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.
8
,
No
.
3
,
J
u
n
e
201
8
,
p
p
.
1
9
0
3
~
1
9
1
2
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v8
i
3
.
p
p
1
9
0
3
-
1912
1903
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
Co
ntent
-
b
a
sed I
m
a
g
e Re
trieva
l S
y
ste
m
f
o
r
a
n I
m
a
g
e G
a
llery
Sea
rch App
lica
tio
n
Nico
le
T
ha
m
L
ey
M
a
i
1
,
Sy
a
h
m
i
Sy
a
hira
n B
in Ah
m
a
d R
id
zua
n
2
,
Z
a
id B
in O
m
a
r
3
1,
2
F
a
c
u
lt
y
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
Un
iv
e
rsiti
T
e
k
n
o
lo
g
i
M
a
lay
sia
,
Jo
h
o
r
Ba
h
ru
,
M
a
lay
sia
3
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
n
d
Co
m
p
u
ter E
n
g
in
e
e
rin
g
F
a
c
u
lt
y
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
Un
iv
e
rsiti
T
e
k
n
o
l
o
g
i
M
a
lay
sia
Jo
h
o
r
Ba
h
ru
,
M
a
lay
sia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
2
7
,
2
0
1
8
R
ev
i
s
ed
A
p
r
2
0
,
2
0
1
8
A
cc
ep
ted
A
p
r
2
6
,
2
0
1
8
Co
n
ten
t
-
b
a
se
d
im
a
g
e
r
e
tri
e
v
a
l
is
a
p
ro
c
e
ss
f
ra
m
e
w
o
rk
th
a
t
a
p
p
li
e
s
c
o
m
p
u
ter
v
isio
n
tec
h
n
i
q
u
e
s
f
o
r
se
a
rc
h
in
g
a
n
d
m
a
n
a
g
in
g
lar
g
e
i
m
a
g
e
c
o
ll
e
c
t
io
n
s
m
o
re
e
ff
ici
e
n
tl
y
.
W
it
h
th
e
g
ro
w
th
o
f
l
a
rg
e
d
ig
it
a
l
i
m
a
g
e
c
o
ll
e
c
ti
o
n
s
tri
g
g
e
r
e
d
b
y
ra
p
id
a
d
v
a
n
c
e
s in
e
lec
tro
n
ic sto
ra
g
e
c
a
p
a
c
it
y
a
n
d
c
o
m
p
u
ti
n
g
p
o
w
e
r
,
th
e
re
is
a
g
ro
w
in
g
n
e
e
d
f
o
r
d
e
v
ice
s
a
n
d
c
o
m
p
u
ter
sy
ste
m
s
to
su
p
p
o
rt
e
ff
icie
n
t
b
ro
w
sin
g
,
se
a
rc
h
in
g
,
a
n
d
re
tri
e
v
a
l
f
o
r
ima
g
e
c
o
ll
e
c
ti
o
n
s.
He
n
c
e
,
th
e
a
im
o
f
th
is
p
ro
jec
t
is
to
d
e
v
e
lo
p
a
c
o
n
ten
t
-
b
a
se
d
im
a
g
e
re
tri
e
v
a
l
s
y
ste
m
t
h
a
t
c
a
n
b
e
im
p
le
m
e
n
ted
in
a
n
im
a
g
e
g
a
ll
e
ry
d
e
sk
to
p
a
p
p
li
c
a
ti
o
n
t
o
a
ll
o
w
e
ff
ici
e
n
t
b
ro
w
sin
g
th
ro
u
g
h
th
re
e
d
if
f
e
re
n
t
se
a
rc
h
m
o
d
e
s:
re
tri
e
v
a
l
b
y
i
m
a
g
e
q
u
e
r
y
,
re
tri
e
v
a
l
b
y
f
a
c
ial
re
c
o
g
n
it
io
n
,
a
n
d
re
tri
e
v
a
l
b
y
tex
t
o
r
tag
s.
In
th
is
p
r
o
jec
t,
th
e
M
P
EG
-
7
-
li
k
e
P
o
w
e
re
d
L
o
c
a
l
ize
d
Co
lo
r
a
n
d
E
d
g
e
Dire
c
ti
v
it
y
De
sc
rip
to
r
is
u
se
d
to
e
x
trac
t
th
e
f
e
a
tu
re
v
e
c
to
rs
o
f
th
e
ima
g
e
d
a
tab
a
s
e
a
n
d
th
e
f
a
c
ial
re
c
o
g
n
it
io
n
sy
ste
m
is
b
u
il
t
a
r
o
u
n
d
t
h
e
Ei
g
e
n
f
a
c
e
s
c
o
n
c
e
p
t.
A
g
ra
p
h
ica
l
u
se
r
in
terf
a
c
e
w
it
h
th
e
b
a
sic
f
u
n
c
ti
o
n
a
li
ty
o
f
a
n
i
m
a
g
e
g
a
ll
e
r
y
a
p
p
li
c
a
ti
o
n
is
a
lso
d
e
v
e
lo
p
e
d
to
im
p
lem
e
n
t
th
e
th
re
e
se
a
rc
h
m
o
d
e
s.
Re
su
lt
s
sh
o
w
th
a
t
th
e
a
p
p
li
c
a
ti
o
n
is
a
b
le
to
re
tri
e
v
e
a
n
d
d
isp
lay
i
m
a
g
e
s
in
a
c
o
ll
e
c
ti
o
n
a
s
th
u
m
b
n
a
il
pr
e
v
ie
w
s
w
it
h
h
ig
h
re
tri
e
v
a
l
a
c
c
u
ra
c
y
a
n
d
m
e
d
iu
m
re
le
v
a
n
c
e
a
n
d
th
e
c
o
m
p
u
tatio
n
a
l
re
q
u
irem
e
n
ts
f
o
r
su
b
se
q
u
e
n
t
se
a
rc
h
e
s
w
e
re
si
g
n
if
ica
n
tl
y
re
d
u
c
e
d
t
h
ro
u
g
h
t
h
e
i
n
c
o
rp
o
ra
ti
o
n
o
f
tex
t
-
b
a
se
d
im
a
g
e
re
tri
e
v
a
l
a
s
o
n
e
o
f
th
e
se
a
rc
h
m
o
d
e
s.
A
ll
in
a
ll
,
th
is
st
u
d
y
in
tro
d
u
c
e
s
a
sim
p
le
a
n
d
c
o
n
v
e
n
ien
t
w
a
y
o
f
o
ff
li
n
e
im
a
g
e
se
a
rc
h
e
s
o
n
d
e
sk
to
p
c
o
m
p
u
ters
a
n
d
p
ro
v
id
e
s
a
ste
p
p
in
g
sto
n
e
to
f
u
tu
re
c
o
n
ten
t
-
b
a
se
d
i
m
a
g
e
re
tri
e
v
a
l
s
y
ste
m
s
b
u
il
t
fo
r
sim
il
a
r
p
u
r
p
o
se
s.
K
ey
w
o
r
d
:
Au
to
-
tag
g
i
n
g
C
o
n
te
n
t
-
b
ased
i
m
a
g
e
r
etr
iev
al
Mp
eg
-
&
p
o
w
er
ed
lo
ca
lized
d
escr
ip
to
r
P
r
in
cip
al
co
m
p
o
n
e
n
t a
n
al
y
s
is
T
ex
t
-
b
ased
i
m
a
g
e
r
etr
iev
al
Co
p
y
rig
h
t
©
2
0
1
8
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
:
Z
aid
B
in
O
m
ar
,
Dep
ar
t
m
en
t o
f
E
lectr
ic
an
d
C
o
m
p
u
ter
E
n
g
i
n
ee
r
i
n
g
,
Facu
lt
y
o
f
E
lectr
ical
E
n
g
in
ee
r
in
g
,
Un
i
v
er
s
iti T
ek
n
o
lo
g
i M
ala
y
s
ia
,
J
o
h
o
r
B
ah
r
u
,
Ma
lay
s
ia
.
E
m
ail: l
s
n
t
l@
cc
u
.
ed
u
.
t
w
1.
I
NT
RO
D
UCT
I
O
N
I
n
r
ec
en
t
y
ea
r
s
,
r
ap
id
ad
v
an
c
es
in
elec
tr
o
n
ic
s
to
r
ag
e
ca
p
ac
it
y
an
d
co
m
p
u
ti
n
g
p
o
w
er
h
av
e
tr
ig
g
er
ed
th
e
g
r
o
w
t
h
o
f
lar
g
e
d
ig
ita
l
i
m
ag
e
co
llectio
n
s
f
o
llo
w
in
g
t
h
e
in
cr
ea
s
e
o
f
u
s
er
s
o
n
t
h
e
i
n
ter
n
et.
O
v
er
th
e
y
ea
r
s
,
w
e
h
a
v
e
s
ee
n
ex
p
o
n
e
n
tial
i
n
cr
ea
s
es
in
n
u
m
b
er
o
f
d
ig
ital
i
m
ag
es
an
d
v
id
eo
b
o
th
o
v
er
th
e
n
et
an
d
ev
e
n
in
o
u
r
o
w
n
d
ev
ice
s
a
s
w
e
atte
m
p
t
to
k
ee
p
m
o
r
e
m
e
m
o
r
ies
th
r
o
u
g
h
p
h
o
to
s
an
d
v
id
eo
s
.
T
h
is
i
n
cr
ea
s
ed
u
s
ag
e
m
a
y
b
e
d
u
e
to
s
ev
er
al
f
ac
to
r
s
s
u
ch
a
s
ed
u
ca
tio
n
,
en
ter
tain
m
e
n
t,
co
m
m
er
cial
p
u
r
p
o
s
es,
an
d
etc.
a
n
d
it
is
n
o
w
ap
p
ar
en
t
th
at
m
o
r
e
a
n
d
m
o
r
e
i
m
a
g
es a
r
e
r
o
u
tin
el
y
u
s
ed
to
co
n
v
e
y
lar
g
e
a
m
o
u
n
t
s
o
f
i
n
f
o
r
m
atio
n
.
Du
e
to
th
e
i
n
cr
ea
s
i
n
g
d
i
f
f
ic
u
lt
y
in
m
a
k
i
n
g
p
r
o
p
er
u
s
e
o
f
th
e
in
f
o
r
m
atio
n
co
n
tai
n
ed
in
d
ig
i
t
al
i
m
a
g
es
an
d
v
id
eo
s
,
ad
v
an
ce
d
i
n
f
o
r
m
a
tio
n
s
y
s
te
m
s
ar
e
n
o
w
m
o
r
e
im
p
o
r
ta
n
t
t
h
an
e
v
er
as
th
e
y
ar
e
n
ee
d
ed
to
m
a
n
ag
e
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
9
0
3
–
1912
1904
i
m
a
g
e
co
llectio
n
s
m
o
r
e
e
f
f
ici
en
tl
y
.
I
m
a
g
e
s
ea
r
c
h
i
n
g
i
s
o
n
e
o
f
th
e
m
o
s
t
i
m
p
o
r
ta
n
t
f
u
n
cti
o
n
s
t
h
at
n
ee
d
s
to
b
e
s
u
p
p
o
r
ted
b
y
d
ev
ices
a
n
d
co
m
p
u
ter
s
y
s
te
m
s
to
allo
w
ef
f
ic
ien
t
b
r
o
w
s
i
n
g
,
s
ea
r
ch
i
n
g
,
an
d
r
etr
iev
al.
W
ith
th
e
r
ap
id
a
d
v
an
ce
m
en
t
i
n
i
m
a
g
e
ca
p
tu
r
in
g
d
ev
ices
o
v
er
th
e
y
ea
r
s
ad
d
ed
w
it
h
th
e
ad
v
en
t
o
f
v
a
r
io
u
s
s
o
cial
m
ed
ia
p
latf
o
r
m
s
,
i
t
h
as
b
ec
o
m
e
a
co
m
m
o
n
c
u
lt
u
r
e
f
o
r
m
e
m
b
er
s
o
f
cu
r
r
en
t
s
o
ciet
y
to
ta
k
e
a
lo
t
o
f
p
h
o
to
s
w
it
h
t
h
eir
p
h
o
n
es
ev
er
y
d
a
y
.
A
cc
o
r
d
in
g
to
a
s
u
r
v
e
y
b
y
p
er
s
o
n
al
m
ed
ia
s
tar
tu
p
Ma
g
i
s
to
,
th
e
av
er
a
g
e
s
m
ar
tp
h
o
n
e
u
s
er
tak
es
ar
o
u
n
d
1
5
0
n
e
w
p
h
o
to
s
ev
er
y
m
o
n
t
h
.
Desp
ite
t
h
is
,
c
u
r
r
en
t
g
a
ller
y
ap
p
licatio
n
s
ca
n
n
o
t
k
ee
p
u
p
w
ith
h
u
g
e
d
atab
ases
o
f
i
m
ag
e
s
an
d
th
e
o
n
l
y
co
m
m
o
n
m
et
h
o
d
av
ailab
le
to
r
etr
iev
e
o
ld
im
a
g
es
is
to
s
cr
o
ll
th
r
o
u
g
h
m
an
y
u
n
w
a
n
ted
i
m
a
g
es
b
e
f
o
r
e
ar
r
iv
in
g
at
t
h
e
d
esire
d
i
m
ag
e
w
h
ic
h
i
s
v
er
y
ted
io
u
s
an
d
ti
m
e
co
n
s
u
m
i
n
g
.
Mo
r
eo
v
er
,
co
n
v
e
n
tio
n
al
C
B
I
R
s
y
s
te
m
s
ar
e
g
e
n
er
all
y
co
m
p
u
tatio
n
all
y
h
ea
v
y
f
o
r
o
f
f
li
n
e
ap
p
licatio
n
s
w
h
er
e
p
er
f
o
r
m
a
n
ce
i
s
e
x
p
ec
ted
to
b
e
f
as
t
w
h
ile
s
till
b
ein
g
ab
le
to
p
r
o
d
u
ce
r
elev
an
t
r
e
s
u
l
ts
a
s
co
m
p
u
ti
n
g
s
p
ee
d
m
a
y
v
ar
y
f
r
o
m
co
m
p
u
ter
to
co
m
p
u
t
er
.
C
o
n
te
n
t
-
B
ased
I
m
a
g
e
R
etr
ie
v
al
(
C
B
I
R
)
i
s
o
n
e
in
s
ta
n
ce
o
f
i
n
f
o
r
m
atio
n
r
etr
iev
al
th
at
ap
p
lies
co
m
p
u
ter
v
i
s
io
n
tec
h
n
iq
u
e
s
t
o
s
o
lv
e
p
r
o
b
lem
s
r
elate
d
to
s
ea
r
ch
i
n
g
a
n
d
m
a
n
ag
in
g
lar
g
e
i
m
ag
e
d
atab
ase
s
.
Ho
w
e
v
er
,
m
o
s
t
C
B
I
R
s
y
s
te
m
s
t
h
at
ai
m
s
to
m
a
n
ag
e
d
ig
ita
l
co
llectio
n
s
in
o
f
f
li
n
e
d
atab
ase
ten
d
to
u
s
e
i
m
a
g
e
co
n
ten
t a
s
q
u
er
y
r
at
h
er
th
a
n
c
o
n
s
id
er
in
g
u
s
er
p
r
ef
er
en
ce
i
n
d
ef
in
i
n
g
th
e
i
m
a
g
e
i
n
q
u
est
io
n
an
d
th
i
s
m
a
y
n
o
t b
e
co
n
v
e
n
ien
t
esp
ec
iall
y
w
h
e
n
it
is
t
h
e
o
n
l
y
a
v
ailab
le
m
o
d
e
o
f
s
ea
r
ch
as
f
u
t
u
r
e
e
f
f
o
r
ts
o
f
s
ea
r
ch
in
g
f
o
r
t
h
e
s
a
m
e
q
u
er
y
i
m
a
g
e
m
a
y
b
e
r
ed
u
n
d
an
t.
He
n
ce
,
t
h
is
p
r
o
j
ec
t
aim
s
to
m
ak
e
i
m
ag
e
g
a
ller
ies
m
o
r
e
o
r
g
an
ized
b
y
in
tr
o
d
u
ci
n
g
a
co
m
b
i
n
atio
n
C
B
I
R
an
d
T
B
I
R
-
b
ased
s
y
s
te
m
f
o
r
m
o
r
e
co
n
v
en
ie
n
t
o
f
f
li
n
e
s
ea
r
ch
e
s
t
h
r
o
u
g
h
au
to
m
at
ic
g
e
n
er
atio
n
o
f
te
x
t
u
al
m
etad
ata
b
y
u
s
in
g
i
n
f
o
r
m
atio
n
o
b
tain
ed
f
r
o
m
u
s
er
i
n
p
u
t
a
n
d
p
r
ev
io
u
s
r
etr
iev
al
r
esu
l
ts
.
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
I
n
m
o
s
t o
f
t
h
e
ea
r
lier
r
etr
ie
v
al
s
y
s
te
m
s
,
v
id
eo
o
r
i
m
a
g
e
co
n
te
n
ts
ar
e
m
an
a
g
ed
b
y
k
e
y
w
o
r
d
s
o
r
tex
tu
a
l
m
etad
ata
[
1
]
.
C
o
n
ten
t
-
b
ased
i
m
a
g
e
r
etr
iev
al
(
C
B
I
R
)
h
o
w
e
v
er
r
elies
o
n
ex
tr
ac
ti
n
g
t
h
e
ap
p
r
o
p
r
iate
ch
ar
ac
ter
is
tic
q
u
a
n
tit
ies
ca
lle
d
„
d
escr
ip
to
r
s
‟
o
r
„
f
ea
t
u
r
es
‟
d
escr
ib
in
g
t
h
e
d
esire
d
co
n
ten
ts
[
2
]
,
[
3
]
.
A
C
B
I
R
s
y
s
te
m
co
n
s
is
t
s
o
f
an
in
ter
f
ac
e
f
o
r
th
e
ac
q
u
i
s
itio
n
o
f
t
h
e
q
u
er
y
i
m
ag
e,
d
atab
ase
s
f
o
r
s
to
r
in
g
in
d
e
x
i
n
g
d
ata
an
d
d
is
tan
ce
m
etr
ics,
a
n
d
a
s
i
m
i
lar
it
y
co
m
p
ar
is
o
n
an
d
r
etr
iev
al
s
y
s
te
m
.
2
.
1
.
F
ea
t
ure
v
ec
t
o
r
ex
t
ra
ct
io
n
So
m
e
co
m
m
o
n
l
y
u
s
ed
f
ea
t
u
r
e
v
ec
to
r
s
in
cl
u
d
e
co
lo
u
r
,
tex
t
u
r
e,
s
h
ap
e,
s
p
atial
lo
ca
tio
n
,
etc.
Du
e
to
its
s
tab
ilit
y
a
n
d
r
o
b
u
s
t
n
es
s
;
ap
p
licatio
n
o
f
co
lo
r
f
ea
t
u
r
es
is
w
i
d
ely
ac
ce
p
ted
i
n
m
o
s
t
C
B
I
R
ap
p
licatio
n
s
.
J
alab
H.
A
[
4
]
i
m
p
le
m
e
n
ted
an
i
m
a
g
e
r
etr
iev
al
s
y
s
te
m
b
ased
o
n
co
lo
r
lay
o
u
t
d
escr
ip
to
r
(
C
L
D)
r
ep
r
esen
tin
g
t
h
e
s
p
atial
d
is
tr
ib
u
t
io
n
o
f
co
lo
r
s
,
J
ay
a
m
ala
K.
P
atil
a
n
d
R
aj
Ku
m
ar
[
5
]
s
u
g
g
e
s
ted
a
p
lan
t
leaf
d
is
ea
s
e
i
m
ag
e
r
etr
iev
al
u
s
i
n
g
c
o
lo
r
m
o
m
en
t
s
,
C
h
atzic
h
r
is
to
f
is
et
al
[
6
]
p
r
o
p
o
s
ed
a
co
lo
u
r
an
d
ed
g
e
d
i
r
ec
tiv
it
y
d
escr
ip
to
r
(
C
E
DD)
in
co
r
p
o
r
atin
g
b
o
th
c
o
lo
r
an
d
tex
t
u
r
e
in
f
o
r
m
at
io
n
i
n
a
h
is
to
g
r
a
m
.
T
ex
t
u
r
e
f
ea
t
u
r
e
co
n
tain
v
al
u
ab
le
in
f
o
r
m
atio
n
o
n
th
e
s
u
r
f
ac
e
s
t
r
u
ctu
r
es
o
f
o
b
j
ec
ts
an
d
th
eir
r
elatio
n
s
h
ip
to
t
h
e
s
u
r
r
o
u
n
d
in
g
e
n
v
ir
o
n
m
e
n
t
[
7
]
.
B
ased
o
n
p
r
ev
io
u
s
r
esear
ch
,
it
is
f
o
u
n
d
th
a
t
th
e
m
o
s
t
i
m
p
o
r
tan
t
tex
t
u
r
e
f
ea
t
u
r
es
ar
e
co
ar
s
en
es
s
,
co
n
tr
ast,
d
ir
ec
tio
n
alit
y
.
T
h
e
Steer
ab
le
P
y
r
a
m
id
Mo
d
el
[
8
]
an
d
Gab
o
r
w
av
ele
t
T
r
an
s
f
o
r
m
(
GW
T
)
[
9
]
ar
e
am
o
n
g
th
e
m
o
s
t
w
id
el
y
u
s
ed
f
ea
t
u
r
es.
U
s
u
all
y
s
h
ap
e
f
ea
tu
r
e
r
ep
r
esen
t
atio
n
s
ar
e
o
n
l
y
u
s
e
f
u
l
a
f
ter
i
m
ag
e
s
eg
m
e
n
tatio
n
.
Kau
p
p
in
e
n
et
al.
h
a
v
e
s
h
o
wn
t
h
at
Fo
u
r
ier
d
escr
ip
to
r
s
u
s
ed
in
2
-
D
s
h
ap
e
clas
s
i
f
icati
o
n
p
er
f
o
r
m
s
b
etter
co
m
p
ar
ed
to
au
to
r
eg
r
es
s
iv
e
m
o
d
ellin
g
b
ased
s
h
ap
e
d
escr
ip
to
r
[
10
].
2
.
2
.
F
a
cia
l
r
ec
o
g
nitio
n
T
h
e
m
o
s
t
co
m
m
o
n
m
eth
o
d
s
o
f
f
ac
ia
l
r
ec
o
g
n
itio
n
ar
e
E
i
g
en
f
ac
e
s
,
Fi
s
h
er
f
ac
es
a
n
d
L
o
ca
l
B
in
ar
y
P
atter
n
s
His
to
g
r
a
m
s
(
L
B
P
H)
.
T
h
e
P
r
in
cip
al
C
o
m
p
o
n
en
t
An
al
y
s
i
s
(
P
C
A
)
p
r
o
p
o
s
ed
b
y
K
ar
l
P
ea
r
s
o
n
(
1
9
0
1
)
an
d
Har
o
ld
Ho
tellin
g
i
s
a
co
r
e
co
m
p
o
n
e
n
t
o
f
t
h
e
E
ig
e
n
f
ac
es
m
et
h
o
d
w
h
ic
h
tr
ies
to
f
o
cu
s
o
n
t
h
e
m
o
s
t
i
m
p
o
r
tan
t
co
m
p
o
n
e
n
ts
o
f
t
h
e
d
ataset,
h
o
w
e
v
er
it
d
o
es
n
o
t
co
n
s
id
er
class
es
[
1
1
]
.
I
n
Fis
h
er
f
ac
e
s
ap
p
r
o
ac
h
,
L
i
n
ea
r
Dis
cr
i
m
i
n
a
n
t
a
n
al
y
s
i
s
is
u
s
ed
to
p
er
f
o
r
m
d
i
m
e
n
s
io
n
alit
y
r
ed
u
c
tio
n
b
y
cla
s
s
es
[
12
]
.
L
B
P
H
an
al
y
s
es
ea
ch
f
ac
e
i
n
th
e
tr
ain
in
g
s
et
s
ep
ar
ately
a
n
d
in
d
ep
en
d
en
tl
y
[
1
3
]
.
A
p
ar
t
f
r
o
m
th
e
tr
ad
itio
n
a
l
m
eth
o
d
s
,
t
h
er
e
ar
e
also
s
o
m
e
m
o
d
er
n
r
esear
ch
d
o
n
e
is
th
i
s
ar
ea
an
d
t
h
e
r
esu
lt
s
ar
e
p
r
o
m
is
i
n
g
.
Dee
p
Fa
ce
[
1
4
]
is
u
s
ed
b
y
Face
b
o
o
k
to
au
t
o
m
atica
ll
y
s
u
g
g
e
s
t
a
tag
f
o
r
f
ac
es
i
n
p
h
o
t
o
s
an
d
v
id
eo
s
.
Face
Net
[
1
5
]
b
y
Go
o
g
le
u
s
e
s
a
E
u
clid
ea
n
s
p
ac
e
f
o
r
i
m
a
g
e
r
ep
r
esen
tatio
n
cr
ea
ted
i
m
ag
e
s
g
e
n
er
ated
th
r
o
u
g
h
a
d
ata
-
m
in
i
n
g
m
eth
o
d
.
2
.
3
.
Si
m
ila
r
a
pp
lica
t
io
n
Gett
y
I
m
ag
e
s
is
an
e
x
te
n
s
i
v
e
w
eb
-
b
ased
g
aller
y
t
h
at
s
el
l
s
h
i
g
h
-
q
u
a
lit
y
s
to
ck
i
m
a
g
es
f
o
r
u
s
e
o
f
ad
v
er
tis
i
n
g
,
m
ar
k
eti
n
g
,
a
n
d
m
o
r
e.
I
t
is
b
ased
o
n
T
B
I
R
b
y
c
o
llectiv
e
ta
g
g
i
n
g
w
h
er
e
s
ev
er
al
h
u
m
an
in
d
e
x
er
s
lo
o
k
at
n
e
w
i
m
a
g
e
a
n
d
en
ter
a
s
s
o
ciate
d
k
e
y
w
o
r
d
s
a
n
d
p
r
ev
i
o
u
s
u
s
er
q
u
er
ies ar
e
co
m
b
i
n
ed
to
f
o
r
m
a
t
h
esa
u
r
u
s
f
o
r
f
u
t
u
r
e
s
ea
r
ch
e
s
.
O
n
e
o
f
t
h
e
ea
r
lies
t c
o
m
m
er
cial
u
s
e
C
B
I
R
s
y
s
te
m
s
is
th
e
Qu
er
y
B
y
I
m
ag
e
C
o
n
te
n
t (
QB
I
C
)
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
C
o
n
ten
t
-
b
a
s
ed
I
ma
g
e
R
etri
ev
a
l S
ystem
F
o
r
A
n
I
ma
g
e
G
a
ller
y
S
ea
r
ch
A
p
p
lica
tio
n
(
N
ico
le
Th
a
m
)
1905
d
ev
elo
p
ed
b
y
I
B
M.
I
t
s
u
p
p
o
r
ts
q
u
er
ie
s
b
ased
o
n
u
s
er
s
k
et
ch
es,
q
u
er
y
i
m
a
g
e
s
,
as
w
ell
a
s
co
lo
r
an
d
te
x
t
u
r
e
p
atter
n
s
s
elec
ted
b
y
t
h
e
u
s
er
[
1
6
]
an
d
u
s
es
a
co
m
b
in
a
tio
n
o
f
co
lo
r
,
tex
tu
r
e,
an
d
s
h
ap
e
as
f
ea
t
u
r
e
d
escr
ip
to
r
w
h
ic
h
i
n
clu
d
e
s
an
i
m
p
r
o
v
ed
v
er
s
io
n
o
f
th
e
T
a
m
u
r
a
tex
tu
r
e
r
ep
r
esen
tatio
n
[
1
7
]
,
an
d
m
aj
o
r
ax
is
o
r
ie
n
tatio
n
.
I
t
h
as
v
er
y
f
a
s
t
p
er
f
o
r
m
an
ce
a
n
d
th
e
r
esu
lt
s
ar
e
in
v
ar
ia
n
t
to
s
m
all
ch
an
g
es
i
n
p
er
s
p
ec
tiv
es,
h
o
w
e
v
er
s
o
m
e
o
f
it
s
w
ea
k
n
ess
e
s
in
cl
u
d
e
s
en
s
iti
v
it
y
to
ill
u
m
in
a
tio
n
ch
a
n
g
es
an
d
n
o
lo
ca
lizatio
n
o
f
co
lo
u
r
.
L
ik
e.
co
m
,
n
o
w
o
w
n
ed
b
y
Go
o
g
le
u
n
d
er
g
o
o
g
le.
co
m
/s
h
o
p
p
in
g
u
s
es
C
B
I
R
to
s
ea
r
c
h
f
o
r
p
r
o
d
u
cts
s
i
m
ilar
to
q
u
er
y
an
d
r
et
u
r
n
r
es
u
lt
s
o
f
s
i
m
ilar
it
e
m
s
w
it
h
lin
k
s
to
r
etailer
s
s
u
c
h
as
Am
az
o
n
.
co
m
.
I
t
also
allo
w
s
u
s
er
s
to
s
elec
t
r
eg
io
n
s
o
f
a
p
r
o
d
u
ct
i
m
a
g
e
to
r
etr
iev
e
p
r
o
d
u
cts r
an
k
ed
b
y
s
i
m
ilar
p
atter
n
s
,
co
lo
u
r
s
an
d
s
h
ap
es.
User
s
ca
n
d
eter
m
i
n
e
w
h
ich
o
f
th
e
s
e
th
r
ee
cr
iter
ia
ar
e
m
o
r
e
i
m
p
o
r
tan
t to
t
h
e
m
to
i
m
p
r
o
v
e
th
e
s
ea
r
ch
r
es
u
lts
.
T
h
e
co
m
p
a
n
y
Go
o
g
le
also
h
a
s
a
n
o
th
er
ap
p
licatio
n
ca
lled
Go
o
g
le
P
h
o
to
s
w
h
ic
h
is
a
c
lo
u
d
-
b
ased
ap
p
licatio
n
th
at
u
s
e
s
f
ac
ial
r
ec
o
g
n
itio
n
to
f
in
d
p
h
o
to
s
o
f
p
eo
p
le
in
t
h
e
g
al
ler
y
.
T
h
is
r
eq
u
ir
es
t
h
e
u
s
er
to
f
i
r
s
t
ap
p
l
y
a
lab
el
to
a
p
h
o
t
o
o
f
s
o
m
eo
n
e
a
n
d
th
e
y
w
il
l
b
e
ab
le
to
s
ea
r
ch
f
o
r
th
a
t
p
ar
ticu
lar
in
d
i
v
id
u
a
l
u
s
i
n
g
t
h
at
lab
el.
User
s
m
a
y
al
s
o
g
o
o
n
lin
e
a
n
d
s
ea
r
c
h
p
h
o
to
s
b
y
co
m
m
o
n
k
e
y
w
o
r
d
s
w
it
h
o
u
t d
ef
i
n
i
n
g
t
h
e
m
.
3.
M
E
T
H
O
DO
L
O
G
Y
T
h
e
s
y
s
te
m
co
n
s
is
ts
o
f
t
h
r
ee
m
ai
n
co
m
p
o
n
en
t
s
:
th
e
g
a
ller
y
in
ter
f
ac
e,
t
h
e
q
u
er
y
p
r
o
ce
s
s
i
n
g
m
o
d
u
le,
an
d
th
e
i
m
ag
e
d
atab
ase
a
n
d
allo
w
s
th
r
ee
m
o
d
es
o
f
s
ea
r
ch
e
s
:
b
y
a
r
ef
er
en
ce
i
m
a
g
e,
a
n
a
m
e,
o
r
a
p
r
ev
io
u
s
l
y
d
ef
in
ed
ta
g
.
Fig
u
r
e
1
s
h
o
w
s
th
e
to
p
-
lev
el
b
lo
ck
d
iag
r
a
m
f
o
r
t
h
e
o
v
er
all
s
y
s
te
m
.
Fig
u
r
e
1
.
T
o
p
-
lev
el
b
lo
ck
d
iag
r
a
m
o
f
th
e
s
y
s
te
m
3
.
1
.
R
et
riev
a
l by
f
ea
t
ure
v
ec
t
o
rs
T
h
e
MP
E
G
-
7
-
lik
e
P
o
w
er
ed
L
o
ca
lized
C
o
lo
r
an
d
E
d
g
e
Dir
ec
tiv
it
y
De
s
cr
ip
to
r
(
SIM
P
L
E
-
C
E
DD)
i
s
u
s
ed
as
o
u
r
f
ea
tu
r
e
v
ec
to
r
[
1
8
]
.
A
s
th
e
s
ize
o
f
a
C
E
DD
d
escr
ip
to
r
is
m
e
m
o
r
y
e
f
f
icie
n
t
(
o
n
l
y
5
4
b
y
te
s
p
er
i
m
a
g
e)
[
6
]
,
an
d
r
eq
u
ir
es
r
elati
v
el
y
lo
w
co
m
p
u
tatio
n
al
p
o
w
e
r
to
ex
tr
ac
t,
it
is
s
u
itab
le
to
b
e
u
s
ed
f
o
r
s
ea
r
c
h
in
g
lar
g
e
i
m
a
g
e
d
atab
ases
s
u
c
h
as
a
lo
ca
l
i
m
a
g
e
g
aller
y
i
n
a
c
o
m
p
u
ter
.
F
u
r
t
h
er
m
o
r
e,
SIM
P
L
E
-
C
E
DD
lo
ca
lizes
th
e
i
m
a
g
e
f
ea
tu
r
e
s
b
y
f
ir
s
t
lo
c
atin
g
f
ea
t
u
re
-
r
ic
h
r
eg
io
n
s
an
d
d
ef
in
e
t
h
ese
p
atch
e
s
as
r
eg
io
n
s
-
of
-
in
ter
e
s
t
o
r
R
OI
b
ef
o
r
e
ex
tr
ac
tio
n
.
A
s
a
r
es
u
lt,
th
e
f
ea
t
u
r
e
v
ec
to
r
b
ec
o
m
es
m
o
r
e
r
o
b
u
s
t
to
i
m
a
g
e
tr
an
s
f
o
r
m
atio
n
s
an
d
allo
w
s
f
aster
ex
ec
u
tio
n
.
T
h
e
ex
tr
ac
ted
f
ea
tu
r
e
v
ec
to
r
s
ar
e
co
m
p
ar
ed
f
o
r
s
i
m
ilar
it
y
u
s
in
g
t
h
e
T
an
im
o
to
co
ef
f
icie
n
t
i
s
d
escr
ib
ed
in
E
q
u
atio
n
(
1
)
w
h
e
r
e
a
an
d
b
ar
e
tw
o
s
ep
ar
ate
p
o
in
ts
.
∑
∑
∑
∑
(
1
)
T
o
f
ilter
th
e
r
es
u
lt
s
to
o
n
l
y
s
h
o
w
th
e
m
o
s
t
s
i
m
ilar
i
m
a
g
es
,
a
cu
s
to
m
v
ar
ian
t
o
f
K
-
m
ea
n
s
i
s
u
s
ed
in
s
tead
o
f
a
lo
w
p
as
s
f
ilter
a
s
t
h
e
o
p
ti
m
u
m
d
i
f
f
er
en
ce
th
r
esh
o
ld
(
m
ea
n
i
n
g
t
h
at
id
ea
ll
y
,
all
v
is
u
all
y
s
i
m
ilar
i
m
a
g
es
s
h
o
u
ld
b
e
as
s
o
ciate
d
w
it
h
d
i
f
f
er
en
ce
v
al
u
es
b
elo
w
th
is
t
h
r
esh
o
ld
)
m
a
y
v
ar
y
w
i
th
d
if
f
er
en
t
d
atasets
,
h
en
ce
K
-
m
ea
n
s
is
u
s
ed
to
ad
ap
t
to
th
e
ch
an
g
es
o
f
t
h
is
t
h
r
esh
o
ld
v
alu
e.
T
h
e
o
b
j
ec
tiv
e
o
f
K
-
m
ea
n
s
clu
s
ter
in
g
is
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
9
0
3
–
1912
1906
to
g
r
o
u
p
n
n
u
m
b
er
o
f
e
le
m
e
n
t
s
o
r
d
ata
p
o
in
t
s
in
to
k
n
u
m
b
er
o
f
c
lu
s
ter
s
b
y
m
i
n
i
m
izi
n
g
t
h
e
t
o
tal
i
n
tr
a
-
clu
s
ter
v
ar
ian
ce
as
in
E
q
u
atio
n
(
2
)
,
w
h
er
e
J
is
t
h
e
o
b
j
ec
tiv
e
f
u
n
ct
io
n
,
x
i
is
t
h
e
o
b
s
er
v
at
io
n
f
o
r
ca
s
e
i,
a
n
d
cj
is
t
h
e
ce
n
tr
o
id
f
o
r
clu
s
ter
j
.
Fo
r
th
is
i
m
p
le
m
e
n
tatio
n
,
th
e
n
u
m
b
er
o
r
clu
s
ter
s
,
k
is
n
o
t
d
ef
i
n
ed
b
u
t
g
r
ad
u
all
y
ad
d
ed
w
h
en
a
p
o
in
t
li
es
o
u
ts
id
e
t
h
e
b
o
u
n
d
ar
ies
o
f
th
e
cl
u
s
ter
r
ad
iu
s
o
f
all
e
x
i
s
ti
n
g
ce
n
tr
o
id
s
d
u
r
i
n
g
th
e
f
ir
s
t
it
er
atio
n
.
T
h
e
clu
s
ter
r
ad
iu
s
is
d
ef
i
n
ed
at
t
h
e
b
eg
i
n
n
in
g
o
f
t
h
e
iter
atio
n
an
d
s
er
v
es
as
t
h
e
d
i
f
f
er
e
n
ce
t
h
r
es
h
o
ld
f
o
r
i
m
a
g
e
r
etr
ie
v
al.
Ho
w
e
v
er
,
th
e
r
etr
ie
v
al
o
u
tp
u
t
w
il
l a
l
w
a
y
s
tak
e
t
h
e
f
ir
s
t c
l
u
s
t
er
o
f
i
m
a
g
e
s
w
it
h
t
h
e
lea
s
t d
ev
iatio
n
r
eg
ar
d
les
s
o
f
w
h
et
h
er
o
r
n
o
t
th
e
y
ar
e
in
h
er
e
n
tl
y
s
i
m
ilar
to
t
h
e
q
u
er
y
i
m
a
g
e.
(
Fig
u
r
e
2
(
a
)
)
T
o
r
eso
lv
e
th
i
s
is
s
u
e,
a
n
ele
m
e
n
t
w
it
h
a
d
ev
ia
tio
n
v
al
u
e
o
f
0
(
to
tall
y
s
i
m
ilar
)
is
ad
d
ed
in
to
t
h
e
ar
r
a
y
b
ef
o
r
e
clu
s
ter
in
g
.
T
h
is
ele
m
e
n
t
d
o
es
n
o
t
co
r
r
esp
o
n
d
to
an
y
i
m
a
g
es
w
h
atso
e
v
er
,
b
u
t
ad
d
ed
to
r
e
tr
iev
e
i
m
ag
e
s
o
n
l
y
w
h
e
n
s
i
m
ilar
i
m
ag
e
s
ex
i
s
t
(
Fig
u
r
e
2
(
b
)
).
(
a)
(
b
)
Fig
u
r
e
2
.
R
etr
iev
al
r
es
u
lt
s
w
it
h
n
o
s
i
m
ilar
i
m
a
g
es,
(
a)
n
o
d
u
m
m
y
ele
m
e
n
t,
(
b
)
w
it
h
d
u
m
m
y
ele
m
e
n
t
3
.
2
.
Ret
riev
a
l by
f
a
ci
a
l r
ec
o
g
niti
o
n
B
ef
o
r
e
th
e
s
y
s
te
m
is
ab
le
to
s
ea
r
ch
f
o
r
i
m
a
g
es
o
f
in
d
i
v
id
u
a
l
s
,
th
e
u
s
er
f
ir
s
t
h
a
s
to
ad
d
a
f
ac
e
in
to
th
e
tr
ain
i
n
g
d
ataset.
Fo
r
th
i
s
tas
k
,
th
e
Haa
r
f
ea
tu
r
e
-
b
ased
ca
s
c
ad
e
class
if
ier
s
is
u
s
ed
f
o
r
f
ac
e
d
etec
tio
n
.
I
m
a
g
e
r
eg
io
n
s
t
h
at
ar
e
lik
el
y
to
co
n
t
ain
f
ac
e
s
b
y
ar
e
lo
ca
ted
b
y
s
c
an
n
i
n
g
th
e
i
m
a
g
e
s
e
v
er
al
ti
m
e
s
at
d
if
f
er
e
n
t
s
ca
le
s
b
y
i
n
cr
ea
s
i
n
g
th
e
R
OI
i
n
s
u
b
s
eq
u
en
t
r
o
u
n
d
s
to
f
i
n
d
t
h
e
f
ac
e.
All
co
n
tain
in
g
r
e
g
io
n
s
w
i
th
d
etec
ted
f
ac
e
s
ar
e
s
av
ed
to
th
e
tr
ain
i
n
g
d
ataset
a
lo
n
g
w
i
th
t
h
e
n
a
m
e
th
a
t
th
e
u
s
er
w
is
h
es
to
id
e
n
t
if
y
it
a
s
.
T
h
is
i
n
f
o
r
m
atio
n
w
ill
b
e
u
s
ef
u
l
la
ter
d
u
r
in
g
te
x
t
-
b
as
ed
s
ea
r
ch
.
T
o
s
p
ee
d
u
p
th
is
p
r
o
ce
s
s
,
th
e
C
an
n
y
ed
g
e
d
etec
to
r
is
u
s
ed
to
i
g
n
o
r
e
i
m
a
g
e
r
eg
io
n
s
w
it
h
to
o
f
e
w
o
r
to
o
m
u
ch
ed
g
es.
Facial
r
ec
o
g
n
itio
n
is
d
o
n
e
u
s
in
g
P
r
in
cip
al
C
o
m
p
o
n
e
n
t
An
al
y
s
i
s
(
P
C
A
)
.
I
n
t
h
is
ca
s
e,
o
u
r
s
et
o
f
tr
ain
i
n
g
i
m
a
g
es
is
co
n
v
er
ted
i
n
to
a
s
et
o
f
ca
lc
u
lated
E
ig
e
n
f
ac
es.
Fo
r
ea
ch
p
r
o
ce
ed
in
g
E
i
g
en
f
ac
es,
th
er
e
ar
e
less
er
f
ea
tu
r
es
a
n
d
m
o
r
e
n
o
i
s
e,
h
en
ce
o
n
l
y
t
h
e
f
e
w
f
ir
s
t
K
E
ig
en
f
ac
e
s
ar
e
s
elec
ted
.
T
h
is
w
a
y
,
th
e
n
u
m
b
er
o
f
v
alu
e
s
n
ee
d
ed
to
r
ec
o
g
n
ize
it
is
r
ed
u
ce
d
an
d
t
h
is
h
elp
s
to
s
p
ee
d
u
p
th
e
r
ec
o
g
n
i
tio
n
p
r
o
ce
s
s
a
n
d
r
ed
u
ce
er
r
o
r
ca
u
s
ed
b
y
n
o
is
e.
As
a
r
esu
lt,
th
e
to
tal
ti
m
e
tak
e
n
to
r
etr
iev
e
i
m
ag
e
s
o
f
in
d
i
v
id
u
a
ls
ca
n
b
e
r
ed
u
ce
d
.
On
ce
all
i
m
a
g
es
ar
e
d
ec
o
m
p
o
s
ed
as
E
i
g
en
v
alu
e
s
,
th
e
E
u
cl
id
ea
n
E
i
g
en
-
d
i
s
tan
ce
b
et
w
ee
n
th
e
q
u
er
y
i
m
a
g
e
an
d
ev
er
y
o
th
er
tr
ain
i
n
g
i
m
a
g
e
in
t
h
e
d
at
ab
ase
is
ca
lcu
lated
as i
n
E
q
u
at
io
n
(
3
)
.
‖
‖
√
∑
)
)
)
(
3
)
3
.
3
.
R
et
riev
a
l by
t
ex
t
B
y
i
n
co
r
p
o
r
atin
g
„
ta
g
s
‟
to
t
h
e
C
B
I
R
s
y
s
te
m
,
t
h
e
co
m
p
u
tati
o
n
ti
m
e
ca
n
b
e
r
ed
u
ce
d
f
o
r
s
u
b
s
eq
u
e
n
t
s
ea
r
ch
es
b
y
f
ir
s
t
tag
g
i
n
g
a
s
in
g
le
o
r
m
u
ltip
le
r
ef
er
e
n
ce
i
m
a
g
es.
T
h
e
q
u
er
y
-
p
r
o
ce
s
s
i
n
g
m
o
d
u
le
th
en
p
r
o
ce
ed
s
to
r
etr
iev
e
t
h
e
m
o
s
t
s
i
m
ilar
i
m
ag
e
s
w
i
th
r
esp
ec
t
to
th
e
q
u
e
r
y
i
m
ag
e
a
n
d
th
e
m
o
s
t
s
i
m
ilar
i
m
a
g
es
w
il
l
t
h
en
b
e
class
i
f
ied
in
to
th
e
s
a
m
e
tag
ca
teg
o
r
y
d
u
r
in
g
au
to
m
atic
tag
g
i
n
g
.
O
n
s
u
b
s
eq
u
en
t
s
ea
r
c
h
es,
t
h
e
u
s
er
m
a
y
s
i
m
p
l
y
in
p
u
t
a
p
r
ev
io
u
s
l
y
d
ef
in
ed
ta
g
an
d
th
e
s
y
s
te
m
p
r
o
ce
ed
s
r
etr
ie
v
e
t
h
e
m
o
s
t
s
i
m
ilar
i
m
a
g
es
w
it
h
t
h
e
s
p
ec
i
f
ied
ta
g
w
it
h
o
u
t p
er
f
o
r
m
i
n
g
e
x
tr
ac
tio
n
an
d
s
i
m
ilar
it
y
co
m
p
u
tat
io
n
.
Au
to
-
tag
g
i
n
g
w
o
r
k
s
al
m
o
s
t
s
i
m
ilar
l
y
f
o
r
r
etr
iev
al
b
y
f
ea
t
u
r
e
v
ec
to
r
s
an
d
f
ac
ial
r
ec
o
g
n
i
tio
n
s
i
n
ce
th
e
o
u
tp
u
t
o
f
b
o
th
p
r
o
ce
s
s
b
lo
ck
s
ar
e
a
s
er
ie
s
o
f
r
elev
a
n
t
i
m
ag
es.
T
ag
g
in
g
i
s
d
o
n
e
w
i
th
a
r
ef
er
en
ce
tab
le,
co
n
tain
i
n
g
in
f
o
r
m
atio
n
th
at
is
u
p
d
ated
b
y
th
e
ap
p
licatio
n
d
u
r
in
g
r
u
n
ti
m
e
an
d
is
n
ee
d
ed
to
r
etr
iev
e
im
a
g
es
q
u
ick
l
y
w
it
h
o
u
t
h
av
in
g
to
ex
t
r
ac
t
an
d
in
s
p
ec
t
th
e
e
m
b
ed
d
e
d
tag
s
o
f
ev
er
y
i
m
a
g
e
in
t
h
e
d
atab
ase.
Na
m
es
a
n
d
n
o
r
m
al
tag
s
ar
e
tr
ea
ted
as
s
e
p
ar
ate
en
titi
e
s
a
n
d
h
e
n
c
e
h
av
e
th
eir
o
w
n
r
esp
ec
ti
v
e
r
e
f
er
e
n
ce
tab
les.
Fo
r
b
o
th
s
ea
r
ch
es,
r
elev
a
n
t
i
m
a
g
es
r
etr
i
ev
ed
ar
e
au
to
m
atica
ll
y
ta
g
g
ed
ac
co
r
d
in
g
to
u
s
er
in
p
u
t.
Du
r
i
n
g
au
to
tag
g
i
n
g
,
th
e
E
XI
F
m
eta
i
n
f
o
r
m
atio
n
o
f
th
e
f
ile
it
s
elf
i
s
ch
a
n
g
ed
to
ad
d
t
h
e
in
p
u
t
ta
g
,
th
e
n
,
th
i
s
i
n
f
o
r
m
atio
n
w
ill
b
e
ad
d
ed
to
th
e
r
ef
er
e
n
ce
tab
le.
T
o
r
etr
iev
e
ta
g
g
ed
i
m
ag
e
s
,
t
h
e
ap
p
lic
atio
n
w
i
ll
s
ca
n
t
h
o
u
g
h
t
h
e
r
e
f
er
en
ce
tab
le
a
n
d
g
et
th
e
i
m
a
g
e
p
at
h
.
Sin
ce
t
h
e
ta
g
m
a
y
v
ar
y
ac
co
r
d
in
g
to
u
s
er
p
r
ef
er
en
ce
,
t
h
e
E
XI
F
m
etad
at
a
is
ch
ec
k
ed
.
I
f
t
h
e
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
C
o
n
ten
t
-
b
a
s
ed
I
ma
g
e
R
etri
ev
a
l S
ystem
F
o
r
A
n
I
ma
g
e
G
a
ller
y
S
ea
r
ch
A
p
p
lica
tio
n
(
N
ico
le
Th
a
m
)
1907
in
f
o
r
m
atio
n
o
f
t
h
at
p
ar
ticu
lar
i
m
a
g
e
n
o
lo
n
g
er
co
n
tai
n
s
th
e
co
r
r
esp
o
n
d
in
g
s
ea
r
c
h
q
u
er
y
o
r
if
t
h
e
i
m
a
g
e
it
s
el
f
n
o
lo
n
g
er
e
x
is
t
s
,
th
e
co
r
r
esp
o
n
d
in
g
r
o
w
o
f
i
n
f
o
r
m
atio
n
w
il
l b
e
r
em
o
v
ed
.
4.
RE
SU
L
T
S
4
.
1
.
P
er
f
o
r
m
a
nce
o
f
CB
I
R
by
f
ea
t
ure
v
ec
t
o
rs
T
o
r
em
o
v
e
th
e
s
u
b
j
ec
tiv
it
y
o
f
h
u
m
a
n
p
er
ce
p
tio
n
i
n
clas
s
i
f
y
i
n
g
i
m
a
g
es,
th
e
C
OI
L
-
1
0
0
d
ata
s
et
is
u
s
e
d
to
test
t
h
e
r
eliab
ili
t
y
o
f
t
h
e
C
B
I
R
s
y
s
te
m
.
T
h
e
r
es
u
lt
i
s
o
b
ta
in
ed
w
it
h
cl
u
s
ter
r
ad
iu
s
o
r
d
if
f
er
en
ce
th
r
es
h
o
ld
o
f
2
0
.
8
o
b
j
ec
ts
as
s
h
o
w
n
i
n
Fi
g
u
r
e
3
ar
e
s
elec
ted
f
r
o
m
th
e
d
ataset
as
q
u
er
y
i
m
a
g
e
an
d
ar
e
ch
o
s
en
b
y
th
eir
v
ar
y
i
n
g
d
if
f
ic
u
lt
y
to
d
if
f
er
e
n
ti
ate
r
an
g
i
n
g
f
r
o
m
a
(
ea
s
ie
s
t)
to
h
(
m
o
s
t d
i
f
f
icu
l
t)
.
Fig
u
r
e
3
.
L
is
t o
f
q
u
er
y
i
m
a
g
es
T
ab
le
1
.
R
etr
iev
al
P
er
f
o
r
m
a
n
c
e
f
o
r
d
if
f
er
e
n
ce
T
h
r
esh
o
ld
o
f
2
0
A
b
C
d
E
f
g
h
N
o
.
o
f
r
e
l
e
v
a
n
t
i
m
a
g
e
s
r
e
t
r
i
e
v
e
d
72
72
72
62
58
72
66
66
P
r
e
c
i
si
o
n
(
%)
9
8
.
6
9
7
.
3
6
4
.
9
1
0
0
1
0
0
6
5
.
5
2
6
.
7
3
2
.
7
R
e
c
a
l
l
o
r
S
e
n
si
t
i
v
i
t
y
(
%)
1
0
0
1
0
0
1
0
0
8
6
.
1
8
0
.
6
1
0
0
9
1
.
7
9
1
.
7
S
p
e
c
i
f
i
c
i
t
y
(
%)
9
9
.
9
9
9
.
9
9
9
.
5
1
0
0
1
0
0
9
9
.
5
9
7
.
5
9
8
.
1
R
e
l
e
v
a
n
c
e
(
%)
9
8
.
6
9
7
.
3
6
4
.
9
1
1
6
.
1
1
2
4
6
5
.
5
2
9
.
1
3
5
.
7
A
c
c
u
r
a
c
y
(
%)
9
9
.
9
9
9
.
9
9
9
.
8
9
3
.
1
9
0
.
3
9
9
.
8
9
4
.
6
9
4
.
9
Fig
u
r
e
4
an
d
Fig
u
r
e
5
ar
e
th
e
R
OC
c
u
r
v
e
an
d
t
h
e
P
r
ec
is
io
n
-
R
ec
all
c
u
r
v
e
f
o
r
th
is
p
ar
tic
u
lar
s
et
o
f
q
u
er
y
i
m
ag
e
s
.
T
h
e
g
r
ap
h
s
ar
e
p
lo
tted
w
it
h
m
u
lt
ip
le
s
ets
o
f
p
r
ec
is
io
n
,
r
ec
all,
an
d
s
p
e
cif
icit
y
v
al
u
es
f
o
r
d
if
f
er
e
n
t t
h
r
esh
o
ld
v
a
lu
e
s
r
an
g
in
g
f
r
o
m
5
to
3
0
.
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
9
0
3
–
1912
1908
Fig
u
r
e
4
.
R
OC
C
u
r
v
e
Fig
u
r
e
5
.
P
r
ec
is
io
n
-
R
ec
al
l c
u
r
v
e
4
.
2
.
P
er
f
o
r
m
a
nce
o
f
CB
I
R
by
f
a
c
ia
l r
ec
o
g
nitio
n
I
m
ag
e
s
f
o
r
th
is
d
ataset
ar
e
p
r
o
d
u
ce
d
b
y
Dr
.
L
ib
o
r
Sp
ac
ek
f
r
o
m
th
e
Dep
ar
t
m
e
n
t
o
f
C
o
m
p
u
te
r
Scien
ce
o
f
U
n
iv
er
s
it
y
o
f
E
s
s
e
x
.
T
h
e
s
et
f
o
r
t
h
is
test
co
n
tai
n
5
6
i
m
a
g
es
o
f
8
d
i
f
f
er
en
t
in
d
i
v
i
d
u
als,
4
m
ale
s
a
n
d
4
f
e
m
ales.
Face
s
o
f
4
i
n
d
iv
id
u
als
ar
e
tr
ai
n
ed
w
it
h
4
i
m
ag
e
s
a
n
d
t
h
ese
i
m
a
g
e
s
ar
e
o
m
itte
d
f
r
o
m
th
e
g
aller
y
d
ata
b
ase
to
test
t
h
e
r
eliab
ilit
y
o
f
t
h
e
f
ac
ial
r
ec
o
g
n
itio
n
s
y
s
t
e
m
.
Fi
g
u
r
e
6
s
h
o
w
s
t
h
e
tr
ai
n
i
n
g
i
m
ag
e
s
f
o
r
f
o
u
r
d
if
f
er
e
n
t i
n
d
iv
id
u
als.
Fig
u
r
e
6
.
T
r
ain
in
g
i
m
ag
e
s
f
o
r
4
d
if
f
er
en
t i
n
d
i
v
id
u
al
s
T
ab
le
2
.
R
etr
iev
al
P
er
f
o
r
m
a
n
c
e
f
o
r
E
ig
en
d
is
ta
n
ce
T
h
r
esh
o
ld
o
f
2500
N
o
.
o
f
r
e
l
e
v
a
n
t
i
m
a
g
e
s re
t
r
i
e
v
e
d
P
r
e
c
i
si
o
n
(
%)
R
e
c
a
l
l
o
r
S
e
n
si
t
i
v
i
t
y
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
R
e
l
e
v
a
n
c
e
(
%)
A
c
c
u
r
a
c
y
(
%)
I
7
1
0
0
1
0
0
1
0
0
1
0
0
1
0
0
J
7
3
1
.
8
1
0
0
7
3
.
2
3
1
.
8
8
6
.
6
K
7
35
1
0
0
7
6
.
8
35
8
8
.
4
L
7
1
0
0
1
0
0
1
0
0
1
0
0
1
0
0
A
v
e
r
a
g
e
r
e
l
e
v
a
n
c
e
f
o
r
t
h
i
s
se
t
=
6
6
.
7
%
A
v
e
r
a
g
e
a
c
c
u
r
a
c
y
f
o
r
t
h
i
s
se
t
=
9
3
.
7
5
%
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
C
o
n
ten
t
-
b
a
s
ed
I
ma
g
e
R
etri
ev
a
l S
ystem
F
o
r
A
n
I
ma
g
e
G
a
ller
y
S
ea
r
ch
A
p
p
lica
tio
n
(
N
ico
le
Th
a
m
)
1909
Fig
u
r
e
7
s
h
o
w
s
o
m
e
f
al
s
e
d
ete
ctio
n
s
o
f
i
n
d
iv
id
u
a
ls
w
h
o
s
e
f
a
ce
s
ar
e
u
n
tr
ain
ed
.
No
te
t
h
at
s
u
b
j
ec
t
I
,
J
,
K,
L
in
t
h
i
s
ex
p
er
i
m
en
t i
s
n
a
m
ed
as „
o
n
e
‟
,
„
t
w
o
‟
,
„
t
h
r
ee
‟
,
an
d
„
f
o
u
r
‟
r
esp
ec
ti
v
el
y
in
t
h
e
d
ete
ctio
n
lab
els.
Fig
u
r
e
7
.
Fals
e
d
etec
tio
n
s
4
.
3
.
G
ra
ph
ica
l u
s
er
i
nte
rf
a
ce
Sh
o
w
n
in
F
ig
u
r
e
8
is
th
e
u
s
er
in
ter
f
ac
e
b
u
ilt
to
i
m
p
le
m
e
n
t
t
h
e
o
v
er
all
s
y
s
te
m
.
T
ab
le
3
d
e
s
cr
ib
es
its
f
u
n
ctio
n
alit
y
b
y
lab
eled
r
eg
io
n
s
i
n
m
o
r
e
d
etail.
Fig
u
r
e
8
.
Ov
er
v
ie
w
o
f
GUI
T
ab
le
3
.
GUI
F
u
n
ctio
n
ali
t
y
b
y
lab
eled
R
eg
io
n
s
R
e
g
i
o
n
F
u
n
c
t
i
o
n
A
S
e
l
e
c
t
g
a
l
l
e
r
y
d
a
t
a
b
a
se
t
o
b
e
se
a
r
c
h
e
d
.
B
L
o
a
d
i
mag
e
s w
i
t
h
o
u
t
p
e
r
f
o
r
mi
n
g
C
B
I
R
o
r
c
l
e
a
r
i
m
a
g
e
t
h
u
m
b
n
a
i
l
s.
C
L
i
st
v
i
e
w
d
i
sp
l
a
y
s i
mag
e
t
h
u
m
b
n
a
i
l
s o
f
l
o
a
d
e
d
o
r
r
e
t
r
i
e
v
e
d
i
mag
e
s.
A
l
l
t
h
u
m
b
n
a
i
l
s a
r
e
se
l
e
c
t
a
b
l
e
a
n
d
c
a
n
b
e
d
o
u
b
l
e
c
l
i
c
k
e
d
f
o
r
a
l
a
r
g
e
r
p
r
e
v
i
e
w
o
f
t
h
e
i
mag
e
.
D
S
e
l
e
c
t
se
a
r
c
h
mo
d
e
(
b
y
n
a
me
,
i
mag
e
,
o
r
t
a
g
)
a
n
d
a
d
d
t
a
g
t
o
se
l
e
c
t
e
d
i
m
a
g
e
.
E
A
p
r
e
v
i
e
w
o
f
t
h
e
q
u
e
r
y
i
ma
g
e
w
i
l
l
b
e
sh
o
w
n
h
e
r
e
w
h
e
n
se
a
r
c
h
i
n
g
b
y
i
mag
e
F
A
l
l
o
w
u
se
r
t
o
a
d
d
f
a
c
e
s a
n
d
t
h
e
i
r
c
o
r
r
e
sp
o
n
d
i
n
g
n
a
me
s
i
n
t
o
t
h
e
d
a
t
a
b
a
se
a
n
d
se
e
w
h
i
c
h
i
n
d
i
v
i
d
u
a
l
s a
r
e
d
e
t
e
c
t
e
d
i
n
t
h
e
se
l
e
c
t
e
d
i
mag
e
.
5.
DIS
CU
SS
I
O
N
T
h
e
s
y
s
te
m
co
n
s
is
ts
o
f
t
h
r
ee
m
ai
n
co
m
p
o
n
en
t
s
:
th
e
g
a
ller
y
in
ter
f
ac
e,
t
h
e
q
u
er
y
p
r
o
ce
s
s
i
n
g
m
o
d
u
le,
an
d
th
e
i
m
ag
e
d
atab
ase
a
n
d
allo
w
s
th
r
ee
m
o
d
es
o
f
s
ea
r
ch
e
s
:
b
y
a
r
ef
er
en
ce
i
m
a
g
e,
a
n
a
m
e,
o
r
a
p
r
ev
io
u
s
l
y
d
ef
in
ed
ta
g
.
Fig
u
r
e
1
s
h
o
w
s
th
e
to
p
-
lev
el
b
lo
ck
d
iag
r
a
m
f
o
r
t
h
e
o
v
er
all
s
y
s
te
m
.
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
9
0
3
–
1912
1910
5
.
1
.
P
er
f
o
r
m
a
nce
a
na
ly
s
is
o
f
t
he
f
ea
t
ure
v
ec
t
o
r
T
h
e
d
escr
ip
to
r
is
s
h
o
w
n
to
h
av
e
v
er
y
h
i
g
h
ac
cu
r
ac
y
as
s
h
o
w
n
in
T
ab
le
1
as
it
is
a
b
le
to
r
etr
iev
e
al
m
o
s
t
all
i
n
s
ta
n
ce
s
o
f
t
h
e
o
b
jects
(
h
ig
h
r
ec
all)
w
h
ile
w
ee
d
i
n
g
o
u
t
a
m
aj
o
r
ity
o
f
u
n
r
elate
d
i
m
ag
e
s
a
m
o
n
g
all
th
e
7
1
2
8
im
ag
e
s
(
h
ig
h
s
p
ec
i
f
icit
y
)
.
Ho
w
e
v
er
,
in
s
o
m
e
i
n
s
t
an
ce
s
it
is
s
til
l
ea
s
y
to
co
n
f
u
s
e
th
e
q
u
er
y
i
m
ag
e
w
it
h
o
t
h
er
u
n
r
elate
d
o
b
j
ec
ts
(
m
ed
iu
m
r
ele
v
a
n
ce
)
in
t
h
e
d
ata
b
ase
w
h
e
n
t
h
e
s
h
ap
e
i
s
i
n
co
n
s
is
ten
t
w
it
h
d
i
f
f
er
en
t
p
er
s
p
ec
tiv
es
an
d
w
h
e
n
th
er
e
a
r
e
im
a
g
es
w
it
h
s
i
m
ilar
co
lo
u
r
an
d
tex
t
u
r
e
in
th
e
d
atab
ase.
T
h
e
r
ea
s
o
n
f
o
r
th
is
i
s
b
ec
au
s
e
t
h
e
r
etr
ie
v
al
s
y
s
te
m
w
il
l
al
w
a
y
s
r
an
k
v
i
s
u
a
ll
y
s
i
m
i
lar
i
m
a
g
es
as
h
ig
h
er
i
n
s
i
m
ilar
it
y
to
h
e
lp
th
e
u
s
er
s
f
ilter
o
u
t i
m
a
g
es t
h
at
ar
e
v
er
y
lik
el
y
u
n
r
elate
d
r
ath
er
th
a
n
d
et
er
m
in
in
g
t
h
e
co
n
te
x
t o
r
d
ef
in
it
io
n
o
f
th
e
i
m
a
g
es.
5
.
2
.
Ana
ly
s
is
o
f
RO
C
a
nd
p
re
cisi
o
n
-
re
ca
ll
c
urv
es a
nd
s
elec
t
io
n o
f
t
hresh
o
ld
v
a
lue
B
ased
o
n
T
ab
le
1
,
th
er
e
ar
e
s
p
ec
ial
ca
s
es
s
u
ch
a
s
o
b
j
ec
t
d
an
d
e
w
h
er
e
t
h
e
p
r
ec
is
io
n
v
a
l
u
e
i
s
1
0
0
%
w
h
ile
r
ec
all
is
less
t
h
an
1
0
0
%.
T
h
is
is
lik
el
y
b
ec
a
u
s
e
s
o
m
e
o
f
th
e
o
th
er
s
i
m
ilar
i
m
a
g
es
in
th
e
q
u
eu
e
d
id
n
o
t
m
ak
e
i
t
in
to
t
h
e
f
ir
s
t
clu
s
ter
d
u
e
to
a
lo
w
d
if
f
er
en
ce
t
h
r
es
h
o
ld
.
W
ith
h
i
g
h
er
t
h
r
es
h
o
ld
s
,
th
e
s
e
n
s
iti
v
it
y
o
r
av
er
ag
e
r
ec
all
w
il
l
b
e
h
i
g
h
er
,
h
o
w
ev
er
th
e
s
p
ec
i
f
icit
y
w
i
ll
b
e
lo
w
er
as
ca
n
b
e
s
ee
n
i
n
Fi
g
u
r
e
3
.
T
h
is
tr
ad
e
-
o
f
f
is
also
ap
p
ar
en
t
i
n
th
e
p
r
ec
is
i
o
n
-
r
ec
all
c
u
r
v
e
w
h
er
e
t
h
e
m
o
r
e
w
e
tr
y
to
in
cr
ea
s
e
r
ec
all
to
f
in
d
m
o
r
e
i
n
s
ta
n
ce
s
o
f
r
elev
an
t i
m
a
g
e
s
,
th
e
m
o
r
e
t
h
e
p
r
ec
is
io
n
v
a
lu
e
d
ec
r
ea
s
es.
I
t
is
o
b
s
er
v
ed
t
h
at
a
cl
u
s
ter
r
a
d
iu
s
o
r
d
if
f
er
e
n
ce
t
h
r
es
h
o
ld
b
et
w
ee
n
2
0
a
n
d
2
5
is
p
r
ef
er
r
ed
as
t
h
e
la
s
t
r
elev
an
t
i
m
ag
e
i
n
t
h
e
d
ataset
w
il
l
te
n
d
h
a
v
e
d
e
v
iatio
n
v
alu
e
s
i
n
b
et
w
ee
n
t
h
at
r
a
n
g
e.
T
h
o
u
g
h
f
o
r
m
o
r
e
d
iv
er
s
e
i
m
a
g
es
h
av
in
g
m
o
r
e
co
lo
u
r
s
an
d
d
iv
er
s
e
b
ac
k
g
r
o
u
n
d
s
,
lar
g
er
th
r
es
h
o
ld
v
al
u
es
w
it
h
in
t
h
at
r
an
g
e
ar
e
m
o
r
e
ef
f
ec
tiv
e.
I
t
is
p
o
s
s
ib
le
to
a
s
s
es
s
th
e
p
r
ec
is
io
n
-
r
ec
all
tr
ad
e
-
o
f
f
s
u
s
in
g
t
h
e
w
ei
g
h
ted
h
ar
m
o
n
ic
m
ea
n
i
n
E
q
u
atio
n
(
4
)
w
h
er
e
F
is
t
h
e
w
eig
h
ted
h
ar
m
o
n
ic
m
ea
n
,
P
an
d
R
is
th
e
p
r
ec
is
io
n
an
d
r
ec
all
v
alu
e
r
esp
ec
ti
v
el
y
f
o
r
th
at
p
ar
ticu
lar
t
h
r
es
h
o
ld
an
d
α
is
th
e
w
ei
g
h
t to
e
m
p
h
as
ize
m
o
r
e
o
n
eith
er
p
r
ec
is
io
n
o
r
r
ec
all.
)
(
4
)
T
o
ch
o
o
s
e
th
e
m
o
s
t
s
u
itab
le
t
h
r
es
h
o
ld
,
a
th
r
esh
o
ld
v
al
u
e
co
r
r
esp
o
n
d
in
g
to
th
e
p
air
o
f
p
r
ec
is
io
n
a
n
d
r
ec
all
v
alu
e
th
at
h
as
t
h
e
h
ig
h
e
s
t
F
s
co
r
e
s
h
o
u
ld
b
e
ch
o
s
e
n
.
I
f
p
r
ec
is
io
n
is
m
o
r
e
i
m
p
o
r
tan
t
t
o
th
e
u
s
er
,
α
s
h
o
u
ld
b
e
lar
g
er
an
d
co
n
v
er
s
e
l
y
i
f
r
e
ca
ll
is
m
o
r
e
i
m
p
o
r
tan
t,
α
s
h
o
u
ld
b
e
lo
w
er
,
alt
h
o
u
g
h
in
t
h
i
s
co
n
tex
t
i
t
en
t
ir
el
y
d
ep
en
d
s
o
n
u
s
er
p
r
ef
er
e
n
ce
.
A
m
o
r
e
b
alan
ce
d
F1
is
d
e
s
cr
ib
ed
in
E
q
u
at
io
n
(
5
)
an
d
i
s
m
o
r
e
co
m
m
o
n
l
y
u
s
ed
b
y
r
esear
ch
er
s
f
ac
i
n
g
th
e
p
r
ec
is
io
n
-
r
ec
all
tr
ad
e
-
o
f
f
p
r
o
b
lem
.
(
5)
5
.
3
.
P
er
f
o
r
m
a
nce
a
na
ly
s
is
o
f
t
he
f
a
cia
l
r
ec
o
g
nitio
n
s
y
s
t
e
m
T
h
e
r
etr
iev
al
p
er
f
o
r
m
a
n
ce
s
h
o
w
s
a
h
ig
h
av
er
a
g
e
ac
cu
r
ac
y
h
o
w
e
v
er
th
e
r
es
u
lt
s
ar
e
n
o
t
v
er
y
co
n
s
is
ten
t
f
o
r
d
if
f
er
en
t
in
d
i
v
id
u
als
a
n
d
th
is
i
s
d
u
e
to
s
e
v
er
al
f
ac
to
r
s
.
No
tice
i
n
Fi
g
u
r
e
5
th
a
t
in
d
i
v
id
u
al
s
J
an
d
K
h
a
v
e
m
o
r
e
v
ar
ia
tio
n
s
i
n
f
a
cial
ex
p
r
es
s
io
n
s
i
n
tr
ai
n
i
n
g
i
m
ag
e
s
.
I
n
ap
p
licatio
n
,
i
t
i
s
i
m
p
o
r
ta
n
t
to
ad
d
m
o
r
e
v
ar
iatio
n
s
i
n
e
x
p
r
ess
io
n
s
f
o
r
ea
ch
i
n
d
iv
id
u
al
an
d
in
d
if
f
er
en
t
li
g
h
t
in
g
co
n
d
itio
n
s
a
s
i
m
a
g
e
s
o
f
p
eo
p
le
ten
d
to
h
av
e
d
i
f
f
er
en
t
e
x
p
r
ess
io
n
s
o
n
d
if
f
er
e
n
t
s
ess
io
n
s
.
Ho
w
e
v
er
,
in
t
h
is
ca
s
e
it
ca
u
s
es
s
o
m
e
co
n
f
u
s
io
n
w
i
th
o
t
h
er
in
d
iv
id
u
als
w
h
o
s
e
f
ac
e
s
ar
e
n
o
t
tr
ain
ed
.
T
h
is
is
o
n
e
o
f
th
e
li
m
itatio
n
s
o
f
t
h
e
E
ig
e
n
f
ac
e
ap
p
r
o
ac
h
as
it
ten
d
s
to
lo
o
k
at
th
e
tr
ai
n
i
n
g
d
ataset
as
a
w
h
o
le
r
ath
er
t
h
a
n
an
a
l
y
zi
n
g
ea
ch
f
ac
e
i
n
t
h
e
s
e
t
in
d
i
v
id
u
all
y
.
Ho
w
ev
er
,
t
h
e
r
ec
o
g
n
itio
n
ca
n
b
e
i
m
p
r
o
v
ed
b
y
ad
d
in
g
m
o
r
e
in
d
iv
id
u
al
f
ac
es
i
n
to
th
e
d
ataset
to
r
ed
u
ce
th
e
n
u
m
b
er
o
f
u
n
k
n
o
w
n
i
n
d
iv
id
u
als o
r
b
y
i
n
c
r
ea
s
in
g
th
e
E
i
g
e
n
d
is
tan
ce
t
h
r
e
s
h
o
ld
to
in
cr
ea
s
e
p
r
ec
is
io
n
.
6.
CO
NCLU
SI
O
N
A
ll
t
h
e
o
b
j
ec
tiv
es
h
av
e
b
ee
n
m
et
a
n
d
t
h
e
t
h
r
ee
d
if
f
e
r
en
t
s
ea
r
c
h
m
o
d
e
s
w
er
e
s
u
cc
ess
f
u
l
l
y
i
m
p
le
m
en
ted
o
n
a
n
i
m
a
g
e
g
a
l
ler
y
s
o
f
t
w
ar
e.
Fo
r
r
etr
iev
al
b
y
q
u
er
y
i
m
a
g
e,
t
h
e
f
ea
tu
r
e
v
e
cto
r
s
ar
e
ex
tr
ac
ted
u
s
i
n
g
SIM
P
L
E
-
C
E
D
D
an
d
r
es
u
lts
s
h
o
w
a
m
ed
i
u
m
av
er
a
g
e
r
elev
an
ce
o
f
7
8
.
9
% a
n
d
a
h
i
g
h
ac
cu
r
ac
y
o
f
9
6
.
5
%.
T
h
is
s
h
o
w
s
t
h
at
t
h
e
s
y
s
te
m
i
s
ef
f
ec
ti
v
e
i
n
f
ilter
i
n
g
o
u
t
u
n
w
a
n
ted
i
m
ag
e
s
h
o
w
ev
er
it
is
p
o
s
s
ib
le
to
f
u
r
th
er
i
m
p
r
o
v
e
th
e
o
v
er
all
r
ele
v
an
ce
b
y
i
n
co
r
p
o
r
atin
g
clas
s
if
icatio
n
alg
o
r
it
h
m
s
to
class
i
f
y
v
i
s
u
al
l
y
d
if
f
er
en
t
i
m
a
g
es
so
th
at
t
h
e
r
etr
ie
v
al
r
es
u
lts
ca
n
b
e
i
m
p
r
o
v
ed
b
y
tak
i
n
g
i
n
to
ac
co
u
n
t
t
h
e
co
n
tex
t
o
r
d
ef
in
i
tio
n
o
f
t
h
e
i
m
a
g
es
th
e
m
s
el
v
es.
Ne
x
t,
r
etr
iev
al
b
y
f
ac
ial
r
ec
o
g
n
itio
n
i
s
b
u
il
t
ar
o
u
n
d
t
h
e
E
ig
e
n
f
ac
es
m
et
h
o
d
.
T
h
e
r
esu
lts
s
h
o
w
a
n
av
er
ag
e
r
ele
v
a
n
ce
o
f
6
6
.
7
%
a
n
d
an
ac
c
u
r
ac
y
o
f
9
3
.
7
%,
h
o
w
e
v
er
i
n
s
o
m
e
ca
s
es
t
h
e
r
es
u
l
ts
ar
e
n
o
t
d
esira
b
le
d
u
e
to
f
alse
d
etec
tio
n
s
f
o
r
u
n
k
n
o
w
n
f
ac
e
s
an
d
th
is
is
a
n
i
n
h
er
e
n
t
w
ea
k
n
e
s
s
i
n
th
e
E
i
g
e
n
f
ac
e
s
m
e
th
o
d
.
As
f
u
tu
r
e
w
o
r
k
,
it
is
r
ec
o
m
m
en
d
e
d
to
ex
p
lo
r
e
o
th
er
m
et
h
o
d
s
o
f
f
ac
ial
r
ec
o
g
n
it
io
n
s
u
c
h
as
L
B
P
H
to
d
eter
m
i
n
e
i
f
it
is
p
o
s
s
ib
le
to
i
n
cr
ea
s
e
t
h
e
p
r
ec
is
io
n
(
an
d
t
h
u
s
,
r
elev
an
ce
)
o
f
t
h
e
r
etr
iev
al
r
e
s
u
l
t
s
.
A
n
d
las
tl
y
,
th
e
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
C
o
n
ten
t
-
b
a
s
ed
I
ma
g
e
R
etri
ev
a
l S
ystem
F
o
r
A
n
I
ma
g
e
G
a
ller
y
S
ea
r
ch
A
p
p
lica
tio
n
(
N
ico
le
Th
a
m
)
1911
co
m
p
u
tatio
n
al
r
eq
u
ir
e
m
e
n
ts
o
f
s
u
b
s
eq
u
en
t
s
ea
r
c
h
es
w
er
e
r
e
d
u
ce
d
b
y
i
n
te
g
r
atin
g
C
B
I
R
with
T
B
I
R
u
s
i
n
g
th
e
co
n
ce
p
t o
f
r
ef
er
en
ce
tab
les
.
RE
F
E
R
E
NC
E
S
[1
]
L
o
n
g
,
F
.
,
Zh
a
n
g
,
H.,
&
F
e
n
g
,
D.
D.,
(
2
0
0
3
),
“
F
u
n
d
a
m
e
n
tals
o
f
c
o
n
ten
t
-
b
a
se
d
im
a
g
e
re
tri
e
v
a
l”
,
In
M
u
lt
ime
d
i
a
In
fo
rm
a
t
io
n
Retrie
v
a
l
a
n
d
M
a
n
a
g
e
me
n
t
,
p
p
.
1
-
2
6
,
S
p
rin
g
e
r
Be
rli
n
He
id
e
lb
e
rg
.
[2
]
R
Zh
a
n
g
,
Y.
J.
,
(
2
0
0
5
),
A
d
v
a
n
c
e
d
T
e
c
h
n
iq
u
e
s f
o
r
Ob
jec
t
-
Ba
se
d
Im
a
g
e
Re
tri
e
v
a
l.
[3
]
Y.
K.
J.
K.
Zu
k
u
a
n
W
EI,
Ho
n
g
y
e
o
n
KIM,
“
A
n
e
ff
icie
n
t
c
o
n
ten
t
b
a
se
d
ima
g
e
re
tri
e
v
a
l
sc
h
e
m
e
,
”
T
EL
KOM
NIKA
(
T
e
lec
o
mm
u
n
ica
ti
o
n
Co
mp
u
ti
n
g
El
e
c
tro
n
ics
a
n
d
C
o
n
tro
l)
,
v
o
l.
1
1
,
n
o
.
1
1
,
p
p
.
6
9
8
6
-
6
9
9
1
,
N
o
v
e
m
b
e
r
2
0
1
3
.
[4
]
Ja
lab
,
H.
A
.
(
2
0
1
1
,
S
e
p
tem
b
e
r),
“
Im
a
g
e
re
tri
e
v
a
l
s
y
ste
m
b
a
se
d
o
n
c
o
lo
r
lay
o
u
t
d
e
sc
rip
to
r
a
n
d
G
a
b
o
r
f
il
ters
”
.
[5
]
Ja
y
a
m
a
la
Ku
m
a
r
P
a
ti
l,
Ra
j
Ku
m
a
r,
(2
0
1
3
),
“
P
lan
t
L
e
a
f
Dise
a
se
Im
a
g
e
R
e
tri
e
v
a
l
Us
in
g
Co
lo
r
M
o
m
e
n
ts”
,
IAE
S
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
Art
if
icia
l
In
telli
g
e
n
c
e
(
IJ
-
AI)
,
v
o
l.
2
,
n
o
.
1
,
p
p.
36
-
4
2
.
[6
]
Ch
a
tzic
h
risto
f
is,
S
.
A
.
,
&
Bo
u
tali
s,
Y.
S
.
,
(2
0
0
8
,
M
a
y
),
“
CEDD:
c
o
lo
u
r
a
n
d
e
d
g
e
d
irec
ti
v
it
y
d
e
sc
ri
p
to
r:
a
c
o
m
p
a
c
t
d
e
sc
rip
to
r
f
o
r
im
a
g
e
in
d
e
x
in
g
a
n
d
re
tri
e
v
a
l
”
.
[7
]
T
a
m
u
ra
,
H.,
M
o
ri,
S
.
,
&
Ya
m
a
w
a
k
i,
T
.
,
(1
9
7
8
),
“
T
e
x
tu
ra
l
f
e
a
t
u
re
s
c
o
r
re
sp
o
n
d
in
g
to
v
isu
a
l
p
e
rc
e
p
ti
o
n
”,
IE
E
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
y
ste
ms
,
M
a
n
,
a
n
d
Cy
b
e
rn
e
ti
c
s
,
v
o
l.
8
,
n
o
.
6
,
p
p
.
460
-
4
7
3
.
[8
]
S
im
o
n
c
e
ll
i,
E.
P
.
,
&
F
re
e
m
a
n
,
W
.
T
.
,
(1
9
9
5
,
Oc
to
b
e
r),
“
T
h
e
ste
e
ra
b
le
p
y
ra
m
id
:
a
f
le
x
ib
le
a
rc
h
it
e
c
tu
re
f
o
r
m
u
l
ti
-
sc
a
le d
e
riv
a
ti
v
e
c
o
m
p
u
tatio
n
”
,
In
ICIP
,
v
o
l.
3
,
p
p
.
4
4
4
-
4
4
7
.
[9
]
B.
S
.
M
a
n
ju
n
a
th
a
n
d
W
.
Y.
M
a
.
“
T
e
x
tu
re
f
e
a
tu
re
s
f
o
r
b
ro
w
sin
g
a
n
d
re
tri
e
v
a
l
o
f
larg
e
i
m
a
g
e
d
a
ta
”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
P
a
tt
e
rn
A
n
a
lys
is
a
n
d
M
a
c
h
in
e
In
telli
g
e
n
c
e
,
(S
p
e
c
ia
l
Iss
u
e
o
n
Dig
it
a
l
L
ib
ra
ries
),
v
o
l.
1
8
,
n
o
.
8
,
A
u
g
u
st 1
9
9
6
,
p
p
.
8
3
7
-
8
4
2
.
[1
0
]
Ka
u
p
p
i
n
e
n
,
H.,
S
e
p
p
a
n
e
n
,
T
.
,
&
P
ietik
a
in
e
n
,
M
,
(
1
9
9
5
)
,
“
A
n
e
x
p
e
ri
m
e
n
tal
c
o
m
p
a
riso
n
o
f
a
u
t
o
re
g
re
ss
iv
e
a
n
d
F
o
u
rier
-
b
a
se
d
d
e
sc
rip
t
o
rs
i
n
2
D
sh
a
p
e
c
las
sif
ica
ti
o
n
”
,
IEE
E
T
r
a
n
sa
c
ti
o
n
s
o
n
P
a
tt
e
rn
An
a
lys
is
a
n
d
M
a
c
h
in
e
In
telli
g
e
n
c
e
,
v
o
l.
1
7
,
n
o
.
2
,
p
p
.
201
-
2
0
7
.
[1
1
]
V
id
y
a
M
S
,
A
ru
l
K,
“
A
u
to
m
a
ted
A
tt
e
n
d
a
n
c
e
S
y
ste
m
T
h
ro
u
g
h
Ei
g
e
n
F
a
c
e
s
Us
in
g
I
m
a
g
e
P
ro
c
e
ss
in
g
”
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
In
fo
rm
a
ti
c
s
a
n
d
Co
mm
u
n
ica
ti
o
n
T
e
c
h
n
o
l
o
g
y
(
IJ
-
ICT
)
,
2
0
1
6
De
c
1
,
v
o
l
.
5
,
n
o
.
3
,
p
p
.
1
1
1
-
11
8.
[1
2
]
F
ish
e
r,
R.
A
.
,
(1
9
3
6
),
“
T
h
e
u
se
o
f
m
u
lt
ip
le
m
e
a
s
u
re
m
e
n
ts
in
tax
o
n
o
m
ic
p
ro
b
lem
s
”
,
An
n
a
ls
o
f
e
u
g
e
n
ics
,
v
o
l.
7
,
n
o
.
2
,
p
p
.
1
7
9
-
1
8
8
[1
3
]
A
h
o
n
e
n
,
T
.
,
Ha
d
id
,
A
.
,
&
P
ietik
ä
in
e
n
,
M
,
(2
0
0
4
),
“
F
a
c
e
re
c
o
g
n
i
ti
o
n
w
it
h
lo
c
a
l
b
in
a
ry
p
a
tt
e
rn
s”
,
C
o
mp
u
ter
v
isio
n
-
e
c
c
v
2
0
0
4
,
p
p
.
4
6
9
-
4
8
1
.
[1
4
]
T
a
ig
m
a
n
,
Y.,
Ya
n
g
,
M
.
,
Ra
n
z
a
to
,
M
.
A
.
,
&
W
o
lf
,
L
,
(2
0
1
4
),
“
De
e
p
f
a
c
e
:
Clo
sin
g
th
e
g
a
p
to
h
u
m
a
n
-
lev
e
l
p
e
rf
o
r
m
a
n
c
e
in
fa
c
e
v
e
ri
f
ic
a
ti
o
n
”
,
In
Pro
c
e
e
d
in
g
s
o
f
th
e
IEE
E
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
Vi
si
o
n
a
n
d
Pa
t
ter
n
Rec
o
g
n
it
io
n
,
p
p
.
1
7
0
1
-
1
7
0
8
.
[1
5
]
S
c
h
ro
f
f
,
F
.
,
Ka
len
ich
e
n
k
o
,
D.,
&
P
h
il
b
i
n
,
J,
(
2
0
1
5
)
,
“
F
a
c
e
n
e
t:
A
u
n
if
ied
e
m
b
e
d
d
in
g
f
o
r
fa
c
e
r
e
c
o
g
n
it
io
n
a
n
d
c
lu
ste
rin
g
”
,
In
Pr
o
c
e
e
d
in
g
s
o
f
th
e
IEE
E
C
o
n
fer
e
n
c
e
o
n
Co
m
p
u
ter
V
isio
n
a
n
d
P
a
tt
e
rn
Rec
o
g
n
it
io
n
,
p
p
.
8
1
5
-
8
2
3
).
[1
6
]
Nib
lac
k
,
C.
W
.
,
Ba
rb
e
r,
R.
,
Eq
u
it
z
,
W
.
,
F
li
c
k
n
e
r,
M
.
D.,
G
las
m
a
n
,
E.
H.
,
P
e
tk
o
v
ic,
D.,
.
.
.
&
Tau
b
in
,
G
,
(1
9
9
3
,
A
p
ril
),
“
QBIC
p
ro
jec
t:
q
u
e
ry
in
g
i
m
a
g
e
s
b
y
c
o
n
ten
t,
u
si
n
g
c
o
lo
r,
tex
tu
re
,
a
n
d
sh
a
p
e
”
,
I
n
IS
&
T
/S
PIE
's
S
y
mp
o
si
u
m
o
n
El
e
c
tro
n
ic
I
ma
g
i
n
g
:
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
,
p
p
.
1
7
3
-
1
8
7
,
In
ter
n
a
ti
o
n
a
l
S
o
c
iety
f
o
r
Op
ti
c
s
a
n
d
P
h
o
t
o
n
ics
.
[1
7
]
T
a
m
u
ra
,
H.,
M
o
ri,
S
.
,
&
Ya
m
a
w
a
k
i,
T
,
(1
9
7
8
),
“
T
e
x
tu
ra
l
f
e
a
t
u
re
s
c
o
r
re
sp
o
n
d
i
n
g
to
v
isu
a
l
p
e
rc
e
p
ti
o
n
”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
y
ste
ms
,
M
a
n
,
a
n
d
Cy
b
e
rn
e
ti
c
s
,
v
o
l.
8
,
n
o
.
6
,
p
p
.
460
-
4
7
3
.
[1
8
]
Ia
k
o
v
id
o
u
,
C.
,
A
n
a
g
n
o
sto
p
o
u
lo
s
,
N.,
Ka
p
o
u
tsis,
A
.
C.
,
Bo
u
tali
s,
Y.,
&
Ch
a
tzic
h
risto
f
is,
S
.
A
.
,
(2
0
1
4
,
J
u
n
e
)
,
“
S
e
a
rc
h
in
g
i
m
a
g
e
s
w
it
h
M
P
EG
-
7
(&
m
p
e
g
-
7
-
li
k
e
)
p
o
w
e
re
d
lo
c
a
li
z
e
d
d
e
sc
rip
to
rs:
th
e
S
IM
P
L
E
a
n
sw
e
r
to
e
ffe
c
ti
v
e
c
o
n
ten
t
b
a
se
d
im
a
g
e
re
tri
e
v
a
l”,
I
n
Co
n
te
n
t
-
Ba
se
d
M
u
lt
ime
d
ia
In
d
e
x
in
g
(
CBM
I),
2
0
1
4
1
2
th
In
ter
n
a
ti
o
n
a
l
W
o
rk
sh
o
p
on
,
p
p
.
1
-
6
,
IEE
E.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Nic
o
le
Th
a
m
La
y
M
e
i
g
ra
d
u
a
ted
w
it
h
a
B
a
c
h
e
lo
r‟s
d
e
g
re
e
in
El
e
c
tro
n
ic
E
n
g
in
e
e
rin
g
f
ro
m
Un
iv
e
rsiti
T
e
k
n
o
lo
g
i
M
a
lay
sia
in
2
0
1
7
.
He
r
w
o
rk
in
tere
sts
a
re
p
ri
m
a
ril
y
in
th
e
a
re
a
o
f
ro
b
o
ti
c
s
a
n
d
a
u
to
n
o
m
o
u
s m
a
c
h
in
e
s.
S
h
e
c
u
rre
n
tl
y
re
sid
e
s in
S
a
b
a
h
,
M
a
lay
sia
S
y
a
h
m
i
S
y
a
h
ir
a
n
B
in
Ah
m
a
d
Rid
z
u
a
n
is
c
u
rre
n
tl
y
d
o
in
g
h
is
P
h
D
a
t
th
e
F
a
c
u
lt
y
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
,
Un
iv
e
rsiti
T
e
k
n
o
lo
g
i
M
a
la
y
sia
.
He
o
b
tain
e
d
h
is
Ba
c
h
e
lo
r‟s
d
e
g
re
e
in
El
e
c
tri
c
a
l,
El
e
c
tro
n
ic
a
n
d
A
u
to
m
a
ti
o
n
E
n
g
in
e
e
rin
g
f
ro
m
Un
iv
e
rsité
d
e
F
ra
n
c
h
e
-
Co
m
té
,
Be
sa
n
ç
o
n
a
n
d
h
is
M
a
ste
r‟s
d
e
g
re
e
in
Ne
tw
o
rk
,
T
e
l
e
c
o
m
m
u
n
ica
ti
o
n
,
M
u
lt
im
e
d
ia
a
n
d
A
u
to
m
a
ti
o
n
f
ro
m
Un
iv
e
rsité
d
e
P
o
it
iers
.
His res
e
a
rc
h
in
tere
st
is
p
rim
a
ril
y
in
i
m
a
g
e
p
ro
c
e
ss
in
g
f
ield
.
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
9
0
3
–
1912
1912
Dr
Z
a
i
d
O
m
a
r
is
a
s
e
n
io
r
lec
tu
re
r
a
t
th
e
F
a
c
u
lt
y
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
,
Un
iv
e
rsiti
T
e
k
n
o
lo
g
i
M
a
la
y
sia
.
He
o
b
tain
e
d
h
is
P
h
D
i
n
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
f
ro
m
I
m
p
e
rial
Co
ll
e
g
e
L
o
n
d
o
n
i
n
2
0
1
2
.
His res
e
a
rc
h
in
tere
sts in
c
lu
d
e
im
a
g
e
p
ro
c
e
ss
in
g
,
m
a
c
h
in
e
lea
rn
in
g
,
a
n
d
m
e
d
ica
l
ima
g
in
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.