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.
7
,
No
.
2
,
A
p
r
il
201
7
,
p
p
.
823
~
8
3
0
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
7
i
2
.
p
p
8
2
3
-
8
3
0
823
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
I
m
a
g
e P
ro
c
ess
ing
f
o
r Rapidly
Ey
e
Dete
c
tion
b
a
sed
o
n Robus
t
H
a
a
r
Sliding
Win
do
w
F
it
ri
Ut
a
m
ini
ng
ru
m
,
Rena
l
d
i P
ri
m
a
s
w
a
ra
,
Yuita
Aru
m
S
a
ri
F
a
c
u
lt
y
o
f
Co
m
p
u
ter S
c
ien
c
e
,
Bra
w
ij
a
y
a
Un
iv
e
rsit
y
,
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
Dec
6
,
2
0
1
6
R
ev
i
s
ed
Feb
15
,
2
0
1
7
A
cc
ep
ted
Mar
3
,
2
0
1
7
Ob
jec
t
De
tec
ti
o
n
u
sin
g
Ha
a
r
C
a
sc
a
d
e
Cla
sif
i
e
r
w
id
e
l
y
a
p
p
li
e
d
in
se
v
e
ra
l
d
e
v
ice
s
a
n
d
a
p
p
li
c
a
ti
o
n
s
a
s
a
m
e
d
iu
m
o
f
in
tera
c
ti
o
n
b
e
tw
e
e
n
h
u
m
a
n
a
n
d
c
o
m
p
u
ter
su
c
h
a
s
a
t
o
o
l
c
o
n
tro
l
t
h
a
t
u
t
il
ize
s
th
e
d
e
tec
ti
o
n
o
f
e
y
e
m
o
v
e
m
e
n
ts.
Ob
v
io
u
sly
sp
e
e
d
a
n
d
p
re
c
isio
n
in
th
e
d
e
tec
ti
o
n
p
ro
c
e
ss
su
c
h
a
s
e
y
e
s,
h
a
s
a
n
e
ffe
c
t
i
f
i
m
p
le
m
e
n
ted
o
n
a
d
e
v
i
c
e
.
I
f
th
e
e
y
e
c
o
u
ld
n
o
t
d
e
tec
t
a
c
c
u
ra
tel
y
,
c
o
n
tro
ll
in
g
d
e
v
ice
s
y
ste
m
s
c
o
u
ld
re
a
c
h
b
a
d
d
e
tec
ti
o
n
a
s
w
e
ll
.
T
h
e
p
ro
p
o
se
d
m
e
th
o
d
c
a
n
b
e
u
se
d
a
s
a
n
a
p
p
ro
a
c
h
to
d
e
tec
t
th
e
e
y
e
r
e
g
io
n
o
f
e
y
e
b
a
se
d
o
n
h
a
a
r
c
las
si
f
ier
m
e
th
o
d
b
y
m
e
a
n
s
o
f
m
o
d
ify
in
g
th
e
d
irec
ti
o
n
o
f
slid
in
g
w
in
d
o
w
.
In
w
h
ich
,
it
w
a
s
in
it
ially
p
lac
e
d
in
th
e
m
id
d
le
p
o
siti
o
n
o
f
i
m
a
g
e
o
n
f
a
c
ial
a
r
e
a
b
y
a
ss
u
m
in
g
th
e
lo
c
a
ti
o
n
o
f
e
y
e
s
a
r
e
a
in
th
e
c
e
n
tral
re
g
io
n
o
f
th
e
im
a
g
e
.
W
h
il
e
th
e
w
in
d
o
w
r
e
g
i
o
n
o
f
c
o
n
v
e
n
ti
o
n
a
l
h
a
a
r
c
a
sc
a
d
e
sc
a
n
t
h
e
w
h
o
le
o
f
i
m
a
g
e
sta
rt
f
ro
m
th
e
le
ft
to
p
c
o
r
n
e
r
.
F
ro
m
th
e
e
x
p
e
ri
m
e
n
t
b
y
u
sin
g
o
u
r
p
ro
p
o
se
d
m
e
th
o
d
,
it
c
a
n
s
p
e
e
d
u
p
t
h
e
t
h
e
c
o
m
p
u
tatio
n
ti
m
e
a
n
d
im
p
ro
v
e
a
c
c
u
ra
c
y
si
g
n
if
ica
n
tl
y
re
a
c
h
to
9
2
,
4
%
.
K
ey
w
o
r
d
:
Dete
ctio
n
E
y
es
Haa
r
ca
s
ca
d
e
class
i
f
ier
Sli
d
in
g
w
i
n
d
o
w
Co
p
y
rig
h
t
©
2
0
1
7
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
:
Fit
r
i U
ta
m
in
in
g
r
u
m
,
Facu
lt
y
o
f
C
o
m
p
u
ter
Scien
ce
,
B
r
a
w
ij
ay
a
U
n
i
v
er
s
it
y
,
Vete
r
an
R
o
ad
No
.
8
Ma
lan
g
,
E
ast J
av
a,
I
n
d
o
n
esia.
E
m
ail:
f
3
_
n
i
n
g
r
u
m
@
u
b
.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
I
n
d
ig
ital
i
m
ag
e
p
r
o
ce
s
s
i
n
g
,
t
h
er
e
ar
e
s
ev
er
al
m
et
h
o
d
s
t
h
at
h
av
e
b
ee
n
u
s
ed
an
d
o
n
e
o
f
t
h
e
m
i
s
h
aa
r
ca
s
ca
d
e
class
if
ier
.
Haa
r
ca
s
ca
d
e
h
as
b
ee
n
w
id
el
y
ap
p
lied
f
o
r
h
u
m
a
n
s
an
d
co
m
p
u
ter
s
in
ter
ac
tio
n
(
HC
I
)
.
T
h
is
i
m
p
le
m
en
ta
tio
n
f
r
eq
u
e
n
tl
y
u
s
ed
f
o
r
th
e
d
etec
tio
n
o
f
o
b
j
ec
t
s
u
c
h
as
f
ac
ial
ar
ea
,
m
o
u
t
h
d
etec
tio
n
,
an
d
e
y
e
d
etec
tio
n
.
E
y
e
g
az
e
tr
ac
k
i
n
g
d
o
in
g
b
y
G
h
an
i
et
al.
[
1
]
f
o
r
co
n
tr
o
llin
g
m
o
u
s
e
p
o
in
ter
,
Gao
et
al.
[
2
]
d
ev
elo
p
ed
a
m
et
h
o
d
b
ased
o
n
h
aa
r
clas
s
i
f
ier
,
d
etec
tio
n
o
f
e
y
e
m
o
v
e
m
e
n
t
f
o
r
co
n
tr
o
llin
g
w
h
ee
lc
h
air
p
r
esen
t
b
y
Si
n
g
h
et
al.
[
3
]
u
s
in
g
E
O
G
s
i
g
n
al,
a
n
d
C
o
etze
r
et
al
[
4
]
i
m
p
le
m
e
n
t
SVM
an
d
ANN
f
o
r
r
ea
l
ti
m
e
e
y
e
d
etec
tio
n
.
T
h
e
p
r
o
ce
s
s
o
f
h
aa
r
ca
s
ca
d
e
i
n
d
etec
tin
g
e
y
es
is
u
s
in
g
th
e
tech
n
i
q
u
e
o
f
th
e
s
lid
i
n
g
w
i
n
d
o
w
o
r
r
ec
tan
g
u
lar
w
in
d
o
w
o
n
th
e
w
h
o
le
i
m
ag
e
a
n
d
f
i
n
d
o
u
t i
f
t
h
er
e
i
s
a
p
ar
t o
f
th
e
i
m
ag
e
th
at
is
s
h
ap
ed
li
k
e
a
n
e
y
e
o
r
n
o
t.
Haa
r
b
ased
o
n
ca
s
ca
d
e
m
o
d
els
i
n
cl
u
d
e
a
weig
h
t
co
n
n
e
ctio
n
f
r
o
m
t
h
e
i
n
p
u
t
to
ea
ch
la
y
er
a
n
d
f
r
o
m
ea
ch
la
y
er
to
t
h
e
s
u
cc
e
s
s
i
v
e
la
y
er
s
[
5
]
.
B
ef
o
r
e
d
o
in
g
t
h
e
d
etec
tio
n
o
f
th
e
e
y
e
ar
ea
,
th
er
e
ar
e
s
o
m
e
p
r
o
ce
s
s
o
f
i
m
a
g
e
p
r
o
ce
s
s
i
n
g
in
ad
v
an
ce
.
R
GB
i
m
ag
e
i
n
itia
ll
y
co
n
v
er
ted
in
to
a
g
r
ay
s
ca
le
im
a
g
e,
th
e
r
es
u
lt
w
ill
b
e
ca
lcu
late
u
s
in
g
in
teg
r
al
i
m
a
g
e
at
th
e
p
ix
el
o
f
g
r
a
y
s
ca
l
e
i
m
a
g
e.
Haa
r
f
ea
t
u
r
es
p
r
o
ce
s
s
g
r
a
y
s
ca
le
i
m
ag
e
i
n
a
r
ec
ta
n
g
u
lar
w
i
n
d
o
w
w
it
h
a
ce
r
tain
s
ize
s
u
c
h
as
2
4
x
2
4
p
i
x
els.
I
n
s
id
e
th
i
s
w
i
n
d
o
w
,
t
h
e
f
ilter
in
g
p
r
o
ce
s
s
p
er
f
o
r
m
ed
to
f
in
d
t
h
e
e
y
e
r
eg
io
n
.
T
h
is
f
ilt
er
i
n
g
w
as
d
o
n
e
u
s
i
n
g
s
elec
ted
h
aa
r
f
ea
t
u
r
e
te
m
p
lat
e.
I
f
th
e
v
al
u
e
o
f
th
e
h
aa
r
f
e
atu
r
es
i
n
a
ce
r
tai
n
r
eg
io
n
b
elo
w
t
h
e
th
r
es
h
o
ld
th
en
it
h
as
to
b
e
clu
s
ter
ed
.
Ho
w
e
v
er
,
if
a
f
ea
t
u
r
e
in
a
r
eg
io
n
is
n
o
t
q
u
ali
f
ied
,
s
o
th
er
e
is
n
o
o
b
j
ec
t
to
b
e
d
etec
ted
in
th
e
ar
ea
o
f
th
e
r
eg
io
n
an
d
its
r
eg
io
n
ar
ea
r
elo
ca
tes
th
e
ar
ea
.
H
o
w
ev
er
,
if
th
e
ar
ea
m
ee
ts
t
h
e
f
ea
tu
r
e,
t
h
en
th
e
r
u
les
ar
e
p
er
f
o
r
m
ed
to
th
e
n
e
x
t
f
ea
t
u
r
e.
I
f
all
r
e
q
u
ir
e
m
e
n
ts
f
u
l
f
illed
f
ea
t
u
r
e,
it
m
ea
n
s
t
h
at
th
er
e
i
s
an
o
b
j
ec
t in
th
e
ar
ea
o
f
th
e
r
eg
i
o
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.
7
,
No
.
2
,
A
p
r
il
2
0
1
7
:
8
2
3
–
8
3
0
824
I
n
d
etec
tin
g
an
o
b
j
ec
t,
Haa
r
f
e
atu
r
e
h
as
a
f
air
l
y
g
o
o
d
ac
cu
r
ac
y
r
ate,
b
u
t
u
s
i
n
g
h
aa
r
f
ea
t
u
r
e
r
eq
u
ir
es
a
lo
n
g
co
m
p
u
tatio
n
ti
m
e.
I
n
ac
co
r
d
an
ce
w
it
h
th
e
h
aa
r
m
et
h
o
d
,
th
is
h
ap
p
en
s
b
ec
au
s
e
h
aa
r
d
o
p
r
o
ce
s
s
in
g
o
n
al
l
p
ar
ts
o
f
t
h
e
i
m
a
g
e
p
ix
el
s
eq
u
e
n
tiall
y
.
I
n
tr
ain
in
g
p
r
o
ce
s
s
h
a
ar
u
s
ed
A
d
aB
o
o
s
t
f
o
r
clas
s
i
f
ie
r
th
e
o
b
j
ec
t
[
6
]
,
[
7
]
.
Ho
w
e
v
er
,
e
y
e
d
etec
tio
n
ca
n
af
f
ec
t
th
e
ac
c
u
r
ac
y
co
n
tr
o
l
o
f
d
ev
ice
o
r
ap
p
licatio
n
th
a
t
u
tili
ze
s
t
h
e
e
y
e
ar
ea
ac
cu
r
atel
y
.
B
esid
es
th
o
s
e
m
et
h
o
d
s
,
o
cc
asio
n
al
l
y
i
m
ag
e
co
m
e
u
p
w
i
th
n
o
i
s
e
o
f
ten
a
f
f
ec
t
th
e
q
u
al
it
y
o
f
t
h
e
i
m
a
g
e.
So
th
at,
it
h
as
an
i
m
p
a
ct
o
n
th
e
ac
cu
r
ac
y
i
n
e
y
e
d
etec
tio
n
[
8
]
.
P
r
ev
io
u
s
r
esear
ch
an
d
m
et
h
o
d
s
o
f
f
er
ed
b
y
Vio
la
J
o
n
es
[
9
]
w
h
ic
h
u
s
e
ce
r
tain
w
i
n
d
o
w
s
ize
s
m
o
v
in
g
at
ea
ch
p
ix
el
f
r
o
m
t
h
e
u
p
p
er
lef
t
co
r
n
er
to
th
e
b
o
tto
m
r
i
g
h
t.
I
n
th
i
s
p
ap
er
,
w
e
p
r
o
p
o
s
ed
a
m
et
h
o
d
b
y
u
t
ilizi
n
g
a
s
lid
i
n
g
w
i
n
d
o
w
a
n
d
h
aa
r
f
ea
t
u
r
es,
w
h
ic
h
t
h
e
d
etec
tio
n
w
i
n
d
o
w
s
tar
t
s
f
r
o
m
th
e
ce
n
ter
o
f
t
h
e
i
m
a
g
e
w
h
ic
h
alr
ea
d
y
in
d
icate
s
t
h
e
ar
ea
o
f
e
y
e
to
r
ed
u
ce
t
h
e
p
r
o
ce
s
s
in
g
ti
m
e
an
d
er
r
o
r
s
i
n
e
y
e
d
etec
tio
n
.
Fo
r
class
i
f
icat
io
n
,
w
e
als
o
p
r
o
p
o
s
ed
m
eth
o
d
b
y
u
s
in
g
s
i
m
p
l
y
n
ea
r
est
d
is
ta
n
ce
co
m
p
ar
ed
h
a
ar
w
it
h
ad
ab
o
o
s
t.
I
t
is
also
co
n
s
id
er
in
g
to
u
s
e
Op
en
C
V
w
it
h
n
o
t
h
i
n
g
c
h
an
g
e
o
f
s
ca
n
i
n
g
w
in
d
o
w
s
r
e
g
io
n
.
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
Af
ter
s
u
cc
e
s
s
f
u
ll
y
d
etec
t
o
b
j
e
cts
i
n
t
h
e
f
ac
ial
ar
ea
,
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
is
p
lacin
g
a
s
lid
i
n
g
w
i
n
d
o
w
in
th
e
m
id
d
le
ar
ea
o
f
t
h
e
f
ac
e
i
m
a
g
e,
n
o
t
f
r
o
m
t
h
e
to
p
le
f
t
co
r
n
er
as
w
as
d
o
n
e
b
y
t
h
e
co
n
v
e
n
tio
n
al
h
aa
r
ca
s
ca
d
e.
P
r
ev
io
u
s
l
y
,
i
m
a
g
es
h
av
e
b
ee
n
th
r
o
u
g
h
g
r
a
y
s
ca
li
n
g
p
r
o
ce
s
s
a
n
d
c
h
an
g
e
t
h
e
p
ix
el
v
al
u
e
i
n
to
th
e
in
te
g
r
al
i
m
a
g
e,
d
eter
m
in
atio
n
R
OI
o
f
e
y
e,
th
a
n
i
n
t
h
is
ar
ea
p
er
f
o
r
m
ed
ca
lcu
lat
io
n
to
g
e
t
f
e
atu
r
es
v
alu
e
o
f
e
y
e.
I
n
th
i
s
ca
s
e
w
e
a
w
ar
e
th
at
t
h
e
r
e
ar
e
s
o
m
e
f
ac
to
r
s
ca
n
d
is
t
u
r
b
in
g
t
h
e
d
etec
tio
n
o
f
o
b
j
ec
ts
s
u
ch
a
s
li
g
h
tin
g
,
s
o
th
at
in
tes
tin
g
p
h
ase,
t
h
e
i
m
a
g
es
h
av
e
g
o
o
d
lig
h
ti
n
g
c
o
n
d
itio
n
s
an
d
tak
e
n
p
ictu
r
es
d
u
r
i
n
g
i
n
th
e
m
o
r
n
in
g
o
r
in
th
e
af
ter
n
o
o
n
.
T
o
ea
s
e
th
e
u
n
d
er
s
tan
d
i
n
g
o
f
e
y
e
d
etec
tio
n
p
r
o
ce
s
s
in
g
e
n
er
al,
d
iag
r
a
m
b
lo
ck
in
F
ig
u
r
e
1
r
ep
r
esen
ts
th
e
m
et
h
o
d
th
at
w
e
p
r
o
p
o
s
ed
.
Fig
u
r
e
1
.
E
y
e
Dete
c
tio
n
B
lo
ck
Diag
r
a
m
o
f
P
r
o
p
o
s
ed
Sy
s
te
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
I
ma
g
e
P
r
o
ce
s
s
in
g
fo
r
R
a
p
i
d
ly
E
ye
Dete
ctio
n
B
a
s
ed
o
n
R
o
b
u
s
t H
a
a
r
S
lid
in
g
W
in
d
o
w
(
F
it
r
i
Uta
min
in
g
r
u
m)
825
2
.
1
.
Sca
lin
g
Scalin
g
p
r
o
ce
s
s
i
s
a
w
a
y
to
r
e
s
ize
a
d
ig
ital
i
m
a
g
e,
it
i
s
n
ec
e
s
s
ar
y
to
all
i
m
a
g
es
w
h
ic
h
ar
e
p
r
o
ce
s
s
ed
h
av
e
t
h
e
s
a
m
e
s
ize.
I
n
t
h
is
s
y
s
te
m
t
h
e
s
ca
li
n
g
p
r
o
ce
s
s
p
er
f
o
r
m
ed
b
y
u
tili
zi
n
g
th
e
s
ca
li
n
g
p
ac
k
ag
e
p
r
o
v
id
ed
o
n
J
av
a.
2
.
2
.
G
ra
y
s
ca
lin
g
I
m
a
g
e
I
n
p
r
ep
r
o
ce
s
s
in
g
p
h
ase,
w
e
n
e
ed
to
r
em
o
d
el
co
lo
r
im
ag
e
i
n
t
o
g
r
ay
s
ca
le
i
m
ag
e,
w
h
ic
h
ap
p
lied
in
h
aa
r
m
et
h
o
d
as
w
ell
[
1
0
]
.
So
th
at,
it
n
ee
d
to
tr
an
s
f
o
r
m
f
r
o
m
R
G
B
im
a
g
e
in
to
g
r
a
y
s
ca
le
i
m
a
g
e
.
T
o
c
o
n
v
er
t
i
m
a
g
e
in
to
g
r
a
y
s
ca
le,
a
ve
r
a
g
e
m
et
h
o
d
is
u
tili
ze
d
,
b
y
ad
d
in
g
u
p
all
th
e
v
al
u
e
o
f
R
G
B
,
th
en
d
iv
id
ed
b
y
3
,
in
o
r
d
e
r
to
o
b
tain
an
av
er
a
g
e
v
al
u
e
o
f
R
G
B
,
th
e
av
er
ag
e
v
al
u
e
o
f
t
h
at
ca
n
b
e
s
aid
as g
r
a
y
s
ca
le.
Gr
a
y
s
ca
le
=
(
R
+
G
+
B
)
/ 3
(
1
)
2
.
3
.
I
nte
g
ra
l I
m
a
g
e
I
n
teg
r
al
i
m
a
g
e
i
s
a
tech
n
iq
u
e
u
s
ed
i
n
th
e
h
aa
r
m
et
h
o
d
f
o
r
ca
lcu
lat
in
g
t
h
e
v
al
u
e
o
f
th
e
r
ec
tan
g
le
f
ea
t
u
r
es
q
u
ic
k
l
y
b
y
c
h
a
n
g
in
g
th
e
v
al
u
e
o
f
ea
c
h
p
ix
el
i
n
th
e
g
r
a
y
s
ca
le
i
m
a
g
e.
T
h
en
g
e
n
er
a
tin
g
v
al
u
e
s
u
m
m
ed
ar
ea
tab
le
s
o
th
at
th
e
r
esu
lts
r
eg
ar
d
ed
as
a
r
e
p
r
esen
tatio
n
o
f
a
n
e
w
i
m
a
g
e.
T
h
is
m
eth
o
d
h
a
s
th
e
ad
v
an
ta
g
e
th
at
r
elativ
el
y
f
aster
co
m
p
u
t
in
g
,
b
ec
au
s
e
it
d
ep
en
d
s
o
n
t
h
e
n
u
m
b
er
o
f
p
i
x
els
i
n
a
s
q
u
ar
e
i
n
s
tead
o
f
ea
c
h
p
i
x
el
va
lu
e
o
f
an
i
m
a
g
e.
Fi
g
u
r
e
2
m
a
y
ca
n
i
m
p
r
o
v
e
th
e
i
n
te
g
r
al
ca
lcu
latio
n
.
(
a)
(
b
)
Fig
u
r
e
2
.
T
h
e
ca
lcu
latio
n
r
esu
l
t o
f
r
ec
tan
g
le
f
ea
t
u
r
e
h
a
v
e
t
h
e
s
a
m
e
v
al
u
e;
(
a)
o
r
ig
in
al
i
m
a
g
e
an
d
(
b
)
r
e
p
r
esen
t
th
e
in
teg
r
al
i
m
a
g
e
T
h
e
r
eg
io
n
r
es
u
lt
v
al
u
e
o
f
o
r
i
g
in
a
l
i
m
ag
e
s
h
o
w
s
i
n
F
ig
u
r
e
2
(
a)
:
5
+2
+
1
+5
+2
+
2
=
1
7
.
B
y
u
s
i
n
g
t
h
at
ca
lcu
latio
n
,
it
n
ee
d
s
f
i
v
e
s
tep
s
ca
lc
u
latio
n
.
I
n
Fig
u
r
e
2
(
b
)
,
r
eg
io
n
r
es
u
lt
v
a
lu
e
o
f
i
n
te
g
r
a
l
i
m
a
g
e:
1
7
-
0
-
0
+0
=1
7
,
s
o
it
is
o
n
l
y
n
ee
d
f
o
u
r
s
t
ep
s
ca
lcu
la
tio
n
.
T
h
is
e
x
p
lain
s
th
at
th
e
ca
lcu
latio
n
o
f
th
e
in
te
g
r
al
i
m
a
g
e
r
eq
u
ir
es
f
e
w
er
s
tep
s
th
a
n
t
h
e
ca
lc
u
latio
n
o
f
ea
c
h
p
ix
e
l
in
a
r
eg
io
n
.
T
h
e
ca
lcu
latio
n
o
f
i
n
te
g
r
al
i
m
ag
e
p
er
f
o
r
m
s
f
aster
i
n
lar
g
e
-
s
ized
i
m
a
g
e
r
ath
er
t
h
an
i
n
s
m
all
-
s
ized
i
m
a
g
e.
2
.
4
.
T
ra
ini
ng
P
ha
s
e
f
o
r
E
y
e
F
ea
t
ure
T
o
d
etec
t
e
y
e
f
ea
tu
r
e
o
n
f
ac
ia
l
ar
ea
,
it
is
n
ec
e
s
s
ar
y
to
tr
ain
o
n
m
a
n
y
e
y
e
i
m
ag
e
s
d
ataset
b
y
v
ar
io
u
s
f
o
r
m
s
o
r
t
y
p
es
o
f
a
p
er
s
o
n
'
s
e
y
es
[
1
1
]
.
T
h
e
d
ataset
c
o
n
tain
i
n
g
e
y
e
i
m
a
g
es
w
it
h
a
lo
o
k
a
h
ea
d
(
f
r
o
n
tal)
a
n
d
t
h
e
v
ar
iatio
n
o
f
t
h
e
le
f
t
a
n
d
r
ig
h
t
v
ie
w
s
o
b
tain
e
d
at
r
an
d
o
m
f
r
o
m
p
eo
p
le
ar
o
u
n
d
t
h
e
r
esear
ch
.
T
r
ain
in
g
p
h
a
s
e
i
s
d
o
n
e
b
y
t
h
e
v
io
la
J
o
n
e
s
et
al
u
s
in
g
h
aa
r
te
m
p
late.
T
h
e
w
i
n
d
o
w
s
ize
u
s
ed
w
as
9
2
x
1
9
p
ix
el
s
i
n
ac
co
r
d
an
ce
b
y
th
e
s
ize
o
f
t
h
e
d
ataset
o
f
e
y
e,
w
h
ile
t
h
e
s
ize
o
f
t
h
e
b
o
x
f
ea
tu
r
e
is
th
e
4
x
3
.
Fig
u
r
e
3
w
il
l c
l
ar
if
y
e
y
e
d
ataset.
Fig
u
r
e
3
.
So
m
e
tr
ain
i
n
g
s
a
m
p
l
e
o
f
e
y
e
w
it
h
s
o
m
e
co
n
d
itio
n
o
f
lig
h
ti
n
g
an
d
s
k
in
co
lo
r
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.
7
,
No
.
2
,
A
p
r
il
2
0
1
7
:
8
2
3
–
8
3
0
826
2
.
5
.
Dire
ct
io
n o
f
Sli
din
g
Win
do
w
I
n
t
h
is
p
ap
er
,
w
in
d
o
w
r
e
g
io
n
i
s
p
lace
d
o
n
t
h
e
to
p
ce
n
ter
ar
e
a
o
f
th
e
f
ac
e
i
m
a
g
e
s
h
o
w
in
F
ig
u
r
e
4
(
b
)
.
T
h
e
s
ize
o
f
t
h
e
w
i
n
d
o
w
r
eg
io
n
t
h
at
u
s
e
i
s
s
a
m
e
a
s
t
h
e
s
ize
o
f
th
e
w
in
d
o
w
t
h
at
i
s
u
s
ed
d
u
r
in
g
th
e
p
r
o
ce
s
s
o
f
tr
ain
i
n
g
d
ata
o
f
e
y
e
r
eg
io
n
.
T
h
e
s
ize
t
h
at
w
e
u
s
e
is
9
2
x
1
9
p
ix
els,
it
is
r
ec
ta
n
g
u
lar
a
n
d
in
ac
co
r
d
an
ce
w
i
th
a
w
i
n
d
o
w
th
at
w
o
u
ld
in
d
icate
t
h
at
th
e
ar
ea
co
n
tain
ed
o
f
e
y
e
r
eg
io
n
.
(
a)
(
b
)
Fig
u
r
e
4
.
I
llu
s
tr
atio
n
o
f
W
in
d
o
w
R
e
g
io
n
,
(
a)
T
h
e
w
i
n
d
o
w
r
e
g
io
n
o
f
co
n
v
e
n
tio
n
al
h
aa
r
ca
s
c
ad
e
s
ca
n
th
e
i
m
a
g
e
s
tar
t f
r
o
m
t
h
e
le
f
t to
p
co
r
n
er
,
(
b
)
Ou
r
p
r
o
p
o
s
ed
m
et
h
o
d
s
tar
t scan
n
i
n
g
t
h
e
i
m
ag
e
f
r
o
m
to
p
ce
n
ter
.
T
o
d
eter
m
i
n
e
th
e
m
id
p
o
in
t
o
f
th
e
s
tar
ti
n
g
o
f
s
ca
n
n
i
n
g
w
in
d
o
w
,
it
is
n
ec
e
s
s
ar
y
to
k
n
o
w
t
h
e
w
id
t
h
o
f
th
e
p
ix
els
i
n
f
ac
ial
ar
ea
an
d
th
e
s
ize
o
f
th
e
b
o
x
w
i
n
d
o
w
u
s
ed
d
u
e
to
th
e
d
eter
m
in
a
tio
n
o
f
th
e
m
id
p
o
in
t
d
er
iv
ed
f
r
o
m
t
h
e
w
id
t
h
o
f
th
e
i
m
a
g
e.
I
n
th
is
ca
s
e
w
e
u
s
e
3
6
0
x
2
4
0
p
ix
els
o
f
t
h
e
i
m
ag
e
s
ize
o
r
t
h
e
o
th
er
w
is
e,
w
h
ile
th
e
s
ize
o
f
th
e
b
o
x
w
i
n
d
o
w
h
aa
r
is
9
2
x
1
9
p
ix
el.
Fig
u
r
e
5
illu
s
tr
ates
th
e
m
et
h
o
d
o
f
d
eter
m
i
n
in
g
t
h
e
s
tar
tin
g
p
o
in
t o
f
s
ca
n
n
i
n
g
w
i
n
d
o
w
.
Fig
u
r
e
5
.
Dete
r
m
in
e
t
h
e
p
o
s
iti
o
n
o
f
th
e
w
i
n
d
o
w
r
e
g
io
n
m
id
p
o
in
t o
f
t
h
e
in
X
p
o
in
t a
n
d
Y
p
o
in
t.
(
X
n
,
Yo
)
= X
–
;
Yo
(
2
)
I
m
ag
e
h
as
w
id
th
(W
-
i
m
a
g
e)
co
n
tain
i
n
g
a
n
u
m
b
er
o
f
p
ix
el,
in
w
h
ic
h
p
ix
el
(
0
,
0
)
lo
ca
ted
in
th
e
to
p
lef
t
co
r
n
er
o
f
a
n
i
m
a
g
e.
T
o
d
eter
m
i
n
e
t
h
e
p
i
v
o
t
p
o
in
t
o
f
r
ec
tan
g
le
w
i
n
d
o
w
,
i
m
a
g
e
w
id
t
h
i
s
m
in
u
s
b
y
w
id
t
h
o
f
t
h
e
r
ec
tan
g
le
w
i
n
d
o
w
(
W
-
w
i
n
d
o
w
)
th
e
n
d
i
v
id
ed
b
y
t
w
o
.
So
it
f
i
g
u
r
e
s
o
u
t
t
h
e
co
o
r
d
in
at
es
f
o
r
(
Xn
,
0
)
a
s
a
r
ef
er
en
ce
p
o
in
t
o
f
r
ec
tan
g
le
w
i
n
d
o
w
.
Fo
r
in
ca
s
e
o
f
i
m
a
g
e
w
it
h
3
6
0
p
ix
el
w
id
th
an
d
9
2
p
ix
el
w
id
t
h
o
f
r
ec
tan
g
le
w
i
n
d
o
w
,
x
p
o
in
t
ca
lcu
lated
f
r
o
m
X
(
3
6
0
-
9
2
)
/
2
=
1
3
4
.
R
ec
tan
g
le
w
i
n
d
o
w
s
t
ar
tin
g
f
r
o
m
t
h
e
d
o
t
p
ix
els (
1
3
4
;0
)
u
n
til t
h
e
w
id
th
o
f
th
e
w
i
n
d
o
w
r
e
g
io
n
e
n
d
s
(
2
2
6
;0
)
w
i
th
a
w
i
n
d
o
w
h
ei
g
h
t th
a
t a
d
j
u
s
ted
.
Dete
ctio
n
p
r
o
ce
s
s
o
f
e
y
e
r
eg
i
o
n
is
o
b
tain
ed
f
r
o
m
th
e
d
i
f
f
er
en
ce
b
et
w
ee
n
th
e
v
al
u
es
o
f
ea
ch
f
ea
t
u
r
e
o
n
t
h
e
te
s
tin
g
d
ata
w
it
h
t
h
e
v
a
lu
es
o
f
ea
c
h
f
ea
t
u
r
e
o
n
t
h
e
tr
ai
n
in
g
d
ata
i
n
a
w
in
d
o
w
r
ec
ta
n
g
u
lar
ar
ea
,
a
n
d
th
e
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
I
ma
g
e
P
r
o
ce
s
s
in
g
fo
r
R
a
p
i
d
ly
E
ye
Dete
ctio
n
B
a
s
ed
o
n
R
o
b
u
s
t H
a
a
r
S
lid
in
g
W
in
d
o
w
(
F
it
r
i
Uta
min
in
g
r
u
m)
827
s
u
m
m
ed
to
g
e
n
er
ate
a
s
i
n
g
le
v
alu
e
f
ea
t
u
r
es
a
n
d
co
m
p
ar
ed
w
h
et
h
er
th
e
v
al
u
e
o
f
th
e
s
e
f
e
atu
r
es
i
s
les
s
t
h
an
a
s
p
ec
if
ied
t
h
r
es
h
o
ld
v
a
lu
e,
ill
u
s
tr
ated
in
Fi
g
u
r
e
6
.
T
h
e
f
ea
t
u
r
es
ar
e
o
b
tain
ed
f
r
o
m
b
lack
an
d
w
h
ite
s
q
u
ar
e
s
o
f
h
aa
r
f
ea
t
u
r
es.
V
a
lu
e
(
V
)
= su
m|
Tes
t
ij
–
Tr
a
in
ij
|
(
3
)
T
esti
ij
is
a
p
ix
el
v
al
u
e
o
f
tes
t
in
g
d
ata
f
ea
tu
r
es
at
p
o
s
itio
n
i
,
j
,
T
r
ain
ij
is
a
p
ix
el
v
al
u
e
o
f
tr
ain
d
ata
f
ea
t
u
r
es a
t p
o
s
itio
n
i,
j
.
Fig
u
r
e
6
.
W
in
d
o
w
r
eg
io
n
s
lid
i
n
g
d
o
w
n
a
n
d
g
e
n
er
ate
s
o
m
e
f
e
atu
r
e
o
f
e
y
e,
ca
lc
u
late
a
n
d
co
m
p
ar
ed
ea
ch
d
ata
w
it
h
lear
n
in
g
d
ata
W
in
d
o
w
b
o
x
e
s
s
h
i
f
ted
e
v
er
y
s
in
g
le
p
i
x
el
i
n
th
e
i
m
a
g
e
o
f
t
h
e
p
ictu
r
e.
T
h
en
t
h
r
o
u
g
h
h
aa
r
f
ea
tu
r
es,
i
t
r
eq
u
ir
es
to
ex
a
m
in
e
w
h
et
h
er
t
h
e
i
m
a
g
e
ar
ea
in
a
w
i
n
d
o
w
b
o
x
d
etec
ted
as
e
y
e
ar
ea
o
r
n
o
t.
I
f
th
e
v
a
lu
e
i
s
less
th
an
t
h
e
th
r
es
h
o
ld
,
th
e
n
th
e
win
d
o
w
b
o
x
w
ill
co
n
ti
n
u
e
to
r
u
n
u
n
til
it
s
to
p
s
.
I
n
th
i
s
s
y
s
te
m
w
e
h
a
v
e
m
o
d
if
ied
th
e
r
u
les
i
n
d
etec
tin
g
e
y
e
r
eg
i
o
n
.
I
f
th
e
d
if
f
er
en
ce
i
n
v
al
u
e
f
ea
tu
r
es
d
o
n
o
t
h
av
e
an
y
v
al
u
e
w
h
ic
h
less
t
h
an
t
h
e
s
p
ec
if
ied
th
r
es
h
o
ld
th
en
it
h
a
s
to
f
in
d
o
u
t
th
e
s
m
alle
s
t
d
if
f
er
en
ce
v
al
u
e.
So
,
t
h
e
v
al
u
e
is
d
etec
ted
as
a
n
ar
ea
o
f
e
y
e.
3.
E
XP
E
R
I
M
E
NT
A
L
RE
SUL
T
AND
DI
SCU
SS
I
O
N
T
esti
n
g
d
ata
w
as
ta
k
en
f
r
o
m
t
h
e
p
eo
p
le
ar
o
u
n
d
t
h
e
u
n
iv
er
s
i
t
y
i
n
Ma
la
n
g
cit
y
.
T
h
e
d
ataset
ca
p
tu
r
ed
in
b
r
i
g
h
t
li
g
h
ti
n
g
co
n
d
itio
n
s
o
n
l
y
a
n
d
w
e
co
llec
ted
3
0
d
atas
et
r
etr
iev
ed
u
s
in
g
Ma
cB
o
o
k
A
ir
w
it
h
r
eso
l
u
tio
n
9
0
0
p
ix
els
an
d
also
f
r
o
m
m
o
b
ile
d
ev
ice
ca
m
er
a
A
s
u
s
Z
en
f
o
n
e
Self
ie
w
it
h
r
eso
lu
tio
n
1
9
2
0
x
1
0
8
0
p
ix
els
.
A
s
a
li
m
ita
tio
n
i
n
t
h
is
s
y
s
te
m
,
th
e
f
ac
ial
i
m
a
g
e
p
o
s
itio
n
s
h
o
u
ld
b
e
in
t
h
e
m
id
d
le.
W
e
also
ad
d
s
ev
er
al
co
n
d
itio
n
s
o
f
i
m
a
g
e
te
s
ti
n
g
to
b
e
te
s
ted
i
n
a
s
y
s
te
m
th
at
h
as
b
ee
n
cr
ea
te
d
.
I
n
t
h
is
s
ec
t
io
n
,
w
e
d
id
s
o
m
e
e
x
p
er
i
m
e
n
t
s
f
o
r
d
etec
tin
g
e
y
e
ar
ea
w
it
h
s
ca
li
n
g
i
m
a
g
e
s
ize
i
n
to
ar
o
u
n
d
o
f
3
6
0
x
2
4
0
p
ix
els
an
d
3
x
4
p
i
x
el
f
o
r
s
ize
o
f
te
m
p
late
h
aa
r
f
ea
t
u
r
e.
T
h
is
ex
p
er
im
e
n
t
i
m
p
le
m
e
n
ted
o
n
I
n
tel
C
o
r
e
i5
-
5
2
0
0
U
p
r
o
ce
s
s
o
r
2
,
7
GHz
u
s
in
g
J
av
a
Netb
ea
n
s
8
.
1
.
T
h
e
f
o
llo
w
i
n
g
Fig
u
r
e
7
s
h
o
w
s
t
h
e
r
es
u
lts
o
f
o
u
r
p
r
o
p
o
s
e
d
m
et
h
o
d
.
(
a)
(
b
)
(
c)
Fig
u
r
e
7
.
E
x
p
er
i
m
en
tal
r
esu
lt
f
r
o
m
d
if
f
er
e
n
t c
o
n
d
itio
n
: (
a)
E
y
e
d
etec
tio
n
r
es
u
lt
w
it
h
d
if
f
er
e
n
t s
k
i
n
t
y
p
e;
(
b
)
E
y
e
d
etec
tio
n
r
es
u
lt
w
it
h
d
if
f
er
en
t b
ac
k
g
r
o
u
n
d
; (
c)
E
y
e
d
etec
tio
n
r
esu
lt
w
it
h
d
if
f
er
en
t o
f
e
y
e
g
az
e
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.
7
,
No
.
2
,
A
p
r
il
2
0
1
7
:
8
2
3
–
8
3
0
828
E
x
p
er
i
m
e
n
tal
r
es
u
lt
o
b
tain
ed
f
r
o
m
ca
lcu
latio
n
u
s
i
n
g
P
er
ce
n
tag
e
o
f
S
u
cc
ess
f
u
l
P
r
ed
icti
o
n
(
P
SP
)
.
A
cc
u
r
ac
y
ca
lc
u
latio
n
i
s
i
m
p
o
r
tan
t
to
m
ea
s
u
r
e
th
e
ab
ilit
y
o
f
a
s
y
s
te
m
t
h
at
h
a
s
b
ee
n
m
ad
e
in
r
ec
o
g
n
izin
g
o
b
j
ec
ts
.
I
n
ex
p
er
i
m
en
tal,
it
r
e
ac
h
s
7
1
.
4
%
d
etec
tio
n
r
ate
a
n
d
1
1
.
1
%
f
alse
r
ate
a
s
s
h
o
w
n
i
n
T
ab
le
1
th
at
b
ased
o
n
th
e
f
o
r
m
u
la
4
.
(
4
)
TP
=
p
er
s
o
n
/ f
ac
ial
i
m
ag
e
a
n
d
s
u
cc
ee
d
ed
to
r
ec
o
g
n
ize
th
e
e
y
e.
TN
=
i
m
ag
e
i
s
n
o
t t
h
e
e
y
e
ar
ea
,
an
d
d
etec
ted
as a
n
ey
e.
(
5
)
T
ab
le
1
.
E
y
e
Dete
ctio
n
R
es
u
lt
o
f
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
M
e
t
h
o
d
D
e
t
e
c
t
i
o
n
R
a
t
e
F
a
l
s
e
R
a
t
e
P
r
o
c
e
ssi
n
g
T
i
me
H
a
a
r
C
a
sc
a
d
e
+
A
d
a
B
o
o
st
6
9
,
2
%
2
5
%
3
,
4
s
P
r
o
p
o
se
d
M
e
t
h
o
d
7
1
,
4
%
1
1
,
1
%
2
,
9
s
As
w
e
ca
n
s
ee
,
t
h
e
e
x
p
er
i
m
en
t
r
es
u
lt
f
r
o
m
T
ab
le
1
,
th
e
p
r
o
p
o
s
ed
m
eth
o
d
co
u
ld
i
m
p
r
o
v
e
th
e
d
etec
tio
n
r
ate
an
d
f
al
s
e
r
ate
b
y
c
h
a
n
g
i
n
g
th
e
d
ir
ec
tio
n
o
f
s
lid
in
g
w
i
n
d
o
w
.
W
h
i
le,
th
e
h
aa
r
ca
s
ca
d
e
m
et
h
o
d
ac
h
iev
e
s
les
s
d
etec
tio
n
r
ate
a
n
d
h
ig
h
er
f
al
s
e
r
ate.
Mo
r
eo
v
er
,
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
h
a
v
e
f
a
s
t
er
p
r
o
ce
s
s
in
g
ti
m
e
th
an
h
aa
r
ca
s
ca
d
e
m
et
h
o
d
.
I
n
t
h
is
p
ap
er
,
w
e
also
ap
p
lied
h
aa
r
ca
s
ca
d
e
m
e
th
o
d
w
h
ic
h
g
en
er
ated
m
a
n
u
a
ll
y
f
o
r
co
m
p
ar
in
g
w
it
h
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
.
T
o
o
b
tain
t
h
e
co
m
p
ar
ab
le
r
es
u
lt
s
,
test
i
n
g
p
r
o
ce
s
s
w
a
s
co
n
d
u
c
ted
w
ith
t
h
e
s
a
m
e
d
atab
ase.
T
h
e
co
m
p
ar
is
o
n
r
es
u
lt
alr
ea
d
y
p
r
esen
ted
in
T
ab
le
2
.
T
h
e
s
y
s
te
m
w
a
s
m
ea
s
u
r
ed
q
u
an
tita
tiv
e
l
y
u
s
i
n
g
th
at
h
a
s
b
ee
n
d
o
n
e
b
y
t
h
e
eq
u
atio
n
s
o
f
ac
c
u
r
ac
y
[
1
2
]
,
[
1
3
]
i
n
f
o
r
m
u
la
5
.
T
h
e
ac
c
u
r
ac
y
o
f
p
r
o
p
o
s
ed
m
et
h
o
d
is
9
3
.
3
%
an
d
co
n
v
e
n
tio
n
al
h
aa
r
ca
s
ca
d
e
m
et
h
o
d
is
8
6
.
6
7
%.
Fro
m
t
h
e
Fi
g
u
r
e
8
(
a)
,
w
e
ca
n
s
ee
t
h
e
d
if
f
er
e
n
ce
th
at
co
n
v
en
t
io
n
al
h
aa
r
ca
s
ca
d
e
m
e
th
o
d
m
i
s
s
ed
s
o
m
e
o
f
e
y
e
d
etec
tio
n
,
an
d
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
ca
n
i
m
p
r
o
v
e
t
h
e
e
y
e
d
etec
tio
n
.
(
a)
(
b
)
Fig
u
r
e
8
.
So
m
e
o
f
e
y
e
d
etec
tio
n
r
esu
lt f
r
o
m
co
m
p
ar
i
s
o
n
m
et
h
o
d
:
(
a)
Haa
r
ca
s
ca
d
e
m
eth
o
d
ca
n
n
o
t d
etec
t
th
e
f
o
u
r
th
a
n
d
s
i
x
th
i
m
a
g
e;
(
b
)
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
ca
n
f
ix
th
e
e
y
e
d
etec
tio
n
T
ab
le
2
.
C
o
m
p
ar
is
o
n
r
es
u
lt
u
s
in
g
Haa
r
C
ascad
e
m
e
th
o
d
an
d
p
r
o
p
o
s
ed
m
et
h
o
d
M
e
t
h
o
d
A
c
c
u
r
a
c
y
H
a
a
r
+
A
d
a
B
o
o
st
9
0
,
8
%
P
r
o
p
o
se
d
M
e
t
h
o
d
9
2
,
4
%
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2088
-
8708
I
ma
g
e
P
r
o
ce
s
s
in
g
fo
r
R
a
p
i
d
ly
E
ye
Dete
ctio
n
B
a
s
ed
o
n
R
o
b
u
s
t H
a
a
r
S
lid
in
g
W
in
d
o
w
(
F
it
r
i
Uta
min
in
g
r
u
m)
829
4.
CO
NCLU
SI
O
N
Min
i
m
izi
n
g
er
r
o
r
s
an
d
s
p
ee
d
i
n
g
u
p
t
h
e
d
etec
tio
n
p
r
o
ce
s
s
ca
n
b
e
h
a
n
d
led
u
s
i
n
g
t
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
in
t
h
is
p
ap
er
.
So
m
e
o
f
test
in
g
i
m
a
g
es
ca
n
n
o
t
b
e
d
etec
ted
b
y
u
s
i
n
g
h
aa
r
co
n
v
en
tio
n
al
m
e
th
o
d
s
,
f
o
r
i
n
s
ta
n
ce
,
s
ev
er
al
i
m
ag
e
s
w
it
h
a
d
if
f
er
en
t
e
y
e
g
az
e
s
o
m
eti
m
es
al
s
o
ca
n
n
o
t
b
e
d
etec
ted
.
B
y
u
s
in
g
th
e
s
c
h
e
m
e
o
f
ch
an
g
i
n
g
t
h
e
p
o
s
itio
n
o
f
t
h
e
s
ca
n
n
i
n
g
w
i
n
d
o
w
in
th
e
ar
e
a
o
f
d
etec
ted
f
ac
e,
th
e
d
etec
tio
n
o
f
e
y
e
ca
n
b
e
i
m
p
r
o
v
ed
ef
f
ec
tiv
e
l
y
,
as
w
ell
as
th
e
co
m
p
u
tatio
n
ti
m
e
ca
n
b
e
im
p
r
o
v
ed
.
T
h
e
d
r
aw
b
ac
k
o
f
th
i
s
p
r
o
p
o
s
e
d
m
et
h
o
d
is
co
m
e
u
p
w
it
h
an
y
d
if
f
er
e
n
t
lig
h
ti
n
g
co
n
d
itio
n
o
f
ca
p
tu
r
ed
im
a
g
e.
T
h
u
s
f
o
r
f
u
r
t
h
er
r
esear
ch
,
it
w
il
l
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
an
ce
o
f
i
m
ag
e
i
n
d
i
f
f
er
e
n
t li
g
h
ti
n
g
co
n
d
itio
n
i
n
o
r
d
er
to
r
u
n
m
o
r
e
q
u
i
ck
l
y
an
d
ac
c
u
r
atel
y
.
B
esid
es th
at,
t
h
e
ap
p
r
o
ac
h
o
f
th
e
an
g
le
o
f
t
h
e
f
ac
e
ca
n
also
b
e
p
er
f
o
r
m
ed
.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
is
w
o
r
k
w
as
p
er
f
o
r
m
ed
i
n
C
o
m
p
u
ter
R
e
s
ea
r
ch
Gr
o
u
p
,
F
ac
u
lt
y
o
f
C
o
m
p
u
ter
Scie
n
ce
,
B
r
a
w
ij
ay
a
Un
i
v
er
s
it
y
.
I
t
w
a
s
s
u
p
p
o
r
ted
b
y
Un
g
g
u
lan
R
esear
c
h
G
r
an
t
Facu
lt
y
o
f
C
o
m
p
u
ter
Scien
ce
,
B
r
a
w
ij
a
y
a
Un
i
v
er
s
it
y
r
esear
ch
f
u
n
d
.
W
e
also
w
o
u
ld
lik
e
to
t
h
an
k
f
o
r
p
eo
p
le
w
h
o
k
ee
n
to
p
r
o
v
id
e
p
ictu
r
es a
s
d
ataset.
RE
F
E
R
E
NC
E
S
[1
]
M
.
U.
G
h
a
n
i,
e
t
a
l.
,
“
Ga
ze
P
o
in
te
r:
A
re
a
l
ti
me
mo
u
se
p
o
i
n
ter
c
o
n
t
ro
l
imp
lem
e
n
ta
ti
o
n
b
a
se
d
o
n
e
y
e
g
a
ze
tra
c
k
in
g
”
2
0
1
3
1
6
t
h
I
n
t.
M
u
lt
i
T
o
p
.
Co
n
f
.
I
NMIC
2
0
1
3
,
p
p
.
1
5
4
–
1
5
9
,
2
0
1
3
.
[2
]
C.
G
a
o
a
n
d
S
.
L
.
L
u
,
“
No
v
e
l
FP
GA
b
a
se
d
h
a
a
r
c
l
a
ss
if
ier
fa
c
e
d
e
tec
ti
o
n
a
lg
o
rith
m
a
c
c
e
ler
a
ti
o
n
”
P
ro
c
.
-
2
0
0
8
I
n
t
.
Co
n
f
.
F
.
P
ro
g
ra
m
.
L
o
g
.
A
p
p
l.
F
P
L
,
p
p
.
3
7
3
–
3
7
8
,
2
0
0
8
.
[3
]
M
.
S
in
g
h
,
e
t
a
l.
,
“
Ey
e
m
o
v
e
m
e
n
t
d
e
tec
ti
o
n
f
o
r
w
h
e
e
lch
a
ir
c
o
n
tro
l
a
p
p
li
c
a
ti
o
n
”
In
t.
Co
n
f.
E
lec
tr.
El
e
c
tro
n
.
S
i
g
n
a
ls
,
Co
mm
u
n
.
Op
ti
m.
EE
S
CO
2
0
1
5
,
p
p
.
4
–
8
,
2
0
1
5
.
[4
]
R.
C.
Co
e
tze
r
a
n
d
G
.
P
.
Ha
n
c
k
e
,
“
E
y
e
d
e
tec
ti
o
n
f
o
r
a
re
a
l
-
ti
m
e
v
e
h
icle
d
riv
e
r
fa
t
ig
u
e
m
o
n
it
o
rin
g
sy
ste
m
,
”
n
o
.
I
V
,
2
0
1
1
.
[5
]
S
.
G
o
y
a
l
a
n
d
G
.
K.
G
o
y
a
l,
“
C
a
s
c
a
d
e
M
o
d
e
ll
in
g
f
o
r
P
re
d
icti
n
g
S
o
lu
b
il
it
y
In
d
e
x
o
f
Ro
ll
e
r
Dri
e
d
G
o
a
t
w
h
o
le
M
il
k
P
o
w
d
e
r,
”
v
o
l
/i
ss
u
e
:
2
(
1
)
,
p
p
.
2
3
–
2
8
,
2
0
1
3
.
[6
]
S
.
Zh
a
n
g
,
e
t
a
l
.
,
“
In
f
o
rm
e
d
Ha
a
r
-
li
k
e
F
e
a
tu
re
s I
m
p
ro
v
e
P
e
d
e
strian
De
tec
ti
o
n
,
”
Cv
p
r
,
2
0
1
4
.
[7
]
D.
Zh
u
,
e
t
a
l.
,
“
Hy
b
rid
M
e
th
o
d
f
o
r
Hu
m
a
n
E
y
e
De
te
c
ti
o
n
,
”
p
p
.
5
3
6
8
–
5
3
7
3
,
2
0
1
4
.
[8
]
F
.
Uta
m
in
in
g
ru
m
,
e
t
a
l.
,
“
M
ix
e
d
g
a
u
ss
ian
a
n
d
im
p
u
lse
n
o
ise
re
m
o
v
a
l
b
a
se
d
o
n
k
e
rn
e
l
o
b
se
rv
a
ti
o
n
a
n
d
e
d
g
e
d
irec
ti
o
n
,
”
In
t
.
J
.
I
n
n
o
v
.
Co
m
p
u
t.
In
f.
C
o
n
tr
o
l
,
v
o
l
/i
ss
u
e
:
11
(
5
)
,
p
p
.
1
5
0
9
–
1
5
2
3
,
2
0
1
5
.
[9
]
P
.
V
io
la an
d
M
.
Jo
n
e
s,
“
Ra
p
id
o
b
jec
t
d
e
tec
ti
o
n
u
sin
g
a
b
o
o
s
ted
c
a
s
c
a
d
e
o
f
si
m
p
le fea
tu
re
s,
”
p
re
se
n
te
d
a
t
2
0
0
1
IEE
E
Co
mp
u
ter
S
o
c
iety
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
V
isio
n
a
n
d
Pa
t
ter
n
Rec
o
g
n
it
io
n
,
v
o
l.
1,
p
p
.
I
-
5
1
1
-
I
-
5
1
8
,
2
0
0
1
.
[1
0
]
M
.
M
e
h
r
u
b
e
o
g
lu
,
e
t
a
l.
,
“
Re
a
l
-
T
i
m
e
E
y
e
T
ra
c
k
in
g
Us
in
g
a
S
m
a
rt
Ca
m
e
ra
,
”
p
p
.
1
–
7.
[1
1
]
Y.
M
.
Ch
e
u
n
g
a
n
d
Q.
P
e
n
g
,
“
Ey
e
Ga
z
e
T
ra
c
k
in
g
w
it
h
a
Web
Ca
m
e
ra
in
a
De
s
k
to
p
En
v
ir
o
n
m
e
n
t,
”
IEE
E
T
ra
n
s
.
Hu
ma
n
-
M
a
c
h
in
e
S
y
st.
,
v
o
l
/i
ss
u
e
:
45
(
4
)
,
p
p
.
4
1
9
–
4
3
0
,
2
0
1
5
.
[1
2
]
S.
W
.
K.
b
.
C.
F
.
T
.
W
e
i
-
Ch
a
o
Li
n
a
,
“
CA
N
N:
A
n
in
tru
sio
n
d
e
tec
ti
o
n
sy
ste
m
b
a
s
e
d
o
n
c
o
m
b
in
in
g
c
lu
ste
r
c
e
n
ters
a
n
d
n
e
a
re
st n
e
ig
h
b
o
rs,
”
K
n
o
w
led
g
e
-
Ba
s
e
d
S
y
ste
m
s,
v
o
l.
7
8
,
p
p
.
1
3
-
2
1
,
2
0
1
5
.
[1
3
]
R.
B.
V
a
n
d
R.
K.
S
e
n
a
p
a
ti
,
“
Brig
h
t
L
e
sio
n
De
tec
ti
o
n
in
Co
l
o
r
F
u
n
d
u
s
Im
a
g
e
s
B
a
se
d
o
n
T
e
x
tu
re
F
e
a
tu
re
s,”
v
o
l
/i
ss
u
e
:
5
(
1
)
,
p
p
.
9
2
–
1
0
0
,
2
0
1
6
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Fi
tr
i
Uta
m
i
n
i
n
g
r
u
m
,
w
a
s
b
o
rn
in
S
u
ra
b
a
y
a
,
Eas
t
Ja
v
a
,
In
d
o
n
e
sia
.
S
h
e
re
c
e
iv
e
d
h
e
r
Ba
c
h
e
lo
r
De
g
re
e
in
El
e
c
tri
c
a
l
En
g
in
e
e
ri
n
g
(BEn
g
.
)
f
ro
m
Na
ti
o
n
a
l
In
stit
u
te
o
f
T
e
c
h
n
o
lo
g
y
f
ro
m
Bra
w
ij
a
y
a
Un
iv
e
rsit
y
,
M
a
lan
g
,
In
d
o
n
e
sia
i
n
2
0
0
7
.
In
a
d
d
it
io
n
,
sh
e
o
b
tain
e
d
th
e
d
e
g
re
e
o
f
Do
c
to
r
o
f
En
g
in
e
e
rin
g
in
th
e
f
ield
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
f
ro
m
Ku
m
a
m
o
to
Un
iv
e
rsit
y
,
Ja
p
a
n
.
S
h
e
h
a
s
b
e
e
n
w
o
rk
in
g
a
s
p
a
rt
ti
m
e
lec
tu
re
r
in
se
v
e
ra
l
in
stit
u
t
io
n
,
s
u
c
h
a
s
S
e
k
o
lah
T
in
g
g
i
T
e
k
n
ik
A
n
g
k
a
tan
Da
ra
t
(S
T
TAD)
f
ro
m
2
0
0
6
u
n
t
il
2
0
0
7
.
S
e
k
o
lah
T
in
g
g
i
T
e
k
n
ik
A
tl
a
s
Nu
sa
n
tara
(S
TT
A
R)
f
ro
m
2
0
0
6
u
n
ti
l
2
0
0
7
.
In
a
d
d
it
i
o
n
,
sh
e
a
lso
h
a
s
b
e
e
n
tea
c
h
in
g
th
e
stu
d
e
n
ts
o
f
V
o
c
a
ti
o
n
a
l
E
d
u
c
a
ti
o
n
De
v
e
lo
p
m
e
n
t
Ce
n
ter
(V
EDC)
M
a
lan
g
-
In
d
o
n
e
sia
in
2
0
0
7
a
n
d
M
a
lan
g
Jo
in
t
Ca
m
p
u
s
(M
JC)
in
2
0
0
7
.
S
h
e
h
a
s
b
e
e
n
f
u
ll
ti
m
e
lec
tu
re
r
i
n
Bra
w
ij
a
y
a
Un
iv
e
rsit
y
sta
rti
n
g
f
ro
m
2
0
0
8
.
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.
7
,
No
.
2
,
A
p
r
il
2
0
1
7
:
8
2
3
–
8
3
0
830
Yu
ita
Ar
u
m
S
a
r
i
,
is
c
u
rre
n
tl
y
w
o
rk
in
g
a
t
Lab
o
ra
to
ry
o
f
In
telli
g
e
n
t
S
y
ste
m
in
Bra
w
ij
a
y
a
Un
iv
e
rsit
y
.
S
h
e
o
b
tain
e
d
Ba
c
h
e
l
o
r
De
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Bra
w
ij
a
y
a
Un
iv
e
rsit
y
,
M
a
lan
g
,
In
d
o
n
e
sia
in
2
0
1
1
.
He
r
M
a
s
ter
De
g
re
e
o
b
tain
e
d
f
ro
m
De
p
a
rt
m
e
n
t
o
f
In
f
o
r
m
a
ti
c
s,
In
stit
u
te
T
e
k
n
o
lo
g
i
S
e
p
u
lu
h
No
p
e
m
b
e
r,
S
u
ra
b
a
y
a
,
In
d
o
n
e
sia
in
2
0
1
5
.
H
er
re
se
a
rc
h
a
r
e
in
f
i
e
ld
s
o
f
c
o
m
p
u
ter v
isio
n
,
d
ig
it
a
l
im
a
g
e
p
ro
c
e
ss
in
g
,
a
n
d
p
a
tt
e
rn
re
c
o
g
n
it
i
o
n
.
Re
n
a
ld
i
Pri
m
a
sw
a
r
a
Pra
e
ty
a
,
is
c
u
rre
n
tl
y
a
s
a
m
a
g
ister
stu
d
e
n
t
in
F
a
c
u
lt
y
o
f
Co
m
p
u
ter
S
c
ien
c
e
Bra
w
ij
a
y
a
Un
iv
e
rsit
y
.
S
h
e
o
b
tain
e
d
Ba
c
h
e
l
o
r
De
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Bra
w
ij
a
y
a
Un
iv
e
rsit
y
,
M
a
lan
g
,
I
n
d
o
n
e
sia
in
2
0
1
2
.
No
w
He
is
c
u
rre
n
tl
y
f
in
a
li
z
in
g
h
is
M
a
ste
r
De
g
re
e
in
Co
m
p
u
ter
S
c
ie
n
c
e
Br
a
w
ij
a
y
a
Un
iv
e
r
sit
y
.
His
re
se
a
r
c
h
a
re
in
f
ield
s
o
f
c
o
m
p
u
ter
v
isio
n
,
sp
e
e
c
h
re
c
o
g
n
i
ti
o
n
,
a
n
d
d
i
g
it
a
l
ima
g
e
p
ro
c
e
ss
in
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.