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.
6
,
No
.
6
,
Dec
em
b
er
201
6
,
p
p
.
2
7
5
5
~
2
7
6
5
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
6i
6
.
1
1
8
7
8
2755
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
JE
C
E
R
ea
l Ti
m
e W
eed
Dete
c
tion
u
sing
a
Bo
o
sted C
a
sca
de
o
f
S
i
m
ple
F
ea
tures
Adil T
a
n
no
uche,
K
ha
lid
S
ba
i,
M
ilo
ud
Ra
h
m
o
un
e
,
Ra
chi
d
Ag
o
un
o
un
e,
Abdel
ha
i
Ra
h
m
a
n
i,
Abdela
li Ra
h
m
a
n
i
L
a
b
o
ra
to
ire d
’Et
u
d
e
d
e
s M
a
téria
u
x
Av
a
n
c
é
s
e
t
A
p
p
li
c
a
ti
o
n
s,
F
S
-
ES
T
,
M
o
u
lay
Is
m
a
il
Un
iv
e
rsit
y
,
B
P
1
1
2
0
1
,
Zi
t
o
u
n
e
,
M
e
k
n
e
s,
M
o
ro
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
7
,
2
0
1
6
R
ev
i
s
ed
A
u
g
14
,
2
0
1
6
A
cc
ep
ted
A
u
g
28
,
2
0
1
6
W
e
e
d
d
e
tec
ti
o
n
is
a
c
ru
c
ial
issu
e
in
p
re
c
isio
n
a
g
ricu
lt
u
re
.
I
n
c
o
m
p
u
ter
v
isio
n
,
v
a
riet
y
o
f
tec
h
n
iq
u
e
s
a
re
d
e
v
e
lo
p
e
d
to
d
e
tec
t,
id
e
n
ti
f
y
a
n
d
lo
c
a
t
e
we
e
d
s
in
d
if
fe
re
n
t
c
u
lt
u
re
s.
In
t
h
is
a
rti
c
le,
w
e
p
re
s
e
n
t
a
re
a
l
-
ti
m
e
n
e
w
w
e
e
d
d
e
tec
ti
o
n
m
e
th
o
d
,
th
r
o
u
g
h
a
n
e
m
b
e
d
d
e
d
m
o
n
o
c
u
lar
v
isio
n
.
O
u
r
a
p
p
ro
a
c
h
is
b
a
se
d
o
n
th
e
u
se
o
f
a
c
a
sc
a
d
e
o
f
d
isc
ri
m
i
n
a
ti
v
e
c
las
si
f
iers
f
o
r
m
e
d
b
y
th
e
Ha
a
r
-
li
k
e
f
e
a
tu
re
s.
T
h
e
q
u
a
li
ty
o
f
th
e
re
su
lt
s
d
e
term
in
e
s
th
e
v
a
li
d
it
y
o
f
o
u
r
a
p
p
ro
a
c
h
,
a
n
d
o
p
e
n
s t
h
e
w
a
y
to
n
e
w
h
o
rizo
n
s in
w
e
e
d
d
e
tec
ti
o
n
.
K
ey
w
o
r
d
:
A
d
a
B
o
o
s
t a
lg
o
r
ith
m
A
r
ti
f
icial
v
i
s
io
n
Haar
-
li
k
e
f
ea
t
u
r
es
P
r
ec
is
io
n
ag
r
icu
l
tu
r
e
W
ee
d
d
etec
tio
n
Co
p
y
rig
h
t
©
2
0
1
6
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
:
A
d
il T
an
n
o
u
ch
e,
Dep
ar
t
m
en
t o
f
E
lectr
ical
an
d
C
o
m
p
u
ter
E
n
g
in
ee
r
i
n
g
,
L
E
M2
A
,
Hig
h
Sc
h
o
o
l o
f
T
ec
h
n
o
lo
g
y
,
Me
k
n
e
s
,
+2
1
2
6
5
0
9
6
0
2
0
7
,
Mo
r
o
cc
o
.
E
m
ail:
ta
n
n
o
u
c
h
e@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
W
ee
d
co
n
tr
o
l
is
a
k
e
y
ele
m
e
n
t
in
p
r
ec
is
io
n
ag
r
icu
l
tu
r
e.
C
h
e
m
ical
w
ee
d
co
n
tr
o
l
is
w
id
el
y
p
r
ac
ticed
to
r
ed
u
ce
in
f
ec
t
io
n
r
ates
an
d
i
m
p
r
o
v
e
cr
o
p
s
.
I
n
th
is
r
eg
ar
d
,
r
ec
e
n
t
co
m
p
u
ter
v
is
io
n
r
esear
ch
h
as
s
p
a
w
n
ed
s
ev
er
a
l
ef
f
ec
tiv
e
tec
h
n
iq
u
e
s
f
o
r
th
e
d
etec
tio
n
an
d
/
o
r
class
i
f
icat
io
n
o
f
w
ee
d
s
.
I
n
[
1
]
th
e
a
u
th
o
r
s
h
a
v
e
d
ev
elo
p
ed
a
r
ap
id
s
h
ap
e
d
escr
ip
to
r
f
o
r
th
e
s
elec
tio
n
o
f
w
ee
d
s
.
I
n
[
2
]
th
e
au
th
o
r
s
elab
o
r
ated
a
m
u
ltil
a
y
e
r
p
er
ce
p
tr
o
n
-
b
ased
tech
n
o
lo
g
y
to
r
ea
l
ti
m
e
d
etec
tio
n
o
f
w
ee
d
s
i
n
g
r
o
w
i
n
g
o
n
i
o
n
s
.
I
n
[
3
]
a
h
y
b
r
id
o
f
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
e
(
SVM)
w
it
h
R
ad
ial
b
asi
s
f
u
n
ctio
n
(
R
B
F)
ar
e
u
s
ed
to
id
en
tif
icatio
n
o
f
f
o
u
r
m
aj
o
r
ty
p
es
o
f
cr
o
p
p
est
w
h
ic
h
o
cc
u
r
s
o
n
tef
f
,
w
h
ea
t,
s
o
r
g
h
u
m
,
b
ar
le
y
an
d
m
aize
.
I
n
[
4
]
th
e
au
th
o
r
s
p
r
o
p
o
s
es
a
n
o
v
el
an
g
u
lar
te
x
t
u
r
e
p
atter
n
(
A
T
P
)
ex
tr
ac
tio
n
m
et
h
o
d
f
o
r
c
r
o
p
an
d
w
ee
d
d
is
cr
i
m
in
at
io
n
t
o
o
v
er
co
m
e
th
e
d
e
f
ec
ts
o
f
b
ac
k
g
r
o
u
n
d
co
m
p
le
x
it
y
an
d
v
ar
iatio
n
o
f
ill
u
m
in
a
tio
n
in
th
e
a
g
r
ic
u
lt
u
r
al
s
ce
n
e.
I
n
[
5
]
th
e
au
th
o
r
s
i
n
tr
o
d
u
ce
a
n
e
w
w
ee
d
d
etec
tio
n
s
y
s
te
m
u
s
in
g
f
r
ac
tal
-
b
ased
lo
w
co
s
t c
o
m
m
o
d
it
y
h
ar
d
w
ar
e
R
asp
b
er
r
y
P
i.
I
n
th
i
s
ar
ticle,
w
e
p
r
ese
n
t
a
n
e
w
m
et
h
o
d
f
o
r
d
etec
tin
g
ad
v
e
n
titi
o
u
s
b
o
ar
d
b
y
m
o
n
o
c
u
lar
v
is
io
n
.
T
h
e
s
p
atial
in
f
o
r
m
atio
n
co
n
tai
n
ed
in
th
e
i
m
ag
e
s
f
r
o
m
t
h
e
f
iel
d
is
d
escr
ib
ed
b
y
th
e
r
ec
tan
g
u
lar
f
ilter
s
ca
lle
d
Haar
-
li
k
e
f
ea
t
u
r
es
[
6
]
-
7
]
.
T
h
en
,
th
i
s
d
is
cr
i
m
in
at
iv
e
d
escr
ip
t
o
r
u
tili
ze
s
a
co
n
ce
p
t
o
f
ca
s
ca
d
e
s
i
m
ilar
to
th
at
p
r
o
p
o
s
ed
in
[
8
-
9
]
to
f
i
n
d
w
e
ed
s
in
c
u
lti
v
ated
la
n
d
.
T
h
e
o
r
ig
in
a
lit
y
o
f
o
u
r
ap
p
r
o
ac
h
lies
in
th
e
s
t
u
d
y
a
n
d
ev
alu
a
tio
n
o
f
t
h
is
tech
n
iq
u
e
o
f
ar
ti
f
icial
v
i
s
io
n
,
w
h
ic
h
i
s
w
i
d
ely
k
n
o
w
n
an
d
r
ec
o
g
n
ized
i
n
o
b
j
ec
t
r
ec
o
g
n
itio
n
in
th
e
ar
ea
o
f
w
ee
d
d
etec
tio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Dec
em
b
er
2
0
1
6
: 2
7
5
5
–
2
7
6
5
2756
2.
M
AT
E
RIAL
A
ND
M
E
T
H
O
D
T
h
e
v
is
io
n
s
y
s
te
m
p
r
ese
n
ted
h
er
e
is
i
n
te
n
d
ed
to
ev
e
n
t
u
all
y
b
e
e
m
b
ed
d
ed
o
n
a
g
r
ic
u
lt
u
r
al
m
ac
h
in
er
y
to
o
p
tim
ize
it
s
o
p
er
atio
n
in
th
e
f
ield
(
Fig
u
r
e
1
)
.
T
h
is
co
n
ce
p
t
is
u
s
ed
in
d
i
f
f
er
e
n
t
w
o
r
k
s
[
2
]
,
[
1
0
]
an
d
[
1
1
]
,
its
in
ter
est
lie
s
in
it
s
ea
s
y
w
a
y
o
f
i
m
p
le
m
e
n
tatio
n
an
d
lo
w
co
s
t.
T
h
is
en
tire
w
o
r
k
i
s
b
ased
o
n
a
d
etailed
an
al
y
s
i
s
o
f
i
m
ag
e
s
f
r
o
m
th
e
f
ie
ld
,
w
it
h
ar
ti
f
icial
v
i
s
io
n
to
o
ls
to
e
x
t
r
ac
t
s
p
ec
if
ic
in
f
o
r
m
a
tio
n
to
t
h
e
d
ec
is
io
n
o
n
th
e
lo
ca
tio
n
o
f
w
ee
d
s
.
Fig
u
r
e
1
.
E
x
p
er
i
m
en
tal
s
et
u
p
R
ea
l
-
ti
m
e
w
ee
d
s
d
etec
tio
n
is
r
eq
u
ir
ed
f
o
r
th
is
t
y
p
e
o
f
ap
p
licatio
n
,
in
s
u
ch
a
w
a
y
a
s
t
o
leav
e
a
r
esp
o
n
s
e
ti
m
e
to
th
e
s
p
r
a
y
er
co
n
tr
o
l.
Ho
w
e
v
er
,
ex
h
au
s
ti
v
e
s
ea
r
c
h
o
f
w
ee
d
s
o
n
th
e
co
m
p
lete
i
m
a
g
e
i
s
p
r
o
h
ib
itiv
e
f
o
r
r
ea
l
-
ti
m
e
ap
p
licatio
n
.
T
o
s
o
lv
e
th
is
p
r
o
b
le
m
,
in
t
h
e
liter
at
u
r
e,
m
o
s
t
o
b
j
ec
t
d
etec
tio
n
m
et
h
o
d
s
ar
e
d
iv
id
ed
in
to
t
w
o
s
ta
g
es a
c
co
r
d
in
g
to
atten
tio
n
al
p
r
o
ce
s
s
e
s
[
1
2
]
:
a.
Dev
elo
p
m
e
n
t
o
f
h
y
p
o
t
h
ese
s
:
t
h
e
s
y
s
te
m
p
r
o
v
id
es
a
v
er
y
s
i
m
p
le
a
n
d
q
u
ic
k
w
a
y
o
f
p
o
ten
t
ial
p
o
s
itio
n
s
o
f
o
b
j
ec
ts
to
b
e
d
etec
ted
in
o
r
d
er
to
r
estrict
th
e
s
ea
r
ch
f
ield
.
b.
Ver
if
y
in
g
h
y
p
o
th
e
s
es:
T
h
e
p
r
ev
io
u
s
l
y
d
ev
elo
p
ed
h
y
p
o
t
h
e
s
es
ar
e
v
alid
ated
b
y
u
s
i
n
g
m
o
r
e
co
m
p
le
x
alg
o
r
ith
m
s
.
T
h
e
ap
p
r
o
ac
h
es
u
s
ed
to
v
ali
d
ate
th
ese
a
s
s
u
m
p
t
io
n
s
ar
e
b
ased
o
n
th
e
u
s
e
o
f
p
r
i
m
it
i
v
es
o
r
th
e
ap
p
ea
r
an
ce
o
f
o
b
j
ec
ts
in
q
u
es
ti
o
n
.
T
h
e
p
r
i
m
it
iv
e
m
et
h
o
d
s
d
ef
i
n
e
a
te
m
p
late
f
o
r
ea
c
h
p
ar
ticu
lar
cla
s
s
.
T
h
u
s
,
th
e
v
al
id
atio
n
is
p
er
f
o
r
m
ed
b
y
co
r
r
elatio
n
b
etw
ee
n
t
h
e
m
o
d
el
an
d
th
e
in
p
u
t
i
m
ag
e.
Seco
n
d
ly
,
m
et
h
o
d
s
b
ased
o
n
ap
p
ea
r
an
ce
lear
n
th
e
ch
ar
a
cter
is
tics
r
elatin
g
to
ea
c
h
c
lass
f
r
o
m
a
s
et
o
f
i
m
a
g
e
s
.
I
n
f
ac
t,
a
v
ec
to
r
o
f
lo
ca
l
o
r
g
lo
b
al
d
escr
ip
to
r
s
d
escr
ib
es
ea
ch
i
m
ag
e
u
s
ed
in
lear
n
i
n
g
.
T
h
en
,
lear
n
in
g
a
cla
s
s
i
f
ier
(
Neu
r
al
Net
w
o
r
k
s
,
Su
p
p
o
r
t
Vec
to
r
Ma
ch
in
e
(
S
V
M)
,
B
ay
es
ian
A
c
t,
etc.
)
g
e
n
er
all
y
p
er
m
i
ts
t
h
e
est
i
m
a
tio
n
o
f
a
b
o
u
n
d
ar
y
d
ec
is
io
n
f
o
r
th
e
ass
ig
n
m
en
t
o
f
a
n
e
w
o
b
j
ec
t
to
a
class
o
r
an
o
t
h
er
.
T
h
is
co
m
m
o
n
l
y
k
n
o
w
n
d
i
s
cr
i
m
i
n
ati
v
e
ca
s
ca
d
e
o
f
w
ea
k
clas
s
i
f
ier
s
tec
h
n
iq
u
e
h
as
b
ee
n
s
u
cc
es
s
f
u
ll
y
u
s
ed
i
n
t
h
e
liter
at
u
r
e
f
o
r
th
e
r
ea
l
-
t
i
m
e
d
etec
tio
n
o
f
v
eh
ic
les
[
1
3
-
1
4
]
an
d
p
ed
estri
an
s
[
1
5
-
1
6
]
.
I
n
ad
d
itio
n
to
t
h
ese
clas
s
i
f
icatio
n
m
et
h
o
d
s
,
th
e
au
t
h
o
r
s
in
[
8
]
p
r
o
p
o
s
e
an
o
r
ig
i
n
al
co
n
ce
p
t
atten
tio
n
al
ap
p
lied
to
th
e
f
ac
e
d
etec
tio
n
.
I
ts
p
r
i
n
cip
le
is
b
ased
o
n
t
h
e
u
s
e
o
f
in
cr
ea
s
i
n
g
co
m
p
le
x
it
y
cla
s
s
i
f
ier
s
ca
s
ca
d
e.
T
h
u
s
,
ea
c
h
s
ta
g
e
o
f
t
h
e
ca
s
ca
d
e
ca
n
i
n
cr
ea
s
in
g
l
y
r
e
s
tr
ict,
t
h
e
r
esear
ch
ar
ea
b
y
r
ej
ec
tin
g
a
n
i
m
p
o
r
tan
t
s
et
o
f
ar
ea
s
n
o
t
co
n
tain
i
n
g
f
ac
e
s
.
T
h
is
m
et
h
o
d
u
s
es
t
h
e
s
o
-
ca
lled
Haar
-
li
k
e
f
ea
tu
r
es
a
ls
o
ca
lled
r
ec
tan
g
u
lar
f
ilter
s
[
6
-
7
]
a
n
d
th
e
lear
n
in
g
al
g
o
r
ith
m
A
d
aB
o
o
s
t
[
1
7
]
.
T
h
is
allo
w
s
th
e
m
to
s
elec
t
o
n
ea
c
h
s
tag
e
a
li
m
ited
n
u
m
b
er
o
f
d
escr
ip
to
r
s
.
T
h
e
o
p
er
atio
n
o
f
t
h
e
s
et,
i
n
r
e
al
ti
m
e,
is
en
s
u
r
ed
b
y
th
e
co
m
b
in
ed
u
s
e
o
f
r
ap
id
ca
lcu
latio
n
Haa
r
-
lik
e
f
ea
t
u
r
es,
b
ased
o
n
th
e
co
n
ce
p
t
o
f
t
h
e
i
n
teg
r
al
i
m
a
g
e,
a
n
d
th
e
d
is
cr
i
m
i
n
ati
v
e
ca
s
ca
d
e.
I
n
th
is
ar
tic
le
w
e
co
n
s
id
er
t
h
e
r
ec
tan
g
u
lar
f
ilter
s
(
Haa
r
-
li
k
e
f
ea
t
u
r
es).
T
h
ese
f
il
ter
s
in
itial
l
y
i
n
tr
o
d
u
ce
d
b
y
[
6
-
7
]
t
o
d
etec
t
p
ed
estrian
s
an
d
v
e
h
i
cles
ar
e
b
ased
o
n
a
d
ec
o
m
p
o
s
itio
n
o
f
th
e
i
m
ag
e
w
it
h
Haa
r
w
a
v
elets.
T
h
en
,
t
h
eir
u
s
e
h
as
i
m
p
r
o
v
ed
s
o
m
u
c
h
i
n
d
i
f
f
er
en
t
s
t
u
d
ies
[
8
-
9
]
,
it
is
n
o
lo
n
g
er
s
tr
ictl
y
co
r
r
esp
o
n
d
to
th
e
w
a
v
elet
t
h
eo
r
y
.
T
h
e
y
ar
e
t
h
en
ap
p
o
in
te
d
Haa
r
-
li
k
e
f
ea
tu
r
e
s
.
C
u
r
r
en
t
l
y
,
t
h
ese
f
il
ter
s
ar
e
co
m
m
o
n
l
y
an
d
w
id
el
y
u
s
ed
in
p
atter
n
r
ec
o
g
n
it
io
n
f
o
r
d
etec
tio
n
o
r
o
b
j
ec
t
r
ec
o
g
n
itio
n
.
O
u
r
ap
p
r
o
ac
h
p
r
o
p
o
s
es
to
in
teg
r
ate
d
is
cr
i
m
i
n
ati
v
e
cla
s
s
i
f
ier
lev
el
d
escr
ip
to
r
s
Haa
r
.
So
w
e
s
t
u
d
ied
th
e
b
eh
a
v
io
r
o
f
d
etec
to
r
s
,
s
i
m
ilar
to
th
e
o
n
e
p
r
o
p
o
s
ed
in
[
8
]
an
d
co
n
s
tr
u
cted
w
it
h
Haa
r
d
escr
ip
to
r
s
ass
o
ciate
d
w
i
th
d
is
cr
i
m
i
n
an
t
clas
s
i
f
ier
s
[
1
4
]
.
T
h
e
ch
o
ice
o
f
d
escr
ip
to
r
s
ca
n
b
e
d
is
cu
s
s
ed
b
u
t
t
h
e
m
ain
o
b
j
ec
tiv
e
o
f
th
is
s
tu
d
y
is
to
s
h
o
w
t
h
e
f
ea
s
ib
il
it
y
o
f
o
u
r
th
eo
r
y
.
I
n
t
h
e
r
e
m
a
in
d
er
o
f
th
i
s
ar
ticle,
w
e
w
ill
d
e
v
el
o
p
o
u
r
id
ea
an
d
p
r
esen
t
m
o
r
e
p
r
ec
is
el
y
h
o
w
w
e
d
ef
in
ed
d
escr
ip
to
r
s
an
d
th
e
ass
o
ciate
d
class
if
ier
u
s
ed
i
n
A
d
aB
o
o
s
t
alg
o
r
ith
m
.
T
h
en
we
w
il
l
ex
p
lai
n
o
u
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
R
ea
l Time
W
ee
d
Dete
ct
io
n
Usi
n
g
a
B
o
o
s
ted
C
a
s
ca
d
e
o
f
S
imp
le
F
ea
tu
r
es
(
A
d
il Ta
n
n
o
u
ch
e
)
2757
i
m
p
le
m
en
ta
tio
n
w
it
h
o
r
w
it
h
o
u
t
a
ca
s
ca
d
e,
th
en
p
r
esen
t
s
ev
er
al
r
esu
lt
s
o
n
t
h
e
b
asis
o
f
e
m
b
ed
d
ed
im
a
g
es.
T
h
ese
r
esu
lts
w
i
ll a
llo
w
u
s
to
an
al
y
ze
t
h
e
r
ea
lized
s
y
s
te
m
an
d
co
n
clu
d
e
o
u
r
ap
p
r
o
ac
h
.
2
.
1
.
Descript
o
r
P
o
s
itiv
e
ex
a
m
p
les
(
w
ee
d
s
)
o
r
n
eg
ati
v
e
(
cr
o
p
an
d
o
th
er
)
,
ar
e
d
is
tr
ib
u
ted
in
a
N
-
d
i
m
e
n
s
io
n
al
s
p
ac
e
w
h
ic
h
d
ep
en
d
s
o
n
th
e
p
ar
a
m
e
ter
s
s
elec
ted
to
r
ep
r
esen
t
t
h
e
i
n
f
o
r
m
atio
n
.
I
n
t
h
e
o
r
ig
i
n
al
s
p
ac
e
(
d
ef
in
ed
b
y
t
h
e
i
m
a
g
e
o
f
g
r
a
y
lev
el
s
)
,
t
h
ese
e
x
a
m
p
le
s
ca
n
b
e
m
i
x
ed
.
T
h
u
s
,
th
e
d
is
tin
c
tio
n
is
m
ad
e
b
y
s
el
ec
tin
g
t
h
e
s
p
ac
e
o
f
r
ep
r
esen
tatio
n
a
n
d
th
e
r
ig
h
t c
l
ass
i
f
ier
.
I
n
th
i
s
w
o
r
k
,
Haa
r
-
li
k
e
f
ea
t
u
r
es
ar
e
u
s
ed
to
d
ef
in
e
a
d
is
cr
im
i
n
ati
n
g
m
o
d
el
t
h
at
s
ep
ar
ate
s
b
et
w
ee
n
r
eg
io
n
s
o
f
w
ee
d
s
g
r
o
w
n
in
t
wo
class
es
w
it
h
a
b
o
u
n
d
ar
y
(
h
y
p
e
r
p
lan
e)
.
T
est
ex
a
m
p
les
ar
e
class
i
f
ied
ac
co
r
d
in
g
to
th
eir
p
o
s
itio
n
i
n
s
p
ac
e
with
r
esp
ec
t
to
th
e
s
ep
ar
atin
g
h
y
p
er
p
lan
e.
T
h
is
s
p
ac
e
ca
n
ex
tr
ac
t
h
i
g
h
-
lev
el
in
f
o
r
m
atio
n
(
o
u
t
lin
e,
te
x
t
u
r
e,
etc)
f
r
o
m
i
m
a
g
e
o
f
g
r
a
y
le
v
els.
I
n
t
h
is
s
ec
tio
n
,
w
e
d
es
cr
ib
e
th
e
s
p
ac
e
o
f
p
ar
am
eter
s
s
elec
ted
to
ca
r
r
y
o
u
t th
e
d
etec
tio
n
o
f
w
ee
d
s
.
Haar
-
li
k
e
f
ea
tu
r
es
o
r
r
ec
tan
g
u
lar
f
ilter
s
,
p
r
o
v
id
e
i
n
f
o
r
m
ati
o
n
ab
o
u
t
t
h
e
d
is
tr
ib
u
t
io
n
o
f
g
r
a
y
le
v
el
s
b
et
w
ee
n
t
w
o
n
ei
g
h
b
o
r
in
g
r
e
g
io
n
s
i
n
t
h
e
i
m
a
g
e.
Fi
g
u
r
e
2
s
h
o
w
s
all
Haa
r
-
li
k
e
f
ilter
s
u
s
e
d
in
o
u
r
w
o
r
k
.
T
h
e
ch
o
s
en
f
i
lter
s
ar
e
t
h
o
s
e
t
w
o
a
n
d
th
r
ee
r
ec
tan
g
les.
T
o
g
et
t
h
e
v
alu
e
(
o
u
tp
u
t)
o
f
a
f
ilter
ap
p
lied
to
a
r
eg
io
n
o
f
t
h
e
i
m
a
g
e,
t
h
e
s
u
m
o
f
t
h
e
p
i
x
el
s
in
th
e
w
h
ite
r
ec
ta
n
g
le
i
s
s
u
b
tr
ac
ted
f
r
o
m
t
h
e
s
u
m
o
f
th
e
p
ix
els
i
n
t
h
e
b
l
u
e
r
ec
tan
g
le
(
m
u
ltip
lied
b
y
a
c
o
ef
f
icie
n
t,
in
t
h
e
ca
s
e
o
f
th
e
f
ilter
th
r
ee
r
id
g
es).
Fig
u
r
e
2
.
Haar
-
L
i
k
e
Fi
lter
A
s
s
e
m
b
l
y
U
s
ed
Fig
u
r
e
3
.
C
o
n
ce
p
t o
f
I
n
te
g
r
al
I
m
ag
e
: I
m
a
g
e
in
teg
r
al
(
1
)
,
C
alcu
lates a
n
y
Sq
u
ar
e
A
r
ea
f
r
o
m
F
o
u
r
R
ef
er
e
n
ce
s
A
,
B
,
C
an
d
D
(
2
)
I
n
[
8
]
,
th
e
au
t
h
o
r
s
in
tr
o
d
u
ce
d
th
e
co
n
ce
p
t
o
f
t
h
e
i
n
t
eg
r
al
i
m
ag
e.
T
h
is
i
s
an
in
ter
m
ed
iar
y
r
ep
r
esen
tatio
n
o
f
t
h
e
in
p
u
t
i
m
ag
e
t
h
at
al
lo
w
s
u
s
to
r
ed
u
ce
t
h
e
co
m
p
u
tat
io
n
t
i
m
e
a
s
s
o
ciate
d
w
it
h
th
e
ap
p
licatio
n
o
f
th
ese
f
ilter
s
(
es
p
ec
iall
y
w
h
e
n
th
e
y
o
v
er
lap
)
.
T
h
e
v
alu
e
o
f
th
e
in
teg
r
al
i
m
ag
e,
n
a
m
ed
ii
(
x
,
y
)
at
th
e
p
o
s
itio
n
(
x
,
y
)
is
t
h
e
s
u
m
o
f
all
t
h
e
p
ix
el
v
a
lu
e
s
o
f
t
h
e
o
r
ig
in
a
l
i
m
ag
e,
ab
o
v
e
a
n
d
t
o
th
e
le
f
t
o
f
(
x
,
y
)
(
Fig
u
r
e
3
(
1
)
)
,
ca
lcu
latio
n
is
p
er
f
o
r
m
ed
ac
co
r
d
in
g
to
th
i
s
f
o
r
m
u
la:
(
)
∑
(
)
(
1
)
w
it
h
i (
x
,
y
)
r
ep
r
esen
ts
t
h
e
l
u
m
i
n
an
ce
o
f
t
h
e
p
i
x
el
i
n
th
e
o
r
i
g
in
a
l i
m
a
g
e
at
t
h
e
p
o
s
it
io
n
(
x
,
y
)
.
T
h
u
s
,
t
h
e
s
u
m
o
f
th
e
p
i
x
els
o
f
an
y
r
ec
ta
n
g
u
lar
ar
ea
o
f
an
i
m
a
g
e
ca
n
b
e
e
v
alu
a
ted
f
r
o
m
f
o
u
r
p
o
i
n
ts
o
f
r
ef
er
en
ce
s
i
n
th
e
co
r
r
esp
o
n
d
in
g
i
n
te
g
r
al
i
m
ag
e.
As
a
n
e
x
a
m
p
le,
i
n
Fig
u
r
e
3
(
2
)
,
w
h
er
e
t
h
e
s
u
m
o
f
t
h
e
v
al
u
es
o
f
th
e
b
l
u
e
ar
ea
zo
n
e
(
A
,
B
,
C
,
D)
is
g
i
v
e
n
b
y
t
h
e
f
o
llo
w
i
n
g
f
o
r
m
u
la:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Dec
em
b
er
2
0
1
6
: 2
7
5
5
–
2
7
6
5
2758
(
)
(
)
(
)
(
)
(
)
(
2
)
T
h
er
ef
o
r
e,
an
d
f
o
r
a
t
w
o
-
r
ec
tan
g
le
f
ilter
,
t
h
e
d
if
f
er
e
n
ce
b
et
w
ee
n
th
e
t
w
o
ad
j
ac
en
t
r
ec
tan
g
le
s
is
obt
ain
ed
v
ia
s
i
x
p
o
in
ts
o
f
r
ef
e
r
en
ce
s
i
n
th
e
i
n
te
g
r
al
i
m
a
g
e.
I
n
t
h
e
ca
s
e
o
f
a
t
h
r
ee
-
r
ec
ta
n
g
le
f
i
lter
,
ei
g
h
t
p
o
in
t
s
o
f
r
ef
er
en
ce
s
ar
e
u
s
ed
.
Fig
u
r
e
4
ill
u
s
tr
ates
a
n
e
x
a
m
p
l
e
o
f
f
ilter
i
n
g
a
p
ict
u
r
e
o
f
o
u
r
d
atab
ase
w
i
th
a
v
er
tical
r
ec
ta
n
g
u
lar
f
i
lter
(2
2
)
p
ix
els
w
id
e.
T
h
e
f
i
g
u
r
e
s
h
o
w
s
th
a
t
t
h
e
s
e
lecte
d
f
ilter
en
h
a
n
ce
s
t
h
e
v
er
tica
l
ed
g
e
s
(
b
r
ig
h
t
p
ix
el
s
)
o
f
th
e
i
m
a
g
e
(
Fig
u
r
e
4
(
2
)
)
.
W
e
n
o
te
th
at
d
o
u
b
lin
g
t
h
e
s
ize
o
f
th
e
f
ilter
s
,
w
e
f
ilter
t
h
e
d
etails
o
f
th
e
o
r
ig
i
n
al
f
ig
u
r
e,
w
h
ile
r
etai
n
i
n
g
t
h
e
m
a
in
o
u
tli
n
es.
Fig
u
r
e
4
.
A
p
p
licatio
n
o
f
v
er
tic
al
Haa
r
-
li
k
e
f
ea
t
u
r
es
f
ilte
r
s
(
2
)
o
n
an
o
r
ig
i
n
al
i
m
ag
e
(
1
)
E
ac
h
d
escr
ip
to
r
(
j)
is
d
ef
in
ed
as
a
f
u
n
ctio
n
f
j
(x
i
,
y
i
,
E
j
,
C
j
)
,
w
h
er
e
(
x
i
,
y
i
)
th
e
p
o
s
it
io
n
in
th
e
th
u
m
b
n
ail,
C
j
i
s
t
h
e
t
y
p
e
Haa
r
-
lik
e
f
ea
t
u
r
es
(
Fi
g
u
r
e
2
)
an
d
E
j
is
th
e
s
ca
le
(
w
e
u
s
e
5
i
n
to
tal:
1
2
,
2
4
,
4
8
,
8
1
6
an
d
1
6
3
2
)
.
Fil
ter
s
r
o
a
m
t
h
e
t
h
u
m
b
n
ails
i
n
s
tep
s
o
f
o
n
e
p
ix
e
l.
Fo
r
a
t
h
u
m
b
n
a
il
i
m
a
g
e
o
f
(
3
2
x
3
2
)
p
ix
els
s
ize,
th
e
Haa
r
s
p
ac
e
p
ar
a
m
eter
s
is
t
h
u
s
o
f
8
1
5
1
d
im
e
n
s
io
n
.
2
.
2
.
Cla
s
s
if
ier
T
h
e
d
im
e
n
s
io
n
al
s
ize
o
f
t
h
e
p
ar
a
m
eter
s
p
ac
es
is
m
u
c
h
g
r
ea
t
er
th
an
t
h
e
n
u
m
b
er
o
f
p
ix
e
ls
o
f
t
h
e
in
p
u
t
i
m
a
g
e.
Usi
n
g
al
l
th
i
s
to
g
et
h
er
to
ac
h
iev
e
t
h
e
clas
s
i
f
icatio
n
i
s
an
in
ad
eq
u
ate
c
h
o
ice
f
r
o
m
th
e
s
tan
d
p
o
in
t
o
f
t
h
e
ex
ec
u
t
io
n
ti
m
e
an
d
r
o
b
u
s
t
n
es
s
,
s
in
ce
s
o
m
e
o
f
t
h
ese
p
ar
a
m
ete
r
s
d
o
n
o
t c
o
n
tain
r
elev
a
n
t i
n
f
o
r
m
atio
n
(
n
o
i
s
e)
.
T
h
e
d
o
p
in
g
alg
o
r
it
h
m
k
n
o
w
n
A
d
ab
o
o
s
t
[
1
7
]
allo
w
ed
th
e
i
m
p
r
o
v
ed
p
er
f
o
r
m
a
n
ce
o
f
s
ev
er
a
l
class
i
f
icatio
n
an
d
d
etec
tio
n
s
y
s
te
m
s
.
I
t
g
i
v
e
s
s
p
ec
if
ic
h
y
p
o
th
eses
co
m
b
i
n
in
g
s
e
v
er
al
w
ea
k
clas
s
if
icatio
n
f
u
n
ctio
n
s
w
i
th
m
o
d
er
ate
ac
cu
r
ac
y
in
g
e
n
er
al.
T
h
is
is
a
n
it
er
ativ
e
al
g
o
r
ith
m
t
h
at
s
ea
r
c
h
e
s
in
t
h
e
d
escr
ip
to
r
v
ec
to
r
s
p
ac
e,
th
e
m
o
s
t
d
is
c
r
i
m
i
n
ati
n
g
w
ea
k
class
i
f
icat
i
o
n
f
u
n
ctio
n
s
to
co
m
b
in
e
th
e
m
in
to
a
s
tr
o
n
g
class
i
f
icatio
n
f
u
n
ctio
n
ac
co
r
d
in
g
to
t
h
is
f
o
r
m
u
la:
*
∑
∑
(
3
)
w
h
er
e
G
an
d
g
ar
e
th
e
s
tr
o
n
g
an
d
w
ea
k
class
i
f
icat
io
n
f
u
n
c
ti
o
n
s
r
esp
ec
tiv
el
y
.
α
is
a
w
ei
g
h
tin
g
f
ac
to
r
f
o
r
ea
ch
w
ea
k
class
if
ier
g
,
S
b
ein
g
th
e
o
v
er
all
class
if
icatio
n
o
f
t
h
e
f
u
n
ct
io
n
th
r
e
s
h
o
ld
G.
Dif
f
er
en
t
v
a
r
iatio
n
s
o
f
t
h
e
d
o
p
in
g
alg
o
r
ith
m
w
er
e
d
ev
el
o
p
ed
:
Dis
cr
ete
A
d
aB
o
o
s
t
[
8
]
R
ea
l
A
d
aB
o
o
s
t
[
1
8
]
Gen
tle
A
d
aB
o
o
s
t,
etc.
a
r
ec
en
t
p
er
f
o
r
m
an
ce
o
f
t
h
ese
a
lter
n
ati
v
es
an
a
l
y
s
is
is
p
r
ese
n
t
ed
in
[
1
9
]
.
W
e
u
s
e
h
er
e
th
e
f
ir
s
t
d
ef
in
ed
b
y
th
e
f
o
llo
w
in
g
p
s
e
u
d
o
alg
o
r
ith
m
:
1
.
Dis
cr
ete
A
d
aB
o
o
s
t a
lg
o
r
ith
m
1.
L
et
N
e
x
a
m
p
le
s
(
)
(
)
*
+
2.
I
n
itialize
*
+
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
R
ea
l Time
W
ee
d
Dete
ct
io
n
Usi
n
g
a
B
o
o
s
ted
C
a
s
ca
d
e
o
f
S
imp
le
F
ea
tu
r
es
(
A
d
il Ta
n
n
o
u
ch
e
)
2759
3.
Fo
r
*
+
Fo
r
ea
ch
d
escr
ip
to
r
(
j
)
,
tr
ain
a
class
i
f
ier
(
G
i
)
.
T
h
e
er
r
o
r
is
g
iv
en
b
y
:
∑
|
(
)
|
C
h
o
o
s
e
(
G
t
)
class
if
ier
h
av
in
g
t
h
e
er
r
o
r
(
t
)
th
e
s
m
aller
;
Up
d
ate
th
e
w
ei
g
h
ts
:
*
(
)
4.
Ou
tp
u
t:
∑
∑
(
)
T
o
u
s
e
it,
w
e
m
u
s
t
n
o
w
d
e
f
in
e
th
e
w
ea
k
clas
s
i
f
icatio
n
f
u
n
c
tio
n
s
f
o
r
t
h
e
t
y
p
e
o
f
d
escr
ip
to
r
s
ch
o
s
e
n
.
W
e
d
ef
in
e
lo
w
clas
s
i
f
icatio
n
f
u
n
ctio
n
ass
o
ciate
d
w
it
h
th
e
d
escr
ip
to
r
(
j)
,
as
a
b
in
ar
y
r
es
p
o
n
s
e
g
i
v
en
b
y
th
e
f
o
llo
w
in
g
eq
u
at
io
n
:
*
(
4
)
w
h
er
e
(
f
j
)
g
i
v
es
th
e
v
al
u
e
o
f
th
e
d
escr
ip
to
r
,
(
θ
j
)
is
th
e
th
r
es
h
o
ld
an
d
(
p
j
)
is
th
e
p
ar
ity
.
Fo
r
ea
ch
d
escr
ip
to
r
(
j)
,
A
d
aB
o
o
s
t
d
ete
r
m
i
n
es
t
h
e
t
h
r
esh
o
ld
o
p
ti
m
al
(
θ
j
)
m
in
i
m
iz
i
n
g
t
h
e
n
u
m
b
er
o
f
m
is
c
lass
if
i
ed
ex
a
m
p
les
o
f
t
h
e
lear
n
in
g
b
ase
(
p
o
s
itiv
e
a
n
d
n
e
g
ati
v
e)
.
2
.
3
.
I
m
ple
m
ent
a
t
io
n
I
n
th
i
s
s
ec
tio
n
,
w
e
d
escr
ib
e
t
h
e
co
n
ten
t o
f
t
h
e
i
m
ag
e
d
atab
as
es u
s
ed
f
o
r
tr
ain
i
n
g
an
d
te
s
ti
n
g
.
T
h
en
w
e
an
al
y
ze
i
n
m
o
r
e
d
etail
s
t
h
e
l
ea
r
n
in
g
o
f
t
h
e
s
elec
ted
d
etec
to
r
.
I
n
u
s
e,
th
e
r
e
s
ea
r
ch
ar
ea
is
r
estricte
d
i
n
t
h
e
i
m
a
g
e
u
s
i
n
g
a
p
r
io
r
k
n
o
w
le
d
g
e
o
f
t
h
e
a
g
r
icu
lt
u
r
al
s
ce
n
e
.
T
h
e
i
m
ag
e
is
tr
a
v
er
s
ed
b
y
a
s
lid
in
g
w
i
n
d
o
w
ev
alu
a
ted
b
y
t
h
e
d
etec
to
r
to
th
e
w
o
r
k
i
n
g
r
eso
lu
tio
n
s
et
to
(
3
2
3
2
)
p
ix
els.
2
.
3
.
1
.
Da
t
a
ba
s
e
T
h
e
d
atab
ase
u
s
ed
is
d
esig
n
ed
f
r
o
m
r
ea
l
i
m
a
g
es
ca
p
t
u
r
ed
m
an
u
al
l
y
a
n
d
co
llected
o
n
th
e
n
et
o
r
f
r
o
m
si
m
u
lat
io
n
ag
r
ic
u
lt
u
r
al
s
ce
n
es
[
2
0
].
I
t
co
n
s
is
ts
o
f
o
v
er
5
0
0
th
u
m
b
n
ails
o
f
d
i
f
f
er
en
t
cr
o
p
s
(
co
r
n
,
w
h
ea
t,
b
ea
n
s
,
p
o
tato
,
b
ea
n
,
etc.
)
,
h
ea
lth
y
a
n
d
in
f
ec
ted
b
y
w
ee
d
s
.
T
h
ese
th
u
m
b
n
a
ils
ta
k
en
at
d
i
f
f
er
en
t
ti
m
e
s
o
f
d
a
y
ar
e
r
an
d
o
m
l
y
tr
an
s
f
o
r
m
ed
b
y
r
o
tatio
n
(
π
)
,
b
y
tr
an
s
latio
n
an
d
s
ca
le
f
ac
to
r
to
in
cr
ea
s
e
t
h
e
s
i
ze
o
f
t
h
e
d
atab
ase.
T
h
en
,
th
i
s
d
atab
ase
w
as
s
o
r
ted
an
d
lab
eled
b
y
h
an
d
i
n
b
o
th
p
o
s
itiv
e
a
n
d
n
e
g
ati
v
e
clas
s
es
r
esp
ec
tiv
el
y
i
n
th
e
p
r
esen
ce
o
r
ab
s
en
ce
o
f
w
ee
d
s
in
th
u
m
b
n
ai
ls
(
Fi
g
u
r
e
5
)
.
Fig
u
r
e
5
.
E
x
a
m
p
le
o
f
p
o
s
iti
v
e
th
u
m
b
n
ails
(
B
x
)
an
d
n
eg
at
iv
e
(
Ax
)
u
s
ed
i
n
o
u
r
s
t
u
d
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Dec
em
b
er
2
0
1
6
: 2
7
5
5
–
2
7
6
5
2760
Sin
ce
th
e
s
m
alle
s
t
d
e
f
i
n
ed
le
ar
n
in
g
s
ca
le
i
s
a
th
u
m
b
n
ai
l
r
e
s
o
lu
tio
n
o
f
(
3
2
3
2
)
p
ix
els,
we
co
n
s
id
er
th
is
s
ize
as
t
h
e
m
in
i
m
u
m
s
ize
d
etec
tio
n
o
f
o
b
j
ec
ts
in
th
e
i
m
a
g
e.
T
h
is
s
ize
co
r
r
esp
o
n
d
s
to
a
u
n
i
tar
y
s
u
b
-
ar
ea
o
f
(
1
m
2
)
i
n
t
h
e
s
ce
n
e
v
ie
w
ed
th
r
o
u
g
h
a
s
ta
n
d
ar
d
R
GB
ca
m
er
a
(
6
4
0
4
8
0
)
r
eso
lu
tio
n
,
o
n
b
o
ar
d
th
e
ca
r
r
ier
v
eh
icle
(
Fig
u
r
e
1
)
.
2
.
3
.
2
.
Det
ec
t
o
r
A
s
i
m
p
le
d
etec
to
r
co
m
p
r
is
es
a
s
tr
o
n
g
class
i
f
ier
,
b
u
ilt
with
a
T
n
u
m
b
er
o
f
d
escr
ip
to
r
s
(
w
ea
k
class
i
f
ier
s
)
.
Fo
u
r
d
etec
to
r
s
w
e
r
e
cr
ea
ted
f
o
r
d
if
f
er
en
t t
y
p
e
s
o
f
Haa
r
-
l
ik
e
f
ea
t
u
r
es c
o
n
s
id
er
ed
(
Fig
u
r
e
2
)
.
T
o
ass
ess
th
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
lear
n
i
n
g
m
e
th
o
d
,
w
e
w
ill
p
er
f
o
r
m
a
cr
o
s
s
-
v
a
lid
atio
n
p
r
o
ce
d
u
r
e.
W
e
tr
ain
th
e
d
etec
to
r
s
o
n
d
if
f
er
en
t
lear
n
in
g
s
a
m
p
le
s
:
th
e
d
is
tr
ib
u
tio
n
o
f
p
o
s
itiv
e
a
n
d
n
eg
at
iv
e
s
a
m
p
les
o
n
lear
n
i
n
g
an
d
v
alid
atio
n
d
ata
is
d
o
n
e
r
a
n
d
o
m
l
y
.
T
h
e
r
ate
o
f
co
r
r
ec
t
d
etec
tio
n
s
(
T
r
u
e
P
o
s
itiv
es
(
T
P
)
)
is
d
ef
in
ed
a
s
th
e
av
er
ag
e
o
f
t
h
e
T
P
o
f
th
e
f
o
u
r
d
etec
to
r
s
.
Si
m
ilar
l
y
,
w
e
u
s
e
t
h
e
av
er
ag
e
o
f
f
alse a
lar
m
s
(
Fal
s
e
P
o
s
itiv
e
(
FP
)
)
.
T
h
en
,
w
e
i
m
p
le
m
en
ted
a
ca
s
c
ad
in
g
ar
ch
itect
u
r
e
[
8
]
co
n
s
is
ti
n
g
o
f
th
e
s
u
cc
e
s
s
io
n
o
f
s
tr
o
n
g
class
i
f
ier
s
.
E
ac
h
s
tr
o
n
g
c
lass
if
ier
is
d
r
i
v
en
b
y
u
s
in
g
th
e
A
d
aB
o
o
s
t
al
g
o
r
ith
m
.
I
n
s
tead
o
f
s
to
p
p
in
g
th
e
iter
ati
v
e
lo
o
p
o
f
A
d
aB
o
o
s
t
f
o
r
m
ax
i
m
u
m
n
u
m
b
er
o
f
d
escr
ip
to
r
s
T
,
w
e
s
et
t
w
o
p
er
f
o
r
m
a
n
ce
th
r
es
h
o
ld
s
:
a
m
in
i
m
u
m
r
ate
o
f
co
r
r
ec
t d
etec
tio
n
s
T
P
m
in
an
d
a
m
ax
i
m
u
m
ac
ce
p
tab
le
r
ate
o
f
f
alse a
lar
m
s
FP
m
ax
.
T
h
e
n
e
g
ati
v
e
b
ase
u
s
ed
to
d
r
i
v
e
t
h
e
s
tr
o
n
g
cla
s
s
i
f
ier
o
n
ac
t
u
al
s
tag
e
i
s
f
o
r
m
ed
o
f
n
e
g
ati
v
e
ex
a
m
p
le
s
th
at
w
er
e
m
is
cla
s
s
i
f
ied
(
i.e
.
co
n
s
id
er
ed
as
w
ee
d
s
)
in
p
r
ev
io
u
s
s
tag
e
s
.
T
h
u
s
,
w
e
h
a
v
e
d
ef
i
n
ed
f
o
u
r
cr
i
ter
ia
to
s
to
p
th
e
lear
n
in
g
o
f
t
h
e
ca
s
ca
d
e:
a.
T
h
e
alg
o
r
ith
m
ac
h
ie
v
ed
th
e
m
ax
i
m
u
m
n
u
m
b
er
o
f
iter
atio
n
s
f
o
r
t
h
e
lear
n
in
g
o
f
t
h
e
s
t
r
o
n
g
cla
s
s
i
f
ier
w
it
h
o
u
t
r
ea
ch
i
n
g
t
h
e
m
ax
i
m
u
m
f
a
ls
e
alar
m
r
ate
o
r
th
e
m
i
n
i
m
u
m
r
ate
o
f
co
r
r
ec
t
d
etec
tio
n
s
.
W
e
s
tate
i
n
o
u
r
r
esu
lt
s
th
at
t
h
er
e
w
as
n
o
n
-
co
n
v
er
g
e
n
ce
o
f
A
d
aB
o
o
s
t in
t
h
e
last
s
tag
e
(
n
o
ted
No
.
C
o
n
v
)
.
b.
T
h
e
ca
s
ca
d
e
r
ea
ch
es
a
f
alse
p
o
s
itiv
e
r
ate
(
FP
)
o
v
er
all
b
elo
w
a
g
iv
e
n
r
ate,
d
ef
i
n
ed
b
y
FP
m
ax
=
0
.
0
5
(
n
o
ted
FP
m
ax
).
c.
I
t
w
as
n
o
t p
o
s
s
ib
le
to
f
i
n
d
a
s
u
f
f
icien
t n
u
m
b
er
o
f
n
eg
a
tiv
e
s
a
m
p
les (
n
o
ted
N.
Sa
m
p
le)
.
d.
W
e
ar
r
iv
ed
to
a
f
ix
ed
n
u
m
b
er
o
f
m
a
x
i
m
u
m
s
tag
e
s
o
f
2
0
(
n
o
ted
E
Max
).
T
h
r
ee
d
if
f
er
en
t
v
er
s
io
n
s
o
f
th
e
ca
s
ca
d
e
ar
e
o
b
tain
ed
b
y
v
ar
y
i
n
g
t
h
e
a
m
o
u
n
t
o
f
n
e
g
ati
v
e
s
a
m
p
les
u
s
ed
f
o
r
th
e
lear
n
i
n
g
: 1
0
0
,
2
0
0
an
d
3
0
0
.
E
v
en
t
u
all
y
,
to
a
v
o
id
t
h
e
n
o
n
-
co
n
v
er
g
e
n
ce
o
f
a
ca
s
ca
d
e
(
w
h
ich
h
ap
p
en
s
f
r
eq
u
e
n
tl
y
w
it
h
A
d
aB
o
o
s
t)
,
w
e
w
ill
m
o
d
if
y
t
h
e
lear
n
i
n
g
cr
iter
io
n
o
f
th
e
clas
s
if
icatio
n
f
u
n
ctio
n
.
if
a
f
ix
ed
l
i
m
itatio
n
att
ac
h
ed
to
ea
ch
s
ta
g
e
o
n
th
e
s
elec
ted
n
u
m
b
er
o
f
d
es
cr
ip
to
r
s
is
ac
h
ie
v
ed
w
i
th
o
u
t
r
e
ac
h
in
g
a
co
n
v
er
g
e
n
ce
(
at
VP
min
an
d
FP
m
ax
le
v
el)
,
th
en
A
d
aB
o
o
s
t
iter
atio
n
is
s
to
p
p
ed
an
d
th
e
class
i
f
icat
io
n
f
u
n
ctio
n
is
r
etai
n
ed
in
t
h
e
s
tate,
th
en
w
e
m
o
v
e
o
n
to
lear
n
in
g
th
e
f
u
n
ctio
n
o
f
th
e
n
ex
t
s
tag
e.
T
h
is
ap
p
r
o
ac
h
d
o
es
n
o
t
b
etr
ay
th
e
f
u
n
d
a
m
e
n
tal
co
n
ce
p
t
o
f
th
e
ca
s
ca
d
e,
s
in
ce
its
p
u
r
p
o
s
e
is
r
etain
ed
,
al
w
a
y
s
ac
ti
n
g
to
eli
m
i
n
ate
a
s
u
b
s
ta
n
tial
p
o
r
tio
n
o
f
n
eg
a
tiv
e
s
a
m
p
le
s
(
n
o
t
co
n
tain
in
g
w
ee
d
)
w
h
ile
k
ee
p
in
g
th
e
p
o
s
iti
v
e
s
a
m
p
l
es
(
co
n
tain
i
n
g
w
ee
d
s
)
.
T
o
s
et
th
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
to
ea
ch
s
tag
e
o
f
th
e
ca
s
ca
d
e,
w
e
ca
n
u
s
e
an
i
n
cr
ea
s
in
g
f
u
n
ctio
n
at
o
u
r
co
n
v
en
ie
n
ce
.
Her
e
w
e
h
a
v
e
ch
o
s
en
to
f
o
llo
w
a
n
ex
p
o
n
en
ti
al
la
w
.
2.
4
.
Rea
l
-
T
i
m
e
P
ro
ce
s
s
W
e
h
av
e
ad
ap
ted
th
e
p
r
o
p
o
s
e
d
m
et
h
o
d
f
o
r
r
ea
l
-
ti
m
e
ap
p
lic
atio
n
.
T
h
e
s
y
s
te
m
co
n
s
is
t
s
o
f
an
o
n
b
o
ar
d
m
o
n
o
cu
lar
ca
m
er
a
o
n
th
e
tr
ac
t
o
r
m
o
v
in
g
i
n
t
h
e
d
ir
ec
tio
n
o
f
s
o
w
i
n
g
li
n
e
s
(
Fi
g
u
r
e
1
)
.
W
e
w
i
s
h
to
d
etec
t
w
ee
d
s
lo
ca
ted
in
f
r
o
n
t
o
f
t
h
e
v
e
h
icle
to
ac
tu
ate
th
e
s
p
r
a
y
i
n
g
o
f
th
e
h
er
b
icid
e.
First,
w
e
d
ef
i
n
e
a
r
eg
io
n
o
f
in
ter
e
s
t.
T
h
en
w
e
u
s
e
"
Sli
d
in
g
W
in
d
o
w
"
to
co
v
er
all
th
is
w
o
r
k
s
p
ac
e
(
s
ee
Fig
u
r
e
6
)
:
w
it
h
an
ass
u
m
p
tio
n
o
f
f
lat
m
o
d
e
an
d
a
p
r
io
r
k
n
o
w
led
g
e
o
f
s
ee
d
lin
g
lin
e
s
g
eo
m
etr
y
.
I
n
li
n
e
w
i
th
th
is
,
th
e
s
tatic
n
at
u
r
e
o
f
th
e
ag
r
icu
lt
u
r
al
s
ce
n
e
(
ag
r
icu
lt
u
r
al
s
ce
n
e
is
al
m
o
s
t
id
en
tical
in
p
r
ec
is
io
n
ag
r
icu
lt
u
r
e)
,
th
e
h
y
p
o
t
h
eses
g
en
er
ate
d
in
3
D
s
p
ac
e
ar
e
p
r
o
j
ec
ted
d
ir
ec
tly
i
n
to
t
h
e
2
D
i
m
a
g
e.
A
t
a
r
eg
io
n
o
f
in
t
er
est
(
3
2
m
2
)
ar
ea
at
th
e
f
r
o
n
t
o
f
t
h
e
tr
ac
to
r
e
m
b
ed
d
in
g
th
e
ca
m
er
a,
a
s
u
b
ar
ea
o
f
(
1
m
2
)
r
ep
r
esen
ts
t
h
e
s
ize
o
f
th
e
s
lid
i
n
g
w
i
n
d
o
w
.
W
e
th
en
o
b
tain
a
s
e
t
f
r
o
m
6
to
1
0
s
u
b
-
w
i
n
d
o
w
ca
n
d
id
ates.
A
ll
th
e
s
e
s
u
b
-
w
i
n
d
o
w
s
ar
e
ad
j
u
s
ted
in
s
ize
(
3
2
3
2
)
p
ix
els.
W
e
th
e
n
ap
p
ly
t
h
e
d
etec
to
r
,
a
r
esp
o
n
s
e
is
g
e
n
er
ated
f
o
r
ea
ch
t
h
u
m
b
n
a
il.
I
f
th
e
p
r
ese
n
ce
o
f
t
h
e
w
ee
d
is
d
etec
ted
o
n
o
n
e
o
f
th
e
m
,
th
e
ar
ea
is
d
ec
lar
ed
co
n
ta
m
i
n
ated
.
T
h
e
p
r
o
ce
s
s
s
to
p
s
f
o
r
th
e
i
m
ag
e
i
n
p
r
o
ce
s
s
an
d
g
o
es
to
th
e
n
ex
t
i
m
a
g
e.
T
h
e
i
m
p
le
m
e
n
tatio
n
was c
ar
r
ied
o
u
t
w
it
h
t
h
e
s
o
f
t
w
ar
e
C
#
an
d
Op
en
C
V.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
R
ea
l Time
W
ee
d
Dete
ct
io
n
Usi
n
g
a
B
o
o
s
ted
C
a
s
ca
d
e
o
f
S
imp
le
F
ea
tu
r
es
(
A
d
il Ta
n
n
o
u
ch
e
)
2761
Fig
u
r
e
6
.
R
eg
io
n
o
f
I
n
ter
est a
n
d
s
lid
in
g
w
i
n
d
o
w
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
NS
I
n
t
h
is
s
ec
tio
n
,
w
e
a
n
al
y
ze
t
h
e
r
esu
lt
s
f
o
r
th
e
d
escr
ip
to
r
b
ased
o
n
Haar
-
l
ik
e
f
ea
t
u
r
es
d
escr
ib
ed
ab
o
v
e.
Fo
r
g
r
a
y
lev
el
s
i
m
ag
e
s
w
it
h
a
r
eso
lu
tio
n
o
f
(
6
4
0
4
8
0
)
p
ix
els
,
th
e
in
d
ices o
f
p
er
f
o
r
m
a
n
ce
co
n
s
id
er
ed
ar
e:
a.
R
ate
o
f
T
r
u
e
P
o
s
itiv
es
(
T
P
)
is
th
e
p
er
ce
n
tag
e
o
f
t
h
u
m
b
n
ails
r
ep
r
esen
tativ
e
s
w
ee
d
co
r
r
ec
tly
id
en
ti
f
ied
o
n
all
th
e
i
m
ag
e
s
o
f
t
h
e
test
s
et
b.
R
ate
o
f
Fal
s
e
P
o
s
itiv
e
(
FP
)
is
ca
lcu
lated
f
r
o
m
t
h
e
av
er
a
g
e
o
f
f
al
s
e
alar
m
s
p
er
i
m
ag
e
(
ca
lcu
lated
o
n
all
i
m
a
g
es o
f
t
h
e
tes
t b
asis
)
.
c.
Av
er
ag
e
p
r
o
ce
s
s
i
n
g
ti
m
e
t
h
u
m
b
n
ail
o
f
(
3
2
3
2
)
p
ix
els
is
e
v
alu
a
ted
o
n
a
r
eg
u
lar
lap
to
p
(
C
o
r
e
i7
@
2
.
6
6
GHz
an
d
6
GB
R
AM
)
.
W
e
v
ar
ied
t
h
e
v
alu
e
o
f
th
e
m
ax
i
m
u
m
n
u
m
b
er
o
f
d
escr
ip
t
o
r
s
(
T
=
5
0
,
1
0
0
an
d
1
5
0
d
escr
ip
to
r
s
)
.
Fig
u
r
e
7
s
h
o
w
s
t
h
e
R
OC
c
u
r
v
e
s
f
o
r
th
e
s
i
m
p
le
d
etec
to
r
(
Fig
u
r
e
2
)
.
Fig
u
r
e
7
.
R
OC
C
u
r
v
e
f
o
r
T
=
{5
0
,
1
0
0
,
1
5
0
}
T
ab
le
1
.
R
esu
lts
Ob
tai
n
ed
f
o
r
th
e
Si
n
g
le
Dete
ct
o
r
f
o
r
th
e
Go
o
d
Dete
ctio
n
R
ate
95%
N
b
.
D
e
scri
p
t
o
r
(
T
)
T
P
(
%)
F
P
(
%)
T
i
me
(
1
0
-
3
s)
50
9
5
.
2
0
.
1
6
18
1
0
0
9
5
.
6
0
.
1
1
40
1
5
0
9
5
.
2
0
.
0
9
64
T
ab
le
1
s
h
o
w
s
t
h
e
r
es
u
lt
s
f
o
r
th
e
s
i
n
g
le
d
etec
to
r
w
it
h
a
th
r
esh
o
ld
o
f
co
r
r
ec
t
d
etec
tio
n
s
b
ased
o
n
p
o
s
itiv
e
v
alid
atio
n
s
et
a
t
9
5
%.
W
e
also
o
b
s
er
v
e
th
at
t
h
e
co
m
p
u
tat
io
n
ti
m
e
p
er
i
m
a
g
e
n
at
u
r
all
y
in
cr
ea
s
es
w
i
t
h
th
e
a
m
o
u
n
t
o
f
d
escr
ip
to
r
s
.
Usi
n
g
a
lar
g
e
n
u
m
b
er
o
f
d
escr
ip
t
o
r
s
r
ef
in
es
t
h
e
s
ep
ar
atio
n
b
o
u
n
d
ar
y
,
b
u
t
b
ec
o
m
es
u
n
r
ea
lized
as
f
ar
as a
n
ap
p
lica
tio
n
in
r
ea
l ti
m
e
is
co
n
ce
r
n
ed
.
Hen
ce
ca
s
ca
d
i
n
g
ar
ch
itect
u
r
al
in
ter
est.
50
60
70
80
90
1
0
0
0
0
.
1
0
.
2
0
.
3
0
.
4
0
.
5
T
r
u
e
P
o
si
t
i
f
s
(
%)
F
a
l
se
P
o
si
t
i
f
s
(
%)
R
O
C
c
u
r
v
e
@
5
0
@
1
0
0
@
1
5
0
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Dec
em
b
er
2
0
1
6
: 2
7
5
5
–
2
7
6
5
2762
T
ab
le
2
in
d
icate
s
th
e
ca
s
ca
d
e
d
etec
to
r
a
r
ch
itectu
r
e
an
d
p
er
f
o
r
m
a
n
ce
.
T
h
e
p
er
f
o
r
m
an
ce
p
ar
am
ete
r
v
alu
e
s
f
o
r
th
e
f
u
n
c
tio
n
Gi
ar
e
th
e
m
i
n
i
m
u
m
r
ate
o
f
co
r
r
ec
t
d
etec
tio
n
s
T
P
m
in
=
0
.
9
9
,
an
d
th
e
m
ax
i
m
u
m
ac
ce
p
tab
le
r
ate
o
f
f
alse
alar
m
s
FP
m
ax
=
0
.
0
0
0
5
.
W
e
o
b
s
er
v
e
th
e
ex
i
s
te
n
ce
o
f
s
i
g
n
if
ic
an
t
d
is
p
ar
it
y
i
n
th
e
a
m
o
u
n
t
o
f
d
escr
ip
to
r
s
ta
g
es
b
y
t
h
e
ca
s
ca
d
e
.
W
e
al
s
o
n
o
te
th
at
al
l
lear
n
in
g
i
s
s
to
p
p
ed
b
ec
au
s
e
o
f
t
h
e
n
o
n
-
co
n
v
er
g
e
n
ce
o
f
t
h
e
A
d
aB
o
o
s
t a
lg
o
r
ith
m
.
T
ab
le
2
.
R
esu
lts
f
o
r
th
e
Dete
ct
o
r
in
C
asca
de
N
b
.
B
a
se
N
e
g
a
t
i
f
N
b
.
S
t
a
g
e
N
b
.
D
e
scri
p
t
o
r
TP
(
%)
FP
(
%)
T
i
me
(
1
0
-
3
s)
S
t
o
p
p
i
n
g
C
r
i
t
e
r
i
o
n
1
0
0
12
4
0
0
9
5
.
4
0
.
0
0
8
A
b
o
u
t
8
N
o
C
o
n
v
.
2
0
0
11
4
7
9
9
6
.
4
0
.
0
0
7
A
b
o
u
t
7
.
5
N
o
C
o
n
v
.
3
0
0
10
2
7
2
9
7
.
7
0
.
0
0
1
A
b
o
u
t
4
N
o
C
o
n
v
.
W
h
en
w
e
i
n
cr
ea
s
e
th
e
n
u
m
b
e
r
o
f
n
e
g
ati
v
e
t
h
u
m
b
n
a
ils
i
n
le
ar
n
in
g
,
t
h
e
p
r
o
ce
s
s
w
ill
co
n
v
e
r
g
e
m
o
r
e
f
o
r
a
n
u
m
b
er
o
f
lo
w
er
s
ta
g
es
.
T
h
is
is
ea
s
il
y
e
x
p
lai
n
ed
:
a
l
ar
g
er
n
e
g
ati
v
e
b
ase
allo
w
s
th
e
g
e
n
er
atio
n
o
f
a
s
tr
o
n
g
e
n
o
u
g
h
b
o
r
d
er
to
elim
i
n
ate,
f
r
o
m
th
e
f
ir
s
t
s
ta
g
e
s
o
f
t
h
e
ca
s
ca
d
e,
a
lar
g
e
n
u
m
b
er
o
f
f
a
ls
e
d
etec
tio
n
s
.
Ver
y
q
u
ick
l
y
,
leav
i
n
g
o
n
l
y
t
h
e
m
o
s
t
d
if
f
icu
l
t
ca
s
es
to
r
ej
ec
t,
h
en
ce
th
e
n
o
n
-
co
n
v
er
g
e
n
ce
o
f
th
e
alg
o
r
ith
m
.
Mo
r
eo
v
er
,
if
w
e
lo
o
k
at
t
h
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
s
elec
ted
in
ea
ch
s
tag
e
(
Fi
g
u
r
e
8
)
,
w
e
f
i
n
d
th
at
t
h
e
Haa
r
d
etec
to
r
r
eq
u
ir
es m
o
r
e
d
escr
ip
to
r
s
to
esti
m
ate
co
r
r
ec
tl
y
b
ette
r
b
o
r
d
er
b
etw
ee
n
class
e
s
.
Fig
u
r
e
8
.
T
h
e
Am
o
u
n
t o
f
th
e
Descr
ip
to
r
s
I
n
clu
d
ed
in
ea
c
h
S
tag
e
b
y
t
h
e
Dete
ct
o
r
f
o
r
1
0
0
N
eg
ati
v
e
T
h
u
m
b
n
a
ils
T
ab
le
3
s
h
o
w
s
th
e
p
er
f
o
r
m
a
n
c
e
ac
h
iev
ed
b
y
o
u
r
m
e
th
o
d
b
y
ap
p
ly
i
n
g
th
e
li
m
iti
n
g
r
u
le
o
f
th
e
n
u
m
b
er
o
f
d
escr
ip
to
r
s
p
er
s
tag
e
in
th
e
co
n
tr
o
lled
ca
s
ca
d
e.
T
h
is
tech
n
iq
u
e
allo
w
ed
u
s
to
i
n
cr
ea
s
e
t
h
e
to
tal
n
u
m
b
er
o
f
d
escr
ip
to
r
s
an
d
to
ac
h
ie
v
e
a
r
ate
o
f
g
o
o
d
d
etec
tio
n
s
o
f
9
9
%
a
n
d
a
f
al
s
e
alar
m
r
ate
o
f
0
.
0
0
0
5
%
b
y
a
v
o
id
in
g
th
e
n
o
n
-
co
n
v
er
g
e
n
ce
o
f
t
h
e
Ad
aB
o
o
s
t
alg
o
r
ith
m
.
T
h
e
r
u
n
n
i
n
g
t
i
m
e
i
s
ac
ce
p
tab
le
f
o
r
an
ap
p
licatio
n
in
r
ea
l
ti
m
e
t
h
r
o
u
g
h
th
e
p
r
in
cip
le
o
f
atten
tio
n
al
ca
s
ca
d
e.
I
n
d
ee
d
,
th
e
av
er
ag
e
p
r
o
ce
s
s
in
g
ti
m
e
f
o
r
an
i
m
a
g
e
is
in
t
h
e
o
r
d
er
o
f
0
.
2
s
ec
o
n
d
.
T
h
u
s
,
t
h
eo
r
etica
ll
y
,
th
e
tr
ac
to
r
ca
n
r
ea
c
h
a
m
a
x
i
m
u
m
s
p
ee
d
ab
o
u
t
(
1
0
m
/s
)
o
r
(
3
6
k
m
/
h
)
.
Fig
u
r
e
9
s
h
o
w
s
a
n
ex
a
m
p
le
o
f
w
ee
d
d
etec
tio
n
i
n
o
n
e
f
r
a
m
e.
T
ab
le
3
.
R
esu
lts
f
o
r
th
e
Dete
c
t
o
r
in
Co
n
tr
o
lled
C
ascad
e
N
b
.
B
a
se
N
e
g
a
t
i
f
N
b
.
D
e
scri
p
t
o
r
TP
(
%)
FP
(
%)
T
i
me
(
1
0
-
3
s)
1
0
0
8
6
6
9
9
.
2
0
.
0
0
0
5
A
b
o
u
t
1
5
.
5
2
0
0
8
1
2
99
<
0
.
0
0
0
5
A
b
o
u
t
1
5
3
0
0
8
0
2
99
<
0
.
0
0
0
5
A
b
o
u
t
1
5
0
20
40
60
80
1
0
0
1
2
0
1
4
0
1
2
3
4
5
6
7
8
9
10
11
D
e
scri
p
t
o
r
S
t
a
g
e
D
e
scri
p
t
o
r
=
f
(
S
t
a
g
e
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
R
ea
l Time
W
ee
d
Dete
ct
io
n
Usi
n
g
a
B
o
o
s
ted
C
a
s
ca
d
e
o
f
S
imp
le
F
ea
tu
r
es
(
A
d
il Ta
n
n
o
u
ch
e
)
2763
T
h
is
Fig
u
r
e
s
h
o
w
s
th
e
s
lid
i
n
g
w
i
n
d
o
w
(
W
h
ite
Sq
u
ar
e)
f
o
c
u
s
ed
o
n
a
p
o
p
u
lated
ar
ea
o
f
w
e
e
d
s
in
t
h
e
r
eg
io
n
o
f
in
ter
est
(
Yello
w
Sq
u
ar
e)
.
T
h
e
r
u
n
n
in
g
ti
m
e
o
f
t
h
is
s
lid
i
n
g
w
i
n
d
o
w
is
(
3
5
.
1
0
-
3
s
)
as
s
h
o
w
n
in
th
e
Fi
g
u
r
e
9
.
Fig
u
r
e
9
.
Scr
ee
n
s
h
o
t o
f
Ou
r
Ap
p
lica
tio
n
4
.
CO
NCLUS
I
O
N
W
e
h
av
e
p
r
o
p
o
s
ed
a
n
e
w
ap
p
r
o
ac
h
b
y
m
ac
h
i
n
e
lear
n
in
g
to
d
etec
t
w
ee
d
s
.
T
h
e
o
r
ig
in
a
l
it
y
o
f
o
u
r
ap
p
r
o
ac
h
lies
in
t
h
e
s
tu
d
y
a
n
d
ev
al
u
atio
n
o
f
ar
ti
f
icial
v
i
s
io
n
tec
h
n
o
lo
g
y
,
w
h
ic
h
i
s
wid
el
y
k
n
o
w
n
an
d
r
ec
o
g
n
ized
in
o
b
j
ec
t
r
ec
o
g
n
itio
n
,
in
w
ee
d
d
etec
tio
n
.
T
h
e
m
et
h
o
d
u
s
es
Haar
-
li
k
e
f
ea
t
u
r
es
to
d
escr
ib
e
th
e
ag
r
icu
l
tu
r
al
s
ce
n
e.
T
h
e
d
escr
ip
tio
n
v
ec
to
r
cr
ea
ted
f
ee
d
s
th
e
A
d
aB
o
o
s
t
alg
o
r
ith
m
.
T
h
en
,
a
r
elev
an
t
d
is
cr
i
m
i
n
ati
v
e
ca
s
ca
d
e
o
f
w
ea
k
clas
s
i
f
ier
s
allo
w
s
th
e
d
etec
t
io
n
o
f
w
ee
d
s
i
n
r
ea
l
-
ti
m
e.
T
h
e
r
esu
lt
s
p
r
o
v
e
th
e
v
alid
it
y
o
f
o
u
r
ap
p
r
o
ac
h
a
n
d
s
h
o
w
r
e
m
ar
k
ab
le
p
er
f
o
r
m
an
c
e.
T
h
e
co
r
r
ec
t
an
d
f
al
s
e
d
etec
tio
n
r
ates
r
ea
ch
th
e
v
alu
e
o
f
9
9
%
a
n
d
0
.
0
0
0
5
%,
r
esp
ec
tiv
el
y
,
w
it
h
a
n
e
x
ec
u
ti
o
n
r
ate
o
f
5
f
p
s
.
R
esear
ch
s
t
u
d
ies
ar
e
c
u
r
r
en
tl
y
u
n
d
er
w
a
y
to
test
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
o
u
r
m
eth
o
d
in
th
e
f
iel
d
.
Oth
er
ap
p
r
o
ac
h
es
w
it
h
m
o
r
e
co
m
p
lex
f
ea
t
u
r
es
(
His
to
g
r
a
m
o
f
Or
ien
ted
Gr
ad
ien
ts
(
HOG)
[
2
1
]
,
Gab
o
r
f
ilter
[
2
2
]
)
co
n
s
id
er
ed
in
d
iv
id
u
all
y
o
r
in
co
m
b
i
n
atio
n
s
,
in
d
is
cr
i
m
in
a
tiv
e
o
r
g
e
n
er
ati
v
e
m
o
d
e
[
1
4
]
ar
e
also
u
n
d
er
s
tu
d
y
b
y
o
u
r
r
esear
ch
g
r
o
u
p
.
RE
F
E
R
E
NC
E
S
[1
]
T
a
n
n
o
u
c
h
e
A
,
S
b
a
i
K,
Ra
h
m
o
u
n
e
M
,
Zo
u
b
ir
A
,
A
g
o
u
n
o
u
n
e
R,
S
a
a
d
a
n
i
R,
Ra
h
m
a
n
i
A
.
A
F
a
st
a
n
d
Eff
icie
n
t
S
h
a
p
e
De
s
c
rip
to
r
f
o
r
a
n
A
d
v
a
n
c
e
d
W
e
e
d
T
y
p
e
Clas
si
f
ic
a
ti
o
n
A
p
p
ro
a
c
h
.
I
n
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
ter
En
g
i
n
e
e
rin
g
(
IJ
ECE
)
.
2
0
1
6
;
6
(3
):
1
1
6
8
-
1
1
7
5
.
[2
]
T
a
n
n
o
u
c
h
e
A
,
S
b
a
i
K,
Ou
n
e
jj
a
r
Y,
Ra
h
m
a
n
i
A
.
A
re
a
l
ti
m
e
e
ff
i
c
ien
t
m
a
n
a
g
e
m
e
n
t
o
f
o
n
i
o
n
s
w
e
e
d
s
b
a
se
d
on
a
m
u
lt
il
a
y
e
r
p
e
rc
e
p
tro
n
n
e
u
ra
l
n
e
t
w
o
rk
s
tec
h
n
iq
u
e
.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
F
a
rm
in
g
a
n
d
Al
li
e
d
S
c
ien
c
e
s
(
IJ
FA
S
)
.
2
0
1
5
;
4
(2
)
:
1
6
1
-
1
6
6
.
[3
]
A
le
m
a
y
e
h
u
DM,
M
e
n
g
istu
A
D,
M
e
n
g
istu
S
G
.
Co
m
p
u
ter
v
isio
n
f
o
r
Et
h
io
p
ian
a
g
ricu
lt
u
ra
l
c
ro
p
p
e
st
id
e
n
ti
f
ica
ti
o
n
.
In
d
o
n
e
sia
n
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
En
g
i
n
e
e
rin
g
a
n
d
Co
mp
u
ter
S
c
ien
c
e
.
2
0
1
6
;
3
(
1
).
[4
]
M
u
ru
g
a
n
D
,
P
re
m
a
P
.
A
No
v
e
l
A
n
g
u
lar
T
e
x
tu
re
P
a
tt
e
rn
(A
T
P
)
Ex
trac
ti
o
n
M
e
t
h
o
d
f
o
r
Cr
o
p
a
n
d
W
e
e
d
Disc
ri
m
in
a
ti
o
n
Us
i
n
g
Cu
rv
e
let
T
r
a
n
sf
o
r
m
a
ti
o
n
.
EL
CVI
A:
e
lec
tro
n
i
c
letter
s
o
n
c
o
mp
u
ter
v
isio
n
a
n
d
i
ma
g
e
a
n
a
lys
is
.
2
0
1
6
;
15
(
1
):
2
7
-
5
9
.
[5
]
S
u
rian
sy
a
h
M
I,
S
u
k
o
c
o
H,
S
o
la
h
u
d
in
M
.
W
e
e
d
De
tec
ti
o
n
Us
in
g
F
ra
c
tal
-
Ba
s
e
d
L
o
w
Co
st
Co
m
m
o
d
it
y
a
rd
wa
re
Ra
sp
b
e
rry
P
i.
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
i
n
e
e
rin
g
a
n
d
Co
mp
u
ter
S
c
ien
c
e
.
2
0
1
6
;
2
(
2
),
4
2
6
-
4
3
0
.
[6
]
Ore
n
M
,
P
a
p
a
g
e
o
r
g
io
u
C,
S
in
h
a
P
,
Os
u
n
a
E,
P
o
g
g
io
T
.
Ped
e
stria
n
d
e
tec
ti
o
n
u
sin
g
wa
v
e
let
tem
p
l
a
tes
.
Co
m
p
u
ter
V
isio
n
a
n
d
P
a
t
tern
Re
c
o
g
n
it
io
n
,
1
9
9
7
.
P
r
o
c
e
e
d
in
g
s.
1
9
9
7
IE
EE
Co
m
p
u
ter
S
o
c
iety
Co
n
fe
re
n
c
e
o
n
(p
p
.
1
9
3
-
1
9
9
)
.
IEE
E.
[7
]
P
a
p
a
g
e
o
rg
io
u
C,
P
o
g
g
io
.
A
trai
n
a
b
le
sy
ste
m
f
o
r
o
b
jec
t
d
e
tec
ti
o
n
.
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
Co
mp
u
ter
Vi
sio
n
.
2
0
0
0
;
38
(
1
):
1
5
-
3
3
.
[8
]
V
io
la
P
,
Jo
n
e
s
M
.
Ra
p
id
o
b
jec
t
d
e
tec
ti
o
n
u
si
n
g
a
b
o
o
ste
d
c
a
sc
a
d
e
o
f
sim
p
le
fea
t
u
re
s
.
In
Co
m
p
u
ter
V
isio
n
a
n
d
P
a
tt
e
r
n
Re
c
o
g
n
it
i
o
n
,
2
0
0
1
.
C
VPR
2
0
0
1
.
P
r
o
c
e
e
d
in
g
s
o
f
t
h
e
2
0
0
1
IEE
E
Co
m
p
u
ter
S
o
c
iety
Co
n
f
e
r
e
n
c
e
o
n
(
Vo
l.
1
,
p
p
.
I
-
5
1
1
).
IEE
E.
[9
]
L
ien
h
a
rt
R,
Ku
ra
n
o
v
A
,
P
isa
re
v
sk
y
V
.
E
m
p
iri
c
a
l
a
n
a
l
y
sis
o
f
d
e
tec
ti
o
n
c
a
sc
a
d
e
s
o
f
b
o
o
ste
d
c
las
s
if
iers
f
o
r
ra
p
id
o
b
jec
t
d
e
tec
ti
o
n
.
In
J
o
in
t
Pa
t
ter
n
Rec
o
g
n
it
io
n
S
y
mp
o
siu
m
(
2
0
0
3
;
p
p
.
2
9
7
-
3
0
4
).
S
p
ri
n
g
e
r
Be
rli
n
He
id
e
lb
e
rg
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
J
E
C
E
Vo
l.
6
,
No
.
6
,
Dec
em
b
er
2
0
1
6
: 2
7
5
5
–
2
7
6
5
2764
[1
0
]
He
rre
ra
P
J,
Do
ra
d
o
J,
Ri
b
e
iro
Á
.
A
No
v
e
l
A
p
p
ro
a
c
h
f
o
r
W
e
e
d
Ty
p
e
Clas
si
f
ica
ti
o
n
Ba
se
d
o
n
S
h
a
p
e
De
sc
rip
to
rs
a
n
d
a
F
u
z
z
y
D
e
c
isio
n
-
M
a
k
in
g
M
e
th
o
d
.
S
e
n
so
rs
.
2
0
1
4
;
14
(8
):
1
5
3
0
4
-
1
5
3
2
4
.
[1
1
]
G
é
e
C,
Bo
ss
u
J,
Jo
n
e
s
G
,
T
ru
c
h
e
tetF
.
Cro
p
/w
e
e
d
d
isc
rim
in
a
ti
o
n
i
n
p
e
rsp
e
c
ti
v
e
a
g
ro
n
o
m
ic
i
m
a
g
e
s.
Co
mp
u
ter
s
a
n
d
El
e
c
tro
n
ics
in
Ag
ric
u
lt
u
re
.
2
0
0
8
;
60
(
1
):
4
9
-
5
9
.
[1
2
]
S
u
n
Z,
Be
b
is
G
,
M
il
ler
R.
On
-
ro
a
d
v
e
h
icle
d
e
tec
ti
o
n
:
A
re
v
i
e
w
.
IEE
E
tra
n
sa
c
ti
o
n
s
o
n
p
a
tt
e
r
n
a
n
a
lys
is
a
n
d
ma
c
h
i
n
e
in
telli
g
e
n
c
e
.
2
0
0
6
;
28
(
5
):
6
9
4
-
7
1
1
.
[1
3
]
A
d
h
ik
a
ri
S
P
,
Yo
o
HJ
,
Kim
H.
Bo
o
sti
n
g
-
b
a
se
d
o
n
-
ro
a
d
o
b
sta
c
le
se
n
sin
g
u
sin
g
d
isc
rim
in
a
ti
v
e
we
a
k
c
las
sif
i
e
rs.
S
e
n
so
rs
.
2
0
0
1
;
11
(4
):
4
3
7
2
-
4
3
8
4
.
[1
4
]
Ne
g
ri
P
,
Clad
y
X
,
Ha
n
if
S
M
,
P
re
v
o
st
L
.
A
c
a
sc
a
d
e
o
f
b
o
o
ste
d
g
e
n
e
ra
ti
v
e
a
n
d
d
isc
rim
in
a
ti
v
e
c
la
ss
i
f
iers
f
o
r
v
e
h
icl
e
d
e
tec
ti
o
n
.
EURA
S
IP
J
o
u
rn
a
l
o
n
A
d
v
a
n
c
e
s in
S
i
g
n
a
l
Pr
o
c
e
ss
in
g
.
2
0
0
8
;
1
3
6
.
[1
5
]
S
iala
M
,
Kh
li
f
a
N,
Bre
m
o
n
d
F
,
Ha
m
ro
u
n
i
K
.
Peo
p
le
d
e
tec
ti
o
n
in
c
o
mp
lex
sc
e
n
e
u
sin
g
a
c
a
sc
a
d
e
o
f
Bo
o
ste
d
c
la
ss
if
ier
s b
a
se
d
o
n
Ha
a
r
-
l
ike
-
fea
tu
re
s
.
In
I
n
telli
g
e
n
t
V
e
h
icle
s S
y
m
p
o
siu
m
,
2
0
0
9
IEE
E
(p
p
.
8
3
-
8
7
).
I
EE
E.
[1
6
]
P
a
v
a
n
i
S
K,
De
lg
a
d
o
D,
F
ra
n
g
i,
A
.
F
.
Ha
a
r
-
li
k
e
f
e
a
tu
re
s
w
it
h
o
p
ti
m
a
ll
y
w
e
i
g
h
ted
re
c
tan
g
les
f
o
r
ra
p
id
o
b
jec
t
d
e
tec
ti
o
n
.
Pa
t
ter
n
Rec
o
g
n
it
i
o
n
.
2
0
1
0;
43
(1
)
:
1
6
0
-
1
7
2
.
[1
7
]
F
re
u
n
d
Y,
S
c
h
a
p
ire RE
.
Ex
p
e
rime
n
ts
w
it
h
a
n
e
w
b
o
o
sti
n
g
a
lg
o
rit
h
m
.
In
Ic
ml
(1
9
9
6
;
9
6
:
1
4
8
-
1
5
6
).
[1
8
]
F
ried
m
a
n
J,
H
a
stie
T
,
T
ib
sh
iran
i
R.
A
d
d
it
iv
e
lo
g
isti
c
re
g
re
s
sio
n
:
a
sta
ti
stica
l
v
ie
w
o
f
b
o
o
sti
n
g
(w
it
h
d
isc
u
ss
io
n
a
n
d
a
re
jo
in
d
e
r
b
y
th
e
a
u
th
o
rs).
T
h
e
a
n
n
a
ls
o
f
sta
ti
st
ics
.
2
0
0
0
;
28
(2
)
:
3
3
7
-
4
0
7
.
[1
9
]
W
u
S
,
Na
g
a
h
a
sh
i
H.
A
n
a
l
y
sis
o
f
g
e
n
e
ra
li
z
a
ti
o
n
a
b
i
li
ty
f
o
r
d
if
fe
re
n
t
a
d
a
b
o
o
stv
a
rian
ts
b
a
se
d
o
n
c
las
sif
ic
a
ti
o
n
a
n
d
re
g
re
ss
io
n
tree
s.
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
Co
mp
u
ter
En
g
i
n
e
e
rin
g
.
2
0
1
5
,
8
.
[2
0
]
Jo
n
e
s
G
,
Gé
e
C,
T
ru
c
h
e
tet
F
.
M
o
d
e
ll
i
n
g
a
g
ro
n
o
m
ic
i
m
a
g
e
s
f
o
r
we
e
d
d
e
tec
ti
o
n
a
n
d
c
o
m
p
a
ris
o
n
o
f
c
ro
p
/w
e
e
d
d
isc
rim
in
a
ti
o
n
a
lg
o
rit
h
m
p
e
rf
o
r
m
a
n
c
e
.
Pre
c
isio
n
A
g
ric
u
lt
u
re
.
2
0
0
9
;
10
(
1
):
1
-
1
5
.
[2
1
]
Da
lal
N,
T
rig
g
s
B.
Histo
g
ra
ms
o
f
o
rie
n
te
d
g
ra
d
ien
ts
f
o
r
h
u
m
a
n
d
e
tec
ti
o
n
.
I
n
2
0
0
5
IEE
E
Co
m
p
u
ter
S
o
c
iety
Co
n
f
e
re
n
c
e
o
n
Co
m
p
u
ter
V
isi
o
n
a
n
d
P
a
tt
e
rn
Re
c
o
g
n
it
i
o
n
(C
VP
R'
0
5
)
.
2
0
0
5
;
1
:
8
8
6
-
8
9
3
.
IEE
E
.
[2
2
]
S
u
n
Z,
Be
b
is
G
,
M
il
ler
R.
O
n
-
ro
a
d
v
e
h
icle
d
e
tec
ti
o
n
u
sin
g
e
v
o
lu
ti
o
n
a
ry
G
a
b
o
r
f
il
ter
o
p
ti
m
iz
a
ti
o
n
.
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
In
tell
ig
e
n
t
T
ra
n
s
p
o
rta
ti
o
n
S
y
ste
m
s
.
2
0
0
5
;
6
(2
)
:
1
2
5
-
1
3
7
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Ad
il
Ta
n
n
o
u
c
h
e
is
c
u
rre
n
tl
y
a
P
h
D
stu
d
e
n
t
a
t
th
e
L
a
b
o
ra
t
o
ired
’
E
tu
d
e
d
e
s
M
a
te
é
riau
x
A
v
a
n
c
e
é
s
e
t
A
p
p
li
c
a
ti
o
n
s,
M
o
u
lay
Is
m
a
il
Un
iv
e
rsit
y
,
F
a
c
u
lt
y
o
f
S
c
ien
c
e
s
in
M
e
k
n
e
s,
M
o
r
o
c
c
o
.
His
re
se
a
rc
h
in
tere
sts
a
re
f
o
c
u
se
d
in
m
a
c
h
in
e
v
isio
n
,
a
rti
f
icia
l
in
telli
g
e
n
c
e
a
n
d
th
e
ir
s
a
p
p
li
c
a
ti
o
n
i
n
a
g
ricu
lt
u
re
.
K
h
a
li
d
S
b
a
i
i
s
a
f
u
ll
p
ro
f
e
ss
o
r
sin
c
e
2
0
0
1
in
El
e
c
tro
n
ics
.
He
re
c
e
iv
e
d
h
is
M
.
sc
.
De
g
re
e
in
El
e
c
tro
n
ics
f
ro
m
V
a
len
c
ien
n
e
U
n
iv
e
rsity
(F
ra
n
c
e
)
in
1
9
9
6
a
n
d
h
i
s
Ha
b
il
it
a
ti
o
n
i
n
P
h
y
sic
s
f
ro
m
M
o
u
lay
Is
m
a
il
Un
iv
e
rsit
y
in
2
0
0
8
.
His
re
se
a
rc
h
in
tere
sts
in
c
l
u
d
e
S
tru
c
tu
ra
l
st
u
d
ies
,
v
ib
ra
ti
o
n
a
l
a
n
d
e
lec
tro
n
ic p
r
o
p
e
rti
e
s o
f
c
a
rb
o
n
n
a
n
o
t
u
b
e
s
M
i
lo
u
d
Ra
h
m
o
u
n
e
is
a
f
u
ll
p
ro
f
e
ss
o
r
a
t
M
o
u
lay
Is
m
a
il
Un
iv
e
rsit
y
.
He
re
c
e
i
v
e
d
h
is
M
sc
.
De
g
re
e
in
a
p
p
li
e
d
m
e
c
h
a
n
ics
f
ro
m
Un
iv
e
rsite
´
M
o
n
tp
e
ll
ier
2
(F
ra
n
c
e
)
a
n
d
h
is
P
h
.
D.
d
e
g
re
e
s
in
M
e
c
h
a
tro
n
ics
f
ro
m
Un
iv
e
rsit
M
o
n
tp
e
ll
ier
2
(F
ra
n
c
e
)
a
n
d
Un
iv
e
rsite´
Ha
s
sa
n
II
M
o
h
a
m
m
e
d
ia,
in
1
9
9
3
a
n
d
1
9
9
6
re
sp
e
c
ti
v
e
l
y
.
His
re
se
a
r
c
h
in
tere
sts
in
c
lu
d
e
stru
c
tu
ra
l
D
y
n
a
m
ics
,
a
c
ti
v
e
c
o
n
tro
l,
a
n
d
sm
a
rt
m
a
teria
ls
.
Evaluation Warning : The document was created with Spire.PDF for Python.