o
f
En
g
in
e
e
rin
g
a
n
d
Ap
p
li
e
d
S
c
ien
c
e
s
,
v
o
l.
1
2
,
n
o
.
2
0
,
p
p
.
5
3
2
3
-
5
3
3
1
,
2
0
1
7
.
[2
9
]
M
a
th
W
o
rk
s,
“
train
Ca
sc
a
d
e
Ob
jec
tDe
tec
to
r
,”
2
0
2
0
.
[
On
li
n
e
]
A
v
a
il
a
b
le
o
n
:
h
tt
p
s:/
/l
a
.
m
a
th
w
o
rk
s.c
o
m
/h
e
lp
/v
is
io
n
/ref
/t
ra
in
c
a
sc
a
d
e
o
b
jec
td
e
tec
to
r
.
h
tm
l.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
10
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
4
7
1
2
-
4720
4720
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Pa
u
la
Us
e
c
h
e
M
u
r
il
lo
is
a
M
e
c
h
a
tro
n
ics
En
g
in
e
e
r
g
ra
d
u
a
ted
w
it
h
h
o
n
o
rs
i
n
2
0
1
7
f
ro
m
th
e
Nu
e
v
a
G
ra
n
a
d
a
M
il
it
a
ry
U
n
iv
e
rsity
in
Bo
g
o
tá,
Co
lo
m
b
ia,
w
h
e
re
sh
e
c
u
rre
n
tl
y
stu
d
ies
a
n
M
.
S
c
.
i
n
M
e
c
h
a
tr
o
n
ics
E
n
g
in
e
e
rin
g
a
n
d
w
o
rk
s
a
s
a
re
se
a
rc
h
a
ss
istan
t
in
th
e
M
e
c
h
a
tro
n
ics
En
g
in
e
e
rin
g
p
r
o
g
ra
m
.
Ro
b
i
n
s
o
n
J
i
m
é
n
e
z
-
M
o
r
e
n
o
is
a
n
e
lec
tro
n
ic
e
n
g
in
e
e
r
g
ra
d
u
a
ted
f
ro
m
th
e
F
ra
n
c
isc
o
Jo
sé
d
e
Ca
ld
a
s
District
Un
iv
e
rsity
in
2
0
0
2
.
He
re
c
e
iv
e
d
a
n
M
.
S
c
.
i
n
E
n
g
in
e
e
rin
g
f
ro
m
th
e
Na
ti
o
n
a
l
Un
iv
e
rsit
y
o
f
Co
lo
m
b
ia i
n
2
0
1
2
a
n
d
P
h
.
D i
n
En
g
in
e
e
rin
g
a
t
th
e
F
ra
n
c
isc
o
Jo
sé
d
e
Ca
ld
a
s D
istri
c
t
Un
iv
e
rsit
y
in
2
0
1
8
.
His
c
u
rre
n
t
r
e
se
a
rc
h
f
o
c
u
se
s
o
n
th
e
u
se
o
f
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
f
o
r
o
b
jec
t
re
c
o
g
n
it
i
o
n
a
n
d
im
a
g
e
p
ro
c
e
ss
in
g
f
o
r
ro
b
o
ti
c
a
p
p
li
c
a
ti
o
n
s
su
c
h
a
s
h
u
m
a
n
-
m
a
c
h
in
e
in
tera
c
ti
o
n
.
J
a
v
ier
E
d
u
a
r
d
o
M
a
r
tin
e
z
B
a
q
u
e
ro
is
a
n
El
e
c
tr
o
n
ic
En
g
in
e
e
r
g
ra
d
u
a
ted
f
ro
m
Un
iv
e
rsit
y
o
f
th
e
L
lan
o
s
in
2
0
0
2
.
P
o
sg
ra
d
u
a
ted
in
El
e
c
tro
n
ic
In
stru
m
e
n
tatio
n
f
ro
m
S
a
n
to
T
o
m
a
s
Un
iv
e
rsit
y
in
2
0
0
4
a
n
d
M
.S
c
.
in
Ed
u
c
a
ti
v
e
Tec
h
n
o
l
o
g
y
a
t
A
u
to
n
o
m
a
o
f
Bu
c
a
ra
m
a
n
g
a
Un
iv
e
rsit
y
in
2
0
1
3
.
His c
u
rre
n
t
re
se
a
rc
h
f
o
c
u
se
s o
n
In
stru
m
e
n
tatio
n
,
A
u
to
m
a
ti
o
n
a
n
d
C
o
n
tr
o
l.
Evaluation Warning : The document was created with Spire.PDF for Python.