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.
11
,
No
.
5
,
Octo
b
er
2
0
2
1
,
p
p
.
4
4
2
3
~
4
4
3
0
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
1
1
i
5
.
pp
4
4
2
3
-
44
30
4423
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
p
-
no
r
m
s o
f
histog
ra
m
of
orient
ed g
ra
dients f
o
r
X
-
ra
y
i
m
a
g
es
Nuha
H
.
H
a
m
a
da
,
F
a
t
en
F
.
K
ha
rba
t
Co
ll
e
g
e
o
f
En
g
in
e
e
rin
g
,
A
l
A
in
U
n
iv
e
rsity
,
A
b
u
Dh
a
b
i,
U
n
it
e
d
A
ra
b
E
m
irate
s
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
n
1
8
,
2
0
2
0
R
ev
i
s
ed
Dec
8
,
2
0
2
0
A
cc
ep
ted
Ma
r
2
2
,
2
0
2
1
L
e
b
e
s
g
u
e
sp
a
c
e
s
(
L
p
o
v
e
r
R
n
)
p
la
y
a
sig
n
i
f
ica
n
t
ro
le i
n
m
a
th
e
m
a
ti
c
a
l
a
n
a
l
y
sis.
T
h
e
y
a
re
w
id
e
l
y
u
se
d
in
m
a
c
h
in
e
lea
rn
i
n
g
a
n
d
a
rti
f
icia
l
in
tell
ig
e
n
c
e
to
m
a
x
i
m
iz
e
p
e
rf
o
r
m
a
n
c
e
o
r
m
in
im
iz
e
e
rro
r.
T
h
e
w
e
ll
-
k
n
o
w
n
h
i
sto
g
ra
m
o
f
o
rien
ted
g
ra
d
ien
ts
(HO
G
)
a
l
g
o
rit
h
m
a
p
p
li
e
s
th
e
2
-
n
o
rm
(Eu
c
li
d
e
a
n
d
istan
c
e
)
to
d
e
tec
t
f
e
a
tu
re
s
in
im
a
g
e
s.
In
th
is
p
a
p
e
r,
w
e
a
p
p
ly
d
iff
e
re
n
t
p
-
n
o
rm
v
a
lu
e
s
to
id
e
n
ti
fy
th
e
im
p
a
c
t
th
a
t
c
h
a
n
g
in
g
th
e
se
n
o
rm
s
h
a
s
o
n
t
h
e
o
rig
in
a
l
a
lg
o
rit
h
m
.
T
h
e
a
i
m
o
f
th
is
m
o
d
ifi
c
a
ti
o
n
is
to
a
c
h
iev
e
b
e
tt
e
r
p
e
rf
o
rm
a
n
c
e
in
c
las
si
fy
in
g
X
-
ra
y
m
e
d
ica
l
ima
g
e
s
re
late
d
to
o
f
COV
ID
-
1
9
p
a
ti
e
n
ts
.
T
h
e
e
ff
ici
e
n
c
y
o
f
th
e
p
-
HO
G
a
lg
o
ri
th
m
is
c
o
m
p
a
re
d
w
it
h
th
e
o
rig
in
a
l
HO
G
d
e
sc
rip
to
r
u
si
n
g
a
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
im
p
le
m
e
n
ted
in
P
y
th
o
n
.
T
h
e
re
su
lt
s
o
f
th
e
c
o
m
p
a
riso
n
s
a
re
p
ro
m
isin
g
,
a
n
d
th
e
p
-
HO
G
a
l
g
o
rit
h
m
sh
o
w
s
g
re
a
ter e
ff
icie
n
c
y
in
m
o
st ca
se
s.
K
ey
w
o
r
d
s
:
Featu
r
e
d
escr
ip
to
r
HOG
alg
o
r
it
h
m
I
m
ag
e
cla
s
s
i
f
icatio
n
p
-
n
o
r
m
s
X
-
r
a
y
i
m
a
g
es
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Nu
h
a
H.
Ha
m
ad
a
C
o
lleg
e
o
f
E
n
g
in
ee
r
i
n
g
A
l
Ain
U
n
i
v
er
s
it
y
Un
ited
A
r
ab
E
m
ir
ates
E
m
ail:
n
u
h
a.
h
a
m
ad
a
@
aa
u
.
ac
.
ae
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
h
is
to
g
r
a
m
o
f
o
r
ien
ted
g
r
ad
ien
ts
(
HOG)
is
a
w
ell
-
k
n
o
w
n
f
ea
tu
r
e
e
x
tr
ac
tio
n
al
g
o
r
ith
m
u
s
ed
esp
ec
iall
y
f
o
r
h
u
m
a
n
d
escr
ip
t
o
r
s
[
1
]
.
T
h
e
HOG
d
escr
ip
to
r
is
b
ased
o
n
th
e
lo
ca
tio
n
a
n
d
o
r
ien
tatio
n
o
f
t
h
e
ed
g
e.
T
h
e
liter
atu
r
e
h
as
u
s
ed
HOG
d
escr
ip
to
r
s
in
co
n
j
u
n
ctio
n
w
ith
li
n
ea
r
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
e
(
SVM)
class
i
f
ier
s
[
2
]
an
d
o
th
er
d
ata
m
i
n
in
g
al
g
o
r
ith
m
s
[
1
]
.
T
h
e
s
tr
en
g
t
h
o
f
th
e
HO
G
alg
o
r
ith
m
lies
in
it
s
ab
ilit
y
to
cap
tu
r
e
ed
g
e
an
d
g
r
ad
ien
t
in
f
o
r
m
at
io
n
w
h
i
le
d
ec
r
ea
s
in
g
t
h
e
w
ei
g
h
t
o
f
ir
r
elev
a
n
t
f
ea
tu
r
es
d
u
e
to
illu
m
i
n
atio
n
co
n
d
itio
n
s
[
3
]
.
Ho
w
e
v
er
,
m
a
n
y
r
esear
c
h
er
s
h
a
v
e
ai
m
ed
to
im
p
r
o
v
e
th
e
HOG
alg
o
r
it
h
m
t
o
en
h
an
ce
d
etec
tio
n
p
er
f
o
r
m
a
n
ce
i
n
ter
m
s
o
f
ac
c
u
r
ac
y
,
co
m
p
u
tatio
n
al
co
s
t
s
,
an
d
class
i
f
icatio
n
.
Fo
r
ex
a
m
p
l
e,
th
e
C
o
HOG
[
4
]
ap
p
r
o
ac
h
u
s
es
p
air
s
o
f
g
r
ad
ie
n
t
o
r
ien
tat
io
n
s
to
f
o
r
m
h
i
s
to
g
r
a
m
s
.
H
OG
-
L
B
P
,
w
h
ic
h
w
a
s
d
ev
elo
p
ed
in
[
5
]
an
d
[
3
]
,
co
m
b
in
e
s
a
lo
ca
l
b
in
ar
y
p
atter
n
(
L
B
P
)
an
d
HO
G
d
esc
r
ip
to
r
to
p
r
o
d
u
ce
b
etter
r
esu
l
ts
.
HOG
w
as
also
en
h
a
n
ce
d
b
y
a
co
m
p
le
m
e
n
tar
y
d
escr
ip
to
r
in
th
e
p
r
o
p
o
s
ed
eHO
G
al
g
o
r
ith
m
[
6
]
to
h
a
n
d
le
t
h
e
s
ca
le
v
ar
iatio
n
o
f
p
ed
estrian
s
.
I
n
[
7
]
,
th
e
a
u
th
o
r
s
r
ed
u
ce
d
th
e
d
i
m
e
n
s
io
n
s
o
f
t
h
e
HOG
f
ea
tu
r
es
b
y
co
m
b
i
n
i
n
g
HOG
an
d
g
r
ee
d
y
alg
o
r
ith
m
s
f
o
r
s
elec
ted
HOG
d
escr
ip
to
r
s
.
Oth
er
w
o
r
k
h
as
also
b
ee
n
co
n
d
u
cted
to
en
h
a
n
ce
HOG
f
ea
t
u
r
es,
in
cl
u
d
in
g
[
8
]
an
d
[
9
]
,
am
o
n
g
o
th
er
s
.
HOG
[
2
]
is
a
n
e
f
f
ec
t
iv
e
f
ea
t
u
r
e
d
escr
ip
to
r
tech
n
iq
u
e
t
h
at
co
m
p
u
tes
ed
g
e
d
ir
ec
tio
n
b
y
d
iv
id
in
g
a
n
i
m
a
g
e
i
n
to
b
lo
ck
s
f
r
o
m
w
h
ic
h
it
e
x
tr
ac
ts
th
e
h
i
s
to
g
r
a
m
g
r
ad
i
en
t
i
n
f
o
r
m
atio
n
.
T
h
e
HOG
f
ea
t
u
r
e
d
escr
ip
to
r
,
d
ef
in
ed
i
n
[
2
]
,
ex
tr
ac
ts
u
s
e
f
u
l
i
n
f
o
r
m
atio
n
f
r
o
m
a
g
i
v
e
n
i
m
a
g
e
an
d
d
is
ca
r
d
s
e
x
tr
a
n
e
o
u
s
i
n
f
o
r
m
atio
n
.
I
n
g
en
er
al,
t
h
e
H
OG
p
r
o
ce
s
s
co
n
tai
n
s
t
h
r
ee
p
h
a
s
es
f
o
r
th
e
d
iv
id
ed
b
lo
ck
s
,
as
d
e
s
cr
ib
ed
in
[
1
]
:
i
)
co
n
d
u
ctin
g
im
a
g
e
n
o
r
m
al
izatio
n
,
ii
)
co
m
p
u
tin
g
t
h
e
i
m
a
g
e
g
r
ad
ien
ts
f
o
r
x
an
d
y
d
ir
ec
tio
n
s
,
an
d
iii
)
co
llectin
g
HO
G
d
escr
ip
to
r
s
f
o
r
all
b
lo
ck
s
.
I
n
th
e
o
r
ig
in
al
alg
o
r
it
h
m
p
r
o
p
o
s
ed
in
[
2
]
,
th
e
E
u
clid
ea
n
n
o
r
m
w
as
u
til
ized
to
ca
lcu
late
t
h
e
g
r
ad
ien
t
m
a
g
n
it
u
d
e.
Sin
ce
p
-
n
o
r
m
s
a
r
e
cr
u
ci
al
in
b
o
th
p
u
r
e
an
d
ap
p
lied
m
at
h
e
m
a
tics
,
o
th
er
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.
11
,
No
.
5
,
Octo
b
e
r
2
0
2
1
:
4
4
2
3
-
4430
4424
n
o
r
m
s
ca
n
b
e
u
s
ed
to
ca
lcu
late
len
g
th
o
r
m
a
g
n
it
u
d
e.
I
n
f
ac
t,
L
p
s
p
ac
es (
o
r
L
eb
es
g
u
e
s
p
ac
es
)
h
av
e
a
k
e
y
r
o
le
in
m
at
h
e
m
a
tical
a
n
al
y
s
is
a
n
d
ar
e
u
s
ed
in
m
a
n
y
d
is
cip
li
n
es,
s
u
c
h
as
co
m
p
u
ter
s
cie
n
ce
,
en
g
in
ee
r
i
n
g
,
p
h
y
s
ic
s
,
s
tatis
t
ics,
an
d
f
in
a
n
ce
[
1
0
]
.
T
h
e
p
-
n
o
r
m
o
r
L
p
-
n
o
r
m
o
f
a
v
ec
t
o
r
=
(
1
,
2
)
in
R
2
is
d
e
f
in
ed
as
f
o
llo
w
s
:
‖
‖
=
(
|
1
|
+
|
2
|
)
1
/
(
1
)
w
h
er
e
p
is
a
r
ea
l
n
u
m
b
er
≥
1
.
I
f
p
=
1
(
ca
lled
th
e
L
1
n
o
r
m
)
,
th
e
d
i
s
tan
ce
b
et
w
ee
n
t
w
o
p
o
in
ts
is
th
e
to
tal
d
is
tan
ce
tr
a
v
eled
:
‖
‖
1
=
|
1
|
+
|
2
|
.
T
h
u
s
,
p
=
1
co
u
n
ts
th
e
to
tal
c
h
an
g
es
in
b
o
th
x
an
d
y
d
ir
ec
tio
n
s
.
I
f
p
=
2
,
w
e
h
a
v
e
t
h
e
E
u
c
lid
ea
n
n
o
r
m
(
L
2
n
o
r
m
)
,
i.e
.
,
th
e
s
h
o
r
test
d
is
tan
ce
b
et
w
ee
n
t
w
o
p
o
in
ts
:
‖
‖
2
=
1
2
+
2
2
.
I
t
is
w
id
el
y
u
s
ed
an
d
is
t
h
e
o
n
l
y
n
o
r
m
i
n
v
ar
ia
n
t
u
n
d
er
a
n
y
u
n
it
ar
y
tr
an
s
f
o
r
m
,
s
u
c
h
as r
o
tatio
n
.
=
∞
co
n
s
id
er
s
th
e
h
i
g
h
est
g
r
ad
ien
t c
h
an
g
e
a
n
d
ig
n
o
r
es t
h
e
s
m
allest o
n
e,
w
h
er
e
∞
n
o
r
m
o
f
=
(
1
,
2
)
is
d
ef
i
n
ed
as:
‖
‖
∞
=
ma
x
{
|
1
|
,
|
2
|
}
(
2
)
Giv
e
n
t
h
e
p
o
w
er
o
f
p
-
n
o
r
m
s
i
n
m
ax
i
m
izin
g
p
er
f
o
r
m
an
ce
o
r
m
i
n
i
m
izi
n
g
er
r
o
r
,
it c
o
u
ld
b
e
ar
g
u
ed
t
h
a
t
th
e
E
u
clid
ea
n
n
o
r
m
(
o
r
2
-
n
o
r
m
)
in
HOG
d
escr
ip
to
r
s
is
n
o
t
n
ec
es
s
ar
il
y
t
h
e
o
n
l
y
c
h
o
ice
f
o
r
d
etec
tin
g
th
e
ac
t
u
al
h
is
to
g
r
a
m
g
r
ad
ien
t
in
a
n
i
m
a
g
e.
I
n
r
ea
lit
y
,
i
m
a
g
e
s
ca
l
in
g
a
n
d
r
eso
lu
tio
n
ar
e
k
n
o
w
n
to
a
f
f
e
ct
th
e
p
er
f
o
r
m
an
ce
o
f
f
ea
t
u
r
e
d
etec
tio
n
.
D
if
f
er
e
n
t
p
-
n
o
r
m
s
m
a
y
en
h
a
n
ce
th
e
ca
p
tu
r
in
g
o
f
ac
tu
al
d
i
s
tan
ce
s
f
o
r
a
s
u
i
tab
le
v
al
u
e
o
f
p.
Dif
f
er
en
t
p
-
n
o
r
m
v
al
u
es
ar
e
ex
p
ec
ted
to
af
f
ec
t
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
HOG
f
ea
t
u
r
e
d
etec
to
r
in
d
if
f
er
en
t
w
a
y
s
.
I
n
,
all
n
o
r
m
s
ar
e
eq
u
iv
ale
n
t
[
1
0
]
—
i.e
.
,
f
o
r
an
y
n
o
r
m
|
|
|
|
an
d
|
|
|
|
,
th
er
e
ar
e
p
o
s
itiv
e
co
n
s
ta
n
ts
c
an
d
k
s
u
c
h
th
at
|
|
|
|
≤
|
|
|
|
≤
|
|
|
|
f
o
r
all
∈
.
I
n
o
th
er
w
o
r
d
s
,
t
h
er
e
is
o
n
l
y
o
n
e
n
o
r
m
to
p
o
lo
g
y
in
.
C
o
n
s
eq
u
e
n
tl
y
,
t
h
e
co
n
v
er
g
en
ce
o
f
a
s
e
q
u
en
ce
o
f
v
ec
to
r
s
i
n
is
in
d
ep
en
d
en
t
o
f
th
e
ch
o
ice
o
f
p
-
n
o
r
m
.
Nev
er
t
h
eles
s
,
d
if
f
er
e
n
t
n
o
r
m
s
o
f
f
er
f
lex
ib
i
lit
y
to
p
r
o
v
e
co
n
v
er
g
en
ce
.
I
n
n
u
m
er
ical
a
n
al
y
s
is
,
ch
o
o
s
in
g
a
s
u
i
tab
le
n
o
r
m
p
la
y
s
a
r
o
le
i
n
e
f
f
icien
tl
y
d
eter
m
in
in
g
co
n
v
er
g
e
n
ce
.
C
o
n
v
er
g
en
ce
i
n
i
n
f
i
n
ite
-
d
i
m
en
s
io
n
al
v
ec
to
r
s
p
ac
es
d
ep
en
d
s
o
n
th
e
c
h
o
ice
o
f
p
-
n
o
r
m
,
as
p
-
n
o
r
m
s
in
i
n
f
i
n
ite
-
d
i
m
e
n
s
i
o
n
al
v
ec
to
r
s
p
ac
es
ar
e
n
o
t
eq
u
iv
alen
t
[
1
0
]
.
Usu
a
ll
y
,
o
n
e
n
o
r
m
is
m
o
r
e
s
u
itab
l
e
th
an
o
th
er
s
f
o
r
s
o
lv
in
g
ce
r
t
ain
p
r
o
b
lem
s
.
Fo
r
in
s
ta
n
ce
,
t
h
e
1
-
n
o
r
m
(
r
ath
er
t
h
an
t
h
e
2
-
n
o
r
m
)
ca
n
b
e
u
s
ed
t
o
f
in
d
t
h
e
to
tal
d
is
ta
n
ce
tr
av
el
ed
in
a
r
ec
tan
g
u
lar
s
tr
ee
t
g
r
id
f
r
o
m
a
lo
ca
tio
n
m
ar
k
ed
as
th
e
o
r
ig
i
n
an
d
th
e
d
esti
n
a
tio
n
p
o
in
t(
x,
y)
.
I
n
ap
p
r
o
x
i
m
at
io
n
t
h
eo
r
y
,
o
p
tim
izatio
n
p
r
o
b
le
m
s
d
ep
en
d
o
n
th
e
ch
o
ice
o
f
p
-
n
o
r
m
a
l
g
o
r
ith
m
to
o
b
tain
o
p
ti
m
al
s
o
l
u
tio
n
s
[
11
]
,
[
1
2
]
.
I
n
s
h
o
r
t,
s
o
lu
t
io
n
s
to
a
p
r
o
b
le
m
c
an
v
ar
y
w
it
h
d
if
f
er
en
t
n
o
r
m
s
.
p
-
n
o
r
m
s
ar
e
w
id
el
y
u
s
ed
i
n
m
ac
h
in
e
lear
n
i
n
g
an
d
ar
ti
f
icia
l
in
te
llig
e
n
ce
a
n
d
ar
e
p
o
w
er
f
u
l
to
o
ls
f
o
r
ev
alu
a
tin
g
an
d
i
m
p
r
o
v
i
n
g
m
ac
h
in
e
lear
n
in
g
m
o
d
els.
P
r
ed
ictio
n
in
m
ac
h
i
n
e
lear
n
i
n
g
r
elies
o
n
d
etec
tin
g
p
atter
n
s
a
n
d
i
n
f
er
e
n
ce
s
,
r
at
h
er
th
a
n
e
x
p
licit
in
s
tr
u
ct
io
n
s
.
Usi
n
g
s
a
m
p
le
d
ata
w
h
e
n
b
u
ild
in
g
m
ac
h
i
n
e
lear
n
in
g
alg
o
r
ith
m
s
r
eq
u
ir
e
s
test
i
n
g
th
e
p
r
ed
ictiv
e
m
o
d
els
to
ac
h
iev
e
t
h
e
b
est
p
er
f
o
r
m
a
n
c
e.
Ma
x
i
m
iz
in
g
th
e
p
er
f
o
r
m
a
n
ce
o
r
m
i
n
i
m
izi
n
g
th
e
er
r
o
r
o
f
a
m
o
d
el,
in
o
th
er
wo
r
d
s
,
ai
m
s
to
m
i
n
i
m
ize
t
h
e
co
s
t
f
u
n
ctio
n
.
No
r
m
s
ar
e
u
s
e
f
u
l
in
m
ea
s
u
r
i
n
g
s
u
ch
er
r
o
r
s
[
1
3
]
.
I
n
ad
d
itio
n
,
s
o
lv
in
g
an
o
p
ti
m
izatio
n
p
r
o
b
lem
m
ea
n
s
f
i
n
d
in
g
t
h
e
in
p
u
t
t
h
at
b
est
m
in
i
m
ize
s
s
o
m
e
o
u
tp
u
t
p
en
alt
y
[
1
4
]
.
No
r
m
s
as
s
ig
n
a
m
ag
n
it
u
d
e
to
t
h
e
s
e
o
u
tp
u
ts
an
d
h
e
n
ce
en
ab
le
p
en
alt
ie
s
to
b
e
m
i
n
i
m
iz
ed
.
I
n
m
ac
h
i
n
e
lear
n
i
n
g
,
d
i
f
f
er
en
t
n
o
r
m
s
ca
n
b
e
u
s
ed
f
o
r
r
eg
u
l
ar
izatio
n
an
d
f
ea
t
u
r
e
s
e
lectio
n
,
as
a
lo
s
s
f
u
n
ctio
n
,
an
d
s
o
o
n
.
C
h
o
o
s
i
n
g
w
h
ic
h
n
o
r
m
to
u
s
e
d
ep
en
d
s
o
n
t
h
e
p
r
o
b
le
m
to
b
e
s
o
l
v
ed
,
as
ea
ch
n
o
r
m
h
as
its
o
w
n
p
r
o
s
an
d
co
n
s
[
1
5
]
.
T
h
e
p
r
in
cip
le
o
f
p
ar
s
i
m
o
n
y
i
n
m
ac
h
in
e
lear
n
i
n
g
i
s
co
m
m
o
n
l
y
u
s
ed
to
cr
ea
te
a
p
r
ed
ictio
n
m
o
d
el
w
it
h
g
o
o
d
s
p
ar
s
e
ap
p
r
o
x
im
a
tio
n
.
R
e
g
u
lar
izatio
n
tec
h
n
iq
u
e
s
w
it
h
d
if
f
er
e
n
t
n
o
r
m
s
ar
e
ap
p
lied
to
a
d
d
r
ess
o
v
er
f
itti
n
g
,
o
u
tlier
s
,
an
d
f
ea
tu
r
e
s
e
lectio
n
in
a
m
o
d
el
[
1
3
]
.
T
h
e
L
1
n
o
r
m
i
s
o
f
te
n
u
s
ed
t
o
ca
lcu
late
th
e
Ma
n
h
atta
n
o
r
tax
icab
d
i
s
tan
ce
,
m
ea
n
ab
s
o
lu
te
er
r
o
r
(
MA
E
)
,
an
d
t
h
e
lea
s
t
ab
s
o
l
u
te
s
h
r
i
n
k
ag
e
a
n
d
s
e
lectio
n
o
p
er
ato
r
(
L
A
SS
O)
.
L
ASSO
u
s
es
L
1
r
eg
u
lar
izatio
n
to
r
ed
u
ce
th
e
h
u
g
e
n
u
m
b
er
o
f
f
ea
t
u
r
es
i
n
a
m
o
d
el
b
y
r
e
m
o
v
in
g
les
s
i
m
p
o
r
ta
n
t
f
ea
t
u
r
es,
s
i
n
ce
L
1
is
r
o
b
u
s
t
to
w
ar
d
s
o
u
t
lier
s
a
n
d
m
i
s
s
i
n
g
d
ata
[
1
3
]
.
On
th
e
o
th
er
h
a
n
d
,
th
e
L
2
n
o
r
m
is
o
f
ten
u
s
ed
to
c
alcu
late
E
u
clid
ea
n
d
is
tan
ce
,
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
an
d
least
s
q
u
ar
es
er
r
o
r
,
an
d
th
e
r
id
g
e
o
p
er
ato
r
,
w
h
ich
u
s
es
L
2
r
eg
u
lar
iz
atio
n
to
h
a
n
d
le
o
v
er
f
itti
n
g
[
1
3
]
.
T
h
er
e
ar
e
m
a
n
y
e
f
f
icie
n
t
m
et
h
o
d
s
av
ailab
le
f
o
r
th
e
w
id
el
y
u
s
ed
L
2
n
o
r
m
;
h
o
w
e
v
er
,
th
e
L
2
n
o
r
m
i
s
s
en
s
itiv
e
to
o
u
t
lier
s
d
u
e
to
en
o
r
m
o
u
s
s
q
u
ar
ed
er
r
o
r
v
alu
es.
Mo
r
eo
v
er
,
ex
tr
ac
tin
g
m
ea
n
i
n
g
f
u
l
f
ea
tu
r
e
s
r
eq
u
ir
es
r
o
b
u
s
t
f
ea
t
u
r
e
s
elec
tio
n
m
eth
o
d
s
th
at
ca
n
eli
m
i
n
ate
n
o
is
y
p
o
in
t
s
.
I
n
[
1
6
]
p
r
o
p
o
s
ed
jo
in
t
L
1,
2
n
o
r
m
m
i
n
i
m
izatio
n
o
n
b
o
th
lo
s
s
f
u
n
ctio
n
an
d
r
eg
u
lar
izatio
n
to
m
a
k
e
f
ea
tu
r
e
s
elec
tio
n
m
o
r
e
ef
f
icie
n
t.
T
h
is
id
ea
r
ef
lect
s
t
h
e
e
f
f
ec
t
o
f
u
s
i
n
g
m
o
r
e
t
h
a
n
o
n
e
n
o
r
m
w
i
th
in
o
n
e
tech
n
iq
u
e.
R
esear
c
h
er
s
h
a
v
e
w
o
r
k
ed
o
n
i
m
p
r
o
v
i
n
g
th
e
f
r
a
m
e
w
o
r
k
o
f
SVM
s
(
alo
n
g
w
it
h
o
th
er
alg
o
r
ith
m
s
)
u
s
i
n
g
p
-
n
o
r
m
s
.
So
m
e
h
a
v
e
p
r
o
p
o
s
ed
a
1
-
n
o
r
m
SVM
to
a
ch
iev
e
m
o
r
e
s
p
ar
s
e
cla
s
s
i
f
ier
s
[
1
7
]
.
Oth
er
s
h
a
v
e
in
tr
o
d
u
ce
d
a
n
e
w
ap
p
r
o
ac
h
u
s
in
g
a
0
<
p
<
1
n
o
r
m
[
1
8
]
,
wh
ich
w
as
s
h
o
w
n
to
b
e
m
o
r
e
e
f
f
ec
tiv
e
t
h
an
t
h
e
1
-
n
o
r
m
SVM.
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
p
-
n
o
r
ms o
f h
is
to
g
r
a
m
o
f o
r
ien
ted
g
r
a
d
ien
ts
fo
r
X
-
r
a
y
ima
g
es
(
N
u
h
a
H.
Ha
ma
d
a
)
4425
I
n
th
i
s
p
ap
er
,
w
e
in
v
e
s
ti
g
ate
d
if
f
er
e
n
t
p
-
n
o
r
m
v
al
u
es
i
n
th
e
HOG
alg
o
r
ith
m
(
p
-
HOG)
t
o
ac
h
iev
e
b
etter
p
er
f
o
r
m
an
ce
i
n
class
if
y
in
g
m
ed
ical
X
-
r
a
y
i
m
ag
es.
T
o
test
d
if
f
er
en
t
n
o
r
m
s
in
th
e
p
r
o
p
o
s
ed
m
o
d
i
f
icat
io
n
,
w
e
u
s
ed
a
d
at
aset
o
f
X
-
r
a
y
i
m
a
g
es
f
r
o
m
C
OVI
D
-
1
9
p
atien
ts
a
n
d
r
ec
o
r
d
ed
th
e
r
esu
lt
s
o
f
co
m
p
ar
i
n
g
t
h
e
o
r
ig
in
al
HOG
an
d
p
-
HOG
alg
o
r
it
h
m
s
u
s
in
g
d
if
f
er
e
n
t
p
-
n
o
r
m
v
al
u
es.
B
o
th
w
er
e
i
m
p
le
m
en
ted
in
P
y
t
h
o
n
.
T
h
e
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
.
I
n
s
ec
tio
n
2
,
w
e
d
escr
ib
e
t
h
e
s
tep
s
o
f
in
c
lu
d
in
g
t
h
e
p
-
n
o
r
m
i
n
th
e
HOG
al
g
o
r
ith
m
an
d
p
r
ese
n
t
th
e
e
x
p
er
i
m
e
n
ts
p
er
f
o
r
m
ed
o
n
th
e
d
ataset.
W
e
d
is
p
la
y
an
d
d
is
cu
s
s
t
h
e
r
esu
lts
in
s
ec
tio
n
3
,
th
en
co
n
clu
d
e
t
h
e
p
ap
er
in
s
ec
tio
n
4
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
2
.
1
.
p
-
H
O
G
a
lg
o
ri
t
h
m
As
m
en
t
io
n
ed
i
n
t
h
e
i
n
tr
o
d
u
ctio
n
,
p
-
n
o
r
m
s
ar
e
w
id
el
y
u
s
ed
i
n
m
ac
h
in
e
lear
n
i
n
g
to
i
m
p
r
o
v
e
p
r
ed
ictiv
e
m
o
d
els.
Us
in
g
t
h
e
p
-
n
o
r
m
s
w
it
h
t
h
e
b
est
p
er
f
o
r
m
an
ce
an
d
ac
c
u
r
ac
y
w
i
ll
a
f
f
ec
t
o
u
r
f
i
n
d
in
g
s
.
I
n
th
is
s
ec
tio
n
,
w
e
p
r
o
p
o
s
e
th
e
p
-
HOG
alg
o
r
ith
m
b
y
ch
a
n
g
in
g
h
o
w
w
e
m
ea
s
u
r
e
d
is
ta
n
ce
u
s
i
n
g
d
if
f
er
en
t
p
-
n
o
r
m
s
in
s
tead
o
f
t
h
e
E
u
c
lid
ea
n
n
o
r
m
.
T
h
e
g
o
al
is
to
im
p
r
o
v
e
t
h
e
H
OG
d
escr
ip
to
r
’
s
d
etec
tio
n
p
r
o
ce
s
s
.
I
n
t
h
e
o
r
ig
in
a
l
HOG
alg
o
r
it
h
m
,
it
is
n
ec
es
s
ar
y
to
ex
tr
ac
t
t
h
e
m
a
in
f
ea
t
u
r
e
d
escr
ip
to
r
to
id
en
tify
i
m
ag
e
f
ea
t
u
r
es.
T
h
e
in
f
o
r
m
atio
n
i
n
ea
c
h
8
-
p
i
x
el
×
8
-
p
ix
el
ce
ll
is
co
m
p
ac
ted
to
a
n
i
n
e
-
d
i
m
e
n
s
io
n
al
s
p
ac
e
co
n
s
i
s
ti
n
g
o
f
n
i
n
e
an
g
u
lar
b
i
n
s
w
h
ich
ar
e
eq
u
all
y
d
i
v
id
ed
o
v
er
0
0
–
180
0
ac
co
r
d
in
g
to
t
h
eir
g
r
ad
ien
t
d
ir
ec
tio
n
s
.
T
h
e
f
o
llo
w
i
n
g
s
tep
s
ex
p
lai
n
t
h
e
u
s
e
o
f
th
e
p
-
n
o
r
m
in
t
h
e
al
g
o
r
ith
m
.
A
ll
s
tep
s
ex
ce
p
t
s
tep
3
ar
e
d
er
i
v
ed
f
r
o
m
th
e
H
O
G
alg
o
r
ith
m
.
Select
th
e
m
ai
n
b
lo
ck
w
it
h
a
s
i
ze
r
atio
o
f
1
:2
.
Div
id
e
t
h
e
m
ai
n
b
lo
ck
i
n
to
8
-
p
ix
el
×
8
-
p
i
x
el
ce
l
ls
to
co
m
p
u
te
t
h
e
h
is
to
g
r
a
m
o
f
g
r
ad
ien
ts
i
n
x
an
d
y
d
ir
ec
tio
n
s
(
d
en
o
ted
as
g
x
an
d
g
y
,
r
esp
ec
tiv
el
y
)
,
as
s
h
o
w
n
i
n
Fi
g
u
r
e
1
.
Fig
u
r
e
1
illu
s
tr
ate
s
th
e
h
is
to
g
r
a
m
g
en
er
ated
f
o
r
a
s
i
n
g
le
ce
ll.
Fig
u
r
e
1
.
His
to
g
r
a
m
g
e
n
er
ated
f
o
r
a
s
in
g
le
ce
ll
C
o
m
p
u
te
th
e
L
p
g
r
ad
ien
t
m
a
g
n
itu
d
e,
g
i
v
en
a
s
|
|
|
|
=
(
|
|
+
|
|
)
1
/
(
3
)
w
h
er
e
p
is
a
r
ea
l n
u
m
b
er
≥
1
,
an
d
th
e
L
p
g
r
ad
ien
t d
ir
ec
tio
n
a
n
g
le
b
y
:
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.
11
,
No
.
5
,
Octo
b
e
r
2
0
2
1
:
4
4
2
3
-
4430
4426
tan
−
1
(
4)
No
r
m
a
lize
t
h
e
b
lo
ck
o
f
1
6
×
1
6
p
ix
els
(
o
r
4
=
2
×
2
ce
lls
)
to
r
ed
u
ce
th
e
v
ar
iatio
n
o
f
g
r
ad
ien
t
m
ag
n
it
u
d
e
th
at
o
cc
u
r
r
ed
b
y
th
e
s
h
ad
o
ws
co
m
es
o
v
er
a
b
r
o
ad
r
an
g
e
.
T
h
e
r
esu
ltin
g
4
×
9
=
36
-
d
im
en
s
io
n
a
l
s
p
ac
e
h
is
to
g
r
a
m
i
s
n
o
r
m
a
lized
to
u
n
i
t le
n
g
th
,
as i
llu
s
tr
ate
d
in
F
ig
u
r
e
2
.
Fig
u
r
e
2
.
His
to
g
r
a
m
g
e
n
er
ated
f
o
r
a
b
lo
ck
o
f
f
o
u
r
ce
lls
2
.
2
.
E
x
peri
m
ent
s
C
o
r
o
n
av
ir
u
s
d
i
s
ea
s
e
2019
(
C
OVI
D
-
1
9
)
is
a
w
id
esp
r
ea
d
d
is
ea
s
e
ca
u
s
ed
by
S
AR
S
-
C
o
V
-
2
[
1
9
]
.
T
h
e
d
is
ea
s
e
f
ir
s
t
h
it
W
u
h
an
,
C
h
i
n
a
,
in
late
Dec
e
m
b
er
2
0
1
9
.
As
th
e
n
u
m
b
er
of
co
n
f
ir
m
ed
ca
s
es
in
cr
ea
s
ed
r
ap
id
ly
,
C
OVI
D
-
1
9
w
a
s
d
ec
lar
ed
a
p
an
d
e
m
ic
o
n
Ma
r
c
h
11,
2020
[
2
0
]
.
C
OVI
D
-
19
can
be
d
ia
g
n
o
s
ed
b
ased
o
n
a
co
m
b
i
n
atio
n
of
s
y
m
p
to
m
s
,
i
n
clu
d
i
n
g
f
e
v
er
(
8
7
.
9
%),
d
r
y
co
u
g
h
(
6
7
.
7
%),
f
atig
u
e
(
3
8
.0
%),
an
d
s
p
u
tu
m
p
r
o
d
u
ctio
n
(
3
3
.
4
%),
am
o
n
g
o
th
er
s
[
2
1
]
.
On
Ma
r
ch
27,
2020,
th
e
W
o
r
ld
Hea
lth
Or
g
an
izatio
n
(
W
HO)
an
n
o
u
n
ce
d
th
at
t
h
e
o
u
tb
r
ea
k
in
clu
d
ed
5
0
9
,
1
6
4
c
o
n
f
ir
m
ed
ca
s
es,
w
h
ic
h
r
esu
l
ted
in
2
3
,
3
3
5
d
ea
th
s
ac
r
o
s
s
201
co
u
n
tr
ies
[
2
2
]
,
[
2
3
]
a
d
ea
th
r
ate
o
f
ap
p
r
o
x
i
m
atel
y
4
.
6
%
.
T
h
e
g
r
o
w
t
h
in
th
e
n
u
m
b
er
of
d
ia
g
n
o
s
ed
ca
s
es
is
d
u
e
to
clo
s
e
co
n
tact
an
d
h
u
m
a
n
-
to
-
h
u
m
a
n
tr
an
s
m
i
s
s
io
n
[
2
0
]
,
[
2
4
]
.
Scien
tis
t
s
all
o
v
er
th
e
w
o
r
ld
ar
e
w
o
r
k
in
g
h
ar
d
to
o
v
er
co
m
e
th
i
s
h
ea
lt
h
cr
is
is
,
w
h
ich
p
o
s
es
a
s
ev
er
e
th
r
ea
t
to
p
u
b
lic
h
ea
lth
esp
ec
iall
y
to
o
ld
er
p
atien
ts
w
it
h
ch
r
o
n
ic
d
is
ea
s
e
s
d
u
e
to
t
h
e
u
n
p
r
ed
ictab
le
j
u
m
p
in
C
OVI
D
-
19
p
atien
ts
.
C
h
est
X
-
r
a
y
s
can
be
u
s
ed
to
d
etec
t
th
e
f
ea
t
u
r
es
of
p
n
e
u
m
o
n
ia
[
2
1
]
,
[
2
4
]
;
th
er
ef
o
r
e,
th
is
r
esear
ch
w
ill
co
n
d
u
ct
co
m
p
ar
i
s
o
n
ex
p
er
im
e
n
t
s
u
s
in
g
a
s
et
of
X
-
r
a
y
i
m
ag
e
s
.
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
p
-
n
o
r
ms o
f h
is
to
g
r
a
m
o
f o
r
ien
ted
g
r
a
d
ien
ts
fo
r
X
-
r
a
y
ima
g
es
(
N
u
h
a
H.
Ha
ma
d
a
)
4427
2
.
2
.
1
.
Da
t
a
s
et
s
elec
t
io
n
I
n
g
e
n
e
r
al,
d
atasets
o
f
C
OVI
D
-
1
9
X
-
r
a
y
i
m
a
g
es
ar
e
s
til
l
ev
o
lv
i
n
g
.
T
h
e
d
ataset
u
s
ed
in
t
h
is
p
ap
er
in
cl
u
d
ed
t
w
o
ca
te
g
o
r
ies,
+C
O
VI
D
-
1
9
an
d
-
C
OVI
D
-
1
9
,
w
h
i
ch
i
n
d
icate
s
ca
n
s
o
f
p
atien
t
s
w
it
h
a
n
d
w
it
h
o
u
t
C
OVI
D
-
1
9
,
r
esp
ec
tiv
el
y
.
T
h
e
+COVI
D
-
1
9
X
-
r
a
y
i
m
a
g
es
u
s
ed
in
t
h
i
s
r
esear
c
h
w
er
e
co
llec
ted
b
y
[
2
5
]
.
I
n
t
h
e
o
r
ig
in
al
d
ata
h
e
p
r
o
v
id
ed
,
o
n
ly
p
o
s
i
tiv
e
C
OVI
D
-
1
9
ca
s
es
wer
e
in
clu
d
ed
.
Oth
er
i
m
ag
e
s
r
elate
d
to
SA
R
S
an
d
ME
R
S
w
er
e
ig
n
o
r
ed
.
T
h
e
to
t
al
i
m
ag
e
s
in
cl
u
d
ed
2
5
+
C
OV
I
D
-
19
i
m
ag
e
s
.
T
h
e
d
ataset
u
s
ed
w
a
s
r
elativ
e
l
y
s
m
al
l;
h
o
w
e
v
er
,
as
th
ese
e
x
p
e
r
i
m
en
ts
w
er
e
p
er
f
o
r
m
ed
as
a
p
r
o
o
f
o
f
co
n
ce
p
t
an
d
s
in
ce
th
i
s
t
y
p
e
o
f
i
m
ag
e
i
s
s
till
n
o
t
attai
n
ab
le
at
a
lar
g
e
s
ca
le,
th
i
s
d
ataset
i
s
co
n
s
id
er
ed
ac
ce
p
tab
le
[
2
6
]
.
T
h
e
-
C
O
VI
D
-
1
9
d
ata
w
a
s
d
o
w
n
lo
ad
ed
f
r
o
m
[
2
7
]
,
w
h
er
e
i
m
a
g
es
o
f
p
n
e
u
m
o
n
ia
w
er
e
co
lle
cted
an
d
s
to
r
ed
in
t
h
e
Kag
g
le
r
ep
o
s
ito
r
y
.
Ho
w
e
v
er
,
[
2
8
]
co
llected
2
5
i
m
ag
e
s
f
r
o
m
t
h
e
r
ep
o
s
ito
r
y
to
av
o
id
n
o
is
y
,
m
is
lab
eled
,
an
d
b
lu
r
r
y
i
m
ag
e
s
.
I
n
th
is
r
esear
c
h
,
w
e
u
s
e
th
e
f
i
n
al
d
ata
f
r
o
m
[
2
8
]
.
2
.
2
.
3
.
E
x
peri
m
ent
a
l
s
et
up
P
y
t
h
o
n
3
.
7
.
3
w
a
s
s
et
u
p
w
it
h
th
e
p
a
ck
ag
e
s
n
ec
es
s
ar
y
s
u
c
h
as
s
k
i
m
a
g
e
,
n
u
m
p
y
,
a
n
d
o
p
en
C
V
to
p
er
f
o
r
m
th
e
e
x
p
er
i
m
en
t
s
p
r
es
en
ted
in
th
is
p
ap
er
.
T
h
e
s
p
ec
if
icatio
n
s
o
f
th
e
co
m
p
u
ter
s
y
s
t
e
m
u
s
ed
w
er
e
as
f
o
llo
w
s
:
I
n
tel
®
C
o
r
e
™
i7
-
8
7
5
0
H
C
P
U
(
3
.
7
0
GHz
,
9
M
C
ac
h
e
)
a
n
d
1
6
.
0
0
GB
R
A
M.
W
e
u
s
ed
1
0
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
to
en
s
u
r
e
m
o
r
e
r
eliab
le
r
esu
lts
f
r
o
m
t
h
e
g
e
n
er
ated
m
o
d
els.
Fo
r
ea
ch
i
m
a
g
e,
b
o
th
t
h
e
o
r
i
g
in
a
l
HO
G
a
n
d
t
h
e
p
-
HOG
f
ea
tu
r
e
d
etec
to
r
d
escr
ip
to
r
s
ar
e
ap
p
lied
.
T
h
e
o
r
ig
in
al
HOG
d
escr
ip
to
r
w
a
s
ex
tr
ac
ted
f
r
o
m
t
h
e
o
r
ig
i
n
a
l
P
y
t
h
o
n
i
m
p
le
m
e
n
tatio
n
in
t
h
e
s
k
lear
n
p
ac
k
a
g
e
,
w
h
ic
h
d
ep
en
d
s
o
n
th
e
s
ca
le
-
i
n
v
ar
ian
t
f
ea
tu
r
e
tr
an
s
f
o
r
m
(
SIF
T
)
alg
o
r
ith
m
[
2
9
]
.
T
h
e
p
-
HO
G
i
m
p
le
m
e
n
tatio
n
w
a
s
b
ased
o
n
th
e
i
m
p
le
m
en
t
atio
n
f
o
u
n
d
in
[
3
0
]
.
T
h
e
m
o
d
if
icatio
n
s
w
er
e
i
m
p
le
m
e
n
ted
in
th
e
m
e
th
o
d
s
,
ad
d
in
g
t
h
e
“
n
o
r
m
”
p
ar
a
m
eter
f
o
r
th
e
m
a
g
n
i
tu
d
e
m
et
h
o
d
an
d
u
s
in
g
t
h
e
n
e
w
v
al
u
e
to
f
i
n
d
th
e
o
r
ie
n
tatio
n
,
g
r
ad
ien
t,
an
d
HOG
ca
lc
u
lat
io
n
s
.
T
h
e
g
en
er
ated
H
OG
a
n
d
p
-
HOG
d
escr
ip
to
r
s
f
o
r
all
i
m
a
g
es
w
er
e
f
ed
s
ep
ar
atel
y
i
n
to
th
e
SVM
alg
o
r
ith
m
to
g
en
er
ate
a
d
i
f
f
er
en
t
m
o
d
el
f
o
r
ea
ch
,
w
h
ic
h
was
later
u
s
ed
i
n
cla
s
s
if
ica
tio
n
.
T
h
e
m
o
d
el
w
a
s
ev
alu
a
ted
u
s
i
n
g
th
e
u
n
s
ee
n
t
esti
n
g
s
et
,
a
n
d
th
e
r
esu
l
ts
we
r
e
r
ec
o
r
d
ed
an
d
c
o
m
p
ar
ed
.
T
o
g
en
er
ate
a
f
u
l
l
p
ictu
r
e
o
f
th
e
p
-
n
o
r
m
’
s
e
f
f
ec
t
o
n
th
e
r
es
u
lt
s
,
d
if
f
er
en
t
p
-
n
o
r
m
v
a
lu
e
s
we
r
e
test
ed
:
p
-
n
o
r
m
=
1
,
2
,
10
,
20
,
a
n
d
∞
.
Mo
r
eo
v
er
,
1
0
-
f
o
ld
cr
o
s
s
-
v
a
lid
atio
n
w
a
s
u
s
ed
to
e
n
s
u
r
e
t
h
e
r
eliab
ilit
y
o
f
th
e
p
r
o
d
u
ce
d
r
esu
lt
s
,
a
n
d
a
t
-
test
w
a
s
u
s
ed
to
r
ec
o
r
d
w
h
et
h
er
th
e
d
if
f
er
e
n
ce
s
b
et
w
ee
n
r
es
u
lt
s
we
r
e
s
tat
is
tica
ll
y
s
ig
n
i
f
ica
n
t.
2
.
2
.
3
.
P
er
f
o
r
m
a
nce
m
ea
s
ure
s
Dif
f
er
en
t
to
o
ls
ar
e
u
s
ed
to
co
m
p
ar
e
r
es
u
lts
o
f
d
i
f
f
er
e
n
t
d
ata
m
in
i
n
g
alg
o
r
it
h
m
s
[
3
1
]
.
O
n
e
s
u
c
h
to
o
l
is
t
h
e
co
n
f
u
s
io
n
m
atr
ix
,
w
h
ic
h
i
s
u
s
ed
as
“
an
in
d
icat
io
n
o
f
th
e
p
r
o
p
er
ties
o
f
a
cla
s
s
i
f
icat
io
n
(
d
is
cr
i
m
in
a
n
t)
r
u
le”
[
3
1
]
.
T
h
e
co
n
f
u
s
io
n
m
a
t
r
ix
h
a
s
f
o
u
r
v
al
u
es
t
h
at
in
d
ica
te
th
e
n
u
m
b
er
o
f
ca
s
es
co
r
r
ec
t
ly
an
d
i
n
co
r
r
ec
t
ly
class
i
f
ied
f
o
r
ea
ch
c
lass
.
I
n
t
h
is
r
esear
c
h
,
th
er
e
ar
e
t
w
o
cl
ass
es:
+
C
OVI
D
-
1
9
an
d
-
C
O
VI
D
-
1
9
.
T
h
e
t
r
u
e
p
o
s
itiv
e
(
T
P
)
r
ate
r
ef
er
s
to
th
e
co
r
r
ec
t
class
if
icatio
n
of
t
h
e
p
o
s
itiv
e
ca
s
es,
a
n
d
t
h
e
f
alse
p
o
s
itiv
e
(
FP
)
r
ate
in
d
icate
s
t
h
e
in
co
r
r
ec
t
class
i
f
i
ca
tio
n
o
f
p
o
s
itiv
e
ca
s
e
s
as
n
eg
ativ
e
.
T
h
e
t
r
u
e
n
eg
ati
v
e
(
T
N)
r
ate
d
escr
ib
es
th
e
co
r
r
ec
t
class
if
icatio
n
of
n
o
r
m
a
l
ca
s
es,
an
d
th
e
f
al
s
e
n
e
g
ati
v
e
(
FN)
r
ate
r
ep
r
esen
ts
th
e
i
n
co
r
r
ec
t
class
i
f
icatio
n
of
n
o
r
m
al
ca
s
es.
Alth
o
u
g
h
ac
cu
r
ac
y
is
n
o
t
t
h
e
o
n
l
y
i
n
d
icat
or
u
s
ed
,
it
ca
n
b
e
co
n
s
id
er
ed
o
n
e
o
f
t
h
e
m
o
s
t
i
m
p
o
r
tan
t.
A
cc
u
r
ac
y
is
co
m
p
u
ted
u
s
in
g
(
5
)
.
A
n
o
t
h
er
in
d
ic
atio
n
is
r
ec
all
o
r
s
en
s
iti
v
it
y
,
w
h
ic
h
s
h
o
w
s
h
o
w
w
ell
t
h
e
p
o
s
iti
v
e
ca
s
e
s
i
s
ca
lc
u
lated
u
s
i
n
g
(
6
)
[
3
1
]
.
=
+
+
+
+
(
5
)
/
=
+
(
6
)
P
r
ec
is
io
n
s
h
o
w
s
h
o
w
m
an
y
p
o
s
itiv
el
y
clas
s
i
f
ied
ca
s
es
w
er
e
r
elev
an
t
.
Hig
h
p
r
ec
is
io
n
i
n
d
i
ca
tes
th
a
t
ca
s
es
lab
eled
as
p
o
s
iti
v
e
we
r
e
in
d
ee
d
p
o
s
itiv
e
,
w
i
th
a
v
er
y
s
m
all
n
u
m
b
er
o
f
FP
s
.
Sp
e
cif
icit
y
i
s
a
n
o
t
h
er
u
s
e
f
u
l
in
d
icato
r
th
a
t
d
escr
ib
es
th
e
n
u
m
b
er
o
f
ca
s
es
w
it
h
o
u
t t
h
e
d
is
ea
s
e
w
h
o
test
n
eg
at
iv
e
[
3
1
]
.
=
+
(
7
)
=
+
(
8
)
Fin
all
y
,
t
h
e
F
-
m
ea
s
u
r
e
co
m
b
in
e
s
p
r
ec
is
io
n
an
d
r
ec
all
t
o
en
s
u
r
e
th
at
th
e
y
ar
e
b
ala
n
ce
d
.
T
h
e
ca
lcu
lati
n
g
f
o
r
th
e
F
-
m
ea
s
u
r
e
is
b
ein
g
as
[
3
1
]
:
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.
11
,
No
.
5
,
Octo
b
e
r
2
0
2
1
:
4
4
2
3
-
4430
4428
−
=
2
∗
∗
+
(
9
)
3.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
T
h
e
r
esu
lts
in
T
ab
le
1
r
ev
ea
l
th
at
u
s
in
g
th
e
li
n
ea
r
k
er
n
e
l
w
it
h
th
e
o
r
ig
i
n
al
HOG
alg
o
r
it
h
m
r
esu
lt
ed
in
9
4
.
8
%
ac
cu
r
ac
y
a
n
d
p
r
ec
is
io
n
an
d
r
ec
all
v
al
u
es
of
9
7
%
an
d
9
1
.
8
%,
r
esp
ec
tiv
el
y
.
Usi
n
g
th
e
p
-
HOG
al
g
o
r
ith
m
(
w
it
h
an
y
n
o
r
m
v
a
lu
e)
in
cr
ea
s
ed
a
cc
u
r
ac
y
to
9
5
%
(
L
2
,
L
∞
)
or
9
6
%
(
L
1
)
.
T
o
en
s
u
r
e
th
at
th
e
d
if
f
er
en
ce
s
we
r
e
s
tatis
t
icall
y
s
ig
n
i
f
ica
n
t,
a
t
-
te
s
t
w
a
s
co
n
d
u
cted
b
et
w
ee
n
t
h
e
o
r
ig
in
a
l
HO
G
a
n
d
p
-
H
OG
r
esu
l
ts
.
T
h
e
t
-
te
s
t
s
h
o
w
ed
s
ig
n
i
f
ica
n
t
d
i
f
f
er
e
n
ce
s
(
p
<
0
.
1
)
in
p
r
ec
is
io
n
,
r
ec
all,
an
d
s
p
ec
i
f
icit
y
(
p
=
0
.
0
3
,
p
=
0
.
0
8
,
an
d
p
=
0.
0
3
,
r
esp
ec
tiv
el
y
)
.
A
lt
h
o
u
g
h
ac
cu
r
ac
y
w
a
s
n
o
t
s
i
g
n
if
ican
t
l
y
d
if
f
er
en
t,
th
e
i
m
p
r
o
v
e
m
e
n
t
s
i
n
r
ec
all
an
d
s
p
ec
if
icit
y
s
ee
m
to
i
n
d
icate
a
g
o
o
d
in
f
l
u
e
n
ce
o
n
p
-
HO
G.
An
o
th
er
s
tep
w
a
s
ta
k
en
to
f
u
r
t
h
er
e
x
p
lo
r
e
th
e
r
ec
o
r
d
ed
r
esu
lts
-
th
at
is
,
to
co
m
p
ar
e
th
e
n
o
r
m
v
alu
e
s
in
t
h
e
p
-
HOG
i
m
p
le
m
en
tatio
n
.
I
n
ter
m
s
o
f
r
ec
all
an
d
ac
cu
r
ac
y
,
L
1
ha
d
s
ig
n
i
f
ica
n
tl
y
b
etter
r
es
u
lt
s
th
a
n
th
e
o
t
h
er
n
o
r
m
v
alu
e
s
(
n
o
r
m=2
an
d
n
o
r
m=1
0
,
r
esp
ec
tiv
el
y
)
.
T
ab
le
1
.
T
h
e
r
esu
lts
o
f
S
VM
w
it
h
li
n
ea
r
k
er
n
e
l u
s
i
n
g
HO
G
an
d
p
-
HOG
w
it
h
d
i
f
f
er
e
n
t n
o
r
m
s
L
i
n
e
a
r
K
e
r
n
e
l
N
o
r
m
P
r
e
c
i
si
o
n
R
e
c
a
l
l
F
-
m
e
a
su
r
e
S
p
e
c
i
f
i
c
i
t
y
A
c
c
u
r
a
c
y
p
-
HOG
1
9
9
.
0
%
9
3
.
8
%
9
6
.
3
%
9
9
.
0
%
9
6
.
0
%
2
9
9
.
0
%
8
9
.
6
%
9
4
.
1
%
9
9
.
0
%
9
5
.
0
%
10
1
0
0
.
0
%
9
1
.
7
%
9
5
.
6
%
1
0
0
.
0
%
9
5
.
5
%
20
1
0
0
.
0
%
9
1
.
7
%
9
5
.
7
%
1
0
0
.
0
%
9
5
.
0
%
∞
1
0
0
.
0
%
8
8
.
3
%
9
3
.
8
%
1
0
0
.
0
%
9
5
.
0
%
HOG
9
7
.
6
%
9
1
.
8
%
9
4
.
6
%
9
7
.
8
%
9
4
.
8
%
T
h
e
r
esu
lts
i
n
T
ab
le
2
r
ev
ea
l
t
h
at
u
s
i
n
g
t
h
e
R
B
F
k
er
n
e
l
w
it
h
th
e
o
r
ig
i
n
al
HOG
al
g
o
r
ith
m
r
esu
lt
ed
in
9
5
.
2
%
ac
cu
r
ac
y
.
T
h
ese
r
es
u
lts
w
er
e
v
er
y
s
i
m
ilar
to
th
e
p
-
HOG
r
es
u
lt
s
u
s
i
n
g
d
i
f
f
er
e
n
t
n
o
r
m
v
a
lu
e
s
,
w
it
h
th
e
e
x
ce
p
t
io
n
of
th
e
L
1
r
esu
lt
,
w
h
ic
h
w
as
h
i
g
h
at
9
7
.
0
%.
T
h
e
d
if
f
er
en
ce
in
ac
cu
r
ac
y
i
s
al
s
o
r
ef
lecte
d
i
n
t
h
e
r
ec
all,
F
-
m
ea
s
u
r
e,
a
n
d
s
p
e
cif
icit
y
s
co
r
es
,
w
h
er
e
th
e
p
-
HOG
(
L
1
)
r
es
u
lt
s
s
u
r
p
ass
(
w
it
h
s
tatis
t
ical
s
ig
n
i
f
ica
n
ce
)
t
h
e
r
e
s
u
lt
s
o
f
t
h
e
o
r
ig
i
n
al
H
OG
al
g
o
r
ith
m
a
n
d
th
o
s
e
o
f
t
h
e
p
-
HO
G
al
g
o
r
ith
m
w
it
h
o
t
h
er
n
o
r
m
v
alu
e
s
.
A
g
ai
n
,
L
1
s
h
o
w
ed
b
etter
r
esu
lts
t
h
a
n
an
y
o
t
h
er
n
o
r
m
v
alu
e
s
an
d
t
h
e
o
r
ig
i
n
al
HOG
alg
o
r
ith
m
,
s
tr
ess
i
n
g
th
e
s
a
m
e
co
n
clu
s
io
n
as b
ef
o
r
e.
T
ab
le
2
.
T
h
e
r
esu
lts
of
S
VM
w
it
h
R
B
F
k
er
n
e
l
u
s
in
g
HOG
a
n
d
p
-
HOG
w
it
h
d
i
f
f
er
e
n
t n
o
r
m
s
RBF
K
e
r
n
e
l
N
o
r
m
P
r
e
c
i
si
o
n
R
e
c
a
l
l
F
-
m
e
a
su
r
e
S
p
e
c
i
f
i
c
i
t
y
A
c
c
u
r
a
c
y
p
-
HOG
1
1
0
0
.
0
%
9
3
.
1
%
9
6
.
4
%
1
0
0
.
0
%
9
7
.
0
%
2
9
8
.
0
%
9
1
.
0
%
9
4
.
4
%
9
8
.
0
%
9
4
.
5
%
10
1
0
0
.
0
%
9
1
.
7
%
9
5
.
7
%
1
0
0
.
0
%
9
5
.
5
%
20
1
0
0
.
0
%
8
9
.
7
%
9
4
.
6
%
1
0
0
.
0
%
9
5
.
0
%
∞
1
0
0
.
0
%
8
9
.
8
%
9
4
.
6
%
1
0
0
.
0
%
9
5
.
0
%
HOG
9
8
.
6
%
9
1
.
7
%
9
5
.
0
%
9
8
.
4
%
9
5
.
2
%
Fin
all
y
,
in
T
ab
le
3
,
th
e
r
esu
lt
s
u
s
i
n
g
th
e
s
i
g
m
o
id
k
er
n
el
s
h
o
w
ed
b
etter
r
esu
lts
in
g
en
er
al
f
o
r
p
-
HOG
o
v
er
t
h
e
o
r
ig
i
n
al
HOG
.
Ho
w
e
v
er
,
b
ased
o
n
th
e
t
-
te
s
t
r
esu
lts
,
t
h
e
d
i
f
f
er
e
n
ce
s
ar
e
n
o
t
s
tati
s
tical
l
y
s
ig
n
i
f
ica
n
t
e
x
ce
p
t
f
o
r
t
h
e
r
e
ca
ll
v
al
u
e
,
w
h
er
e
p
-
H
OG
s
h
o
w
s
s
tatis
ticall
y
s
ig
n
i
f
ica
n
t
i
m
p
r
o
v
e
m
en
t.
I
n
g
en
er
al,
e
x
p
lo
r
in
g
d
if
f
er
en
t
p
-
n
o
r
m
v
a
lu
e
s
en
h
a
n
ce
d
th
e
S
VM
’s
p
er
f
o
r
m
a
n
ce
i
n
clas
s
if
y
in
g
i
m
a
g
es.
T
h
i
s
r
esu
lt
r
e
v
ea
ls
t
h
at
u
s
i
n
g
p
-
HO
G
w
i
th
L
1
to
d
etec
t
+
C
OVI
D
-
1
9
X
-
r
a
y
i
m
ag
e
s
is
p
r
o
m
i
s
i
n
g
.
Ou
r
r
esu
lts
s
tr
es
s
th
at
L
1
i
s
r
o
b
u
s
t
to
w
ar
d
s
o
u
tli
er
s
an
d
co
n
s
eq
u
e
n
tl
y
ac
h
ie
v
es
b
etter
r
es
u
lts
in
d
etec
tin
g
+
C
OVI
D
-
19
X
-
r
a
y
i
m
a
g
es
.
Dete
cti
n
g
ed
g
es
i
n
X
-
r
ay
s
o
cc
u
r
s
w
h
en
co
lo
r
s
ch
a
n
g
e
f
r
o
m
w
h
ite
to
b
lack
(
o
r
g
r
a
y
)
,
o
r
v
ice
v
er
s
a
,
w
h
ic
h
th
u
s
h
as a
lar
g
e
g
r
ad
ien
t
m
ag
n
it
u
d
e
co
m
p
ar
ed
w
i
th
c
h
an
g
i
n
g
co
lo
r
s
g
r
ad
u
all
y
.
T
ab
le
3
.
T
h
e
r
esu
lts
o
f
S
VM
w
it
h
s
ig
m
o
id
k
er
n
e
l u
s
in
g
HO
G
an
d
p
-
HOG
w
it
h
d
i
f
f
er
e
n
t n
o
r
m
s
si
g
mo
i
d
K
e
r
n
e
l
N
o
r
m
P
r
e
c
i
si
o
n
R
e
c
a
l
l
F
-
m
e
a
su
r
e
S
p
e
c
i
f
i
c
i
t
y
A
c
c
u
r
a
c
y
p
-
HOG
1
1
0
0
.
0
%
9
2
.
4
%
9
6
.
1
%
1
0
0
.
0
%
9
6
.
0
%
2
9
8
.
0
%
9
1
.
8
%
9
4
.
8
%
9
8
.
0
%
9
5
.
0
%
10
1
0
0
.
0
%
9
2
.
8
%
9
6
.
2
%
1
0
0
.
0
%
9
5
.
5
%
20
1
0
0
.
0
%
9
0
.
3
%
9
4
.
9
%
1
0
0
.
0
%
9
5
.
0
%
∞
1
0
0
.
0
%
9
0
.
2
%
9
4
.
8
%
1
0
0
.
0
%
9
5
.
5
%
HOG
9
8
.
8
%
9
1
.
8
%
9
5
.
1
%
9
8
.
8
%
9
5
.
1
%
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
p
-
n
o
r
ms o
f h
is
to
g
r
a
m
o
f o
r
ien
ted
g
r
a
d
ien
ts
fo
r
X
-
r
a
y
ima
g
es
(
N
u
h
a
H.
Ha
ma
d
a
)
4429
4.
CO
NCLU
SI
O
N
I
n
th
i
s
p
ap
er
,
w
e
p
r
o
p
o
s
ed
p
-
HOG
f
ea
t
u
r
e
d
etec
tio
n
.
T
h
e
m
ai
n
id
ea
o
f
p
-
HOG
i
s
to
ap
p
l
y
d
if
f
er
en
t
n
o
r
m
v
al
u
es
w
h
en
ca
lcu
lat
in
g
g
r
ad
ien
t
m
a
g
n
itu
d
e.
Gr
ad
ien
t
co
m
p
u
tatio
n
d
etec
to
r
p
er
f
o
r
m
an
ce
is
s
e
n
s
iti
v
e
to
h
o
w
g
r
ad
ien
ts
ar
e
co
m
p
u
ted
.
Di
f
f
er
e
n
t
p
-
n
o
r
m
s
we
r
e
test
ed
o
n
X
-
r
a
y
i
m
ag
e
s
o
f
C
O
VI
D
-
1
9
ca
s
es.
O
u
r
e
x
p
er
i
m
e
n
tal
r
e
s
u
l
ts
s
h
o
w
t
h
a
t
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
-
H
OG
d
escr
ip
t
o
r
w
as
s
i
g
n
if
ican
tl
y
h
i
g
h
er
t
h
an
t
h
e
HOG
d
escr
ip
to
r
in
ter
m
s
o
f
p
r
ec
is
io
n
,
ac
cu
r
ac
y
,
a
n
d
s
p
ec
if
icit
y
,
r
e
g
ar
d
less
o
f
SVM
k
e
r
n
el.
T
h
e
L
1
n
o
r
m
s
h
o
w
s
t
h
e
b
est
r
e
s
u
l
ts
,
e
m
p
h
a
s
iz
in
g
th
e
r
o
b
u
s
t
n
ess
o
f
th
is
n
o
r
m
an
d
i
ts
r
e
s
is
ta
n
ce
to
o
u
tli
e
r
s
in
X
-
r
a
y
i
m
a
g
es
,
w
h
ic
h
m
ai
n
l
y
in
c
lu
d
e
th
e
co
lo
r
s
w
h
ite,
b
lack
,
an
d
g
r
a
y
.
Ho
w
e
v
er
,
o
n
e
o
f
o
u
r
li
m
i
tatio
n
s
in
th
is
r
esear
ch
w
a
s
s
m
al
l
s
a
m
p
le
s
ize
d
u
e
to
r
estr
ictio
n
s
on
r
elea
s
i
n
g
C
O
VI
D
-
1
9
d
ata.
I
n
f
u
t
u
r
e
w
o
r
k
,
w
e
i
n
ten
d
to
i
n
v
esti
g
at
e
th
e
ef
f
ec
ti
v
e
n
es
s
o
f
p
-
H
OG
f
o
r
d
etec
ti
n
g
o
th
er
c
h
est d
is
ea
s
e
s
in
X
-
r
a
y
i
m
a
g
es.
RE
F
E
R
E
NC
E
S
[1
]
F
.
F
.
Kh
a
rb
a
t,
T
.
El
a
m
s
y
,
A
.
M
a
h
m
o
u
d
a
n
d
R.
A
b
d
u
ll
a
h
,
"
Im
a
g
e
F
e
a
tu
re
De
tec
to
rs
f
o
r
De
e
p
fa
k
e
V
id
e
o
De
tec
ti
o
n
,
"
2
0
1
9
IEE
E/
ACS
1
6
t
h
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
ter
S
y
ste
ms
a
n
d
Ap
p
li
c
a
ti
o
n
s
(
AICCS
A)
,
A
b
u
Dh
a
b
i
,
Un
it
e
d
A
ra
b
Em
irate
s,
2
0
1
9
,
p
p
.
1
-
4
,
d
o
i:
1
0
.
1
1
0
9
/A
ICCS
A
4
7
6
3
2
.
2
0
1
9
.
9
0
3
5
3
6
0
.
[2
]
N.
Da
lal
a
n
d
B.
T
rig
g
s,
"
Histo
g
ra
m
s
o
f
o
rien
ted
g
ra
d
ien
ts
f
o
r
h
u
m
a
n
d
e
tec
ti
o
n
,
"
2
0
0
5
IEE
E
Co
mp
u
ter
S
o
c
iet
y
Co
n
fer
e
n
c
e
o
n
Co
mp
u
ter
Vi
sio
n
a
n
d
Pa
tt
e
rn
Rec
o
g
n
it
io
n
(
C
VP
R
'
0
5
)
,
S
a
n
Die
g
o
,
CA
,
US
A
,
v
o
l
.
1
,
2
0
0
5
,
p
p
.
8
8
6
-
893
,
d
o
i:
1
0
.
1
1
0
9
/C
VP
R.
2
0
0
5
.
1
7
7
.
[3
]
M
.
G
h
o
rb
a
n
i,
A
.
T
.
T
a
r
g
h
i
a
n
d
M
.
M
.
De
h
sh
i
b
i,
"
HO
G
a
n
d
L
BP
:
T
o
w
a
rd
s
a
ro
b
u
st
f
a
c
e
re
c
o
g
n
it
io
n
sy
ste
m
,
"
2
0
1
5
T
e
n
th
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Dig
it
a
l
I
n
fo
rm
a
ti
o
n
M
a
n
a
g
e
me
n
t
(
ICDIM
)
,
Je
ju
,
Ko
re
a
(S
o
u
t
h
),
2
0
1
5
,
p
p
.
1
3
8
-
1
4
1
,
d
o
i:
1
0
.
1
1
0
9
/ICDIM
.
2
0
1
5
.
7
3
8
1
8
6
0
.
[4
]
T
.
W
a
t
a
n
a
b
e
,
S
.
It
o
,
a
n
d
K.
Yo
k
o
i,
“
Co
-
o
c
c
u
rre
n
c
e
Histo
g
ra
m
s
o
f
Orie
n
ted
G
ra
d
ien
ts
f
o
r
P
e
d
e
strian
De
tec
ti
o
n
,
”
i
n
Ad
v
a
n
c
e
s i
n
Ima
g
e
a
n
d
V
id
e
o
T
e
c
h
n
o
lo
g
y
,
p
p
.
3
7
-
4
7
,
2
0
0
9
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
5
4
0
-
9
2
9
5
7
-
4
_
4
.
[5
]
X
.
W
a
n
g
,
T
.
X
.
Ha
n
a
n
d
S
.
Ya
n
,
"
A
n
HO
G
-
L
BP
h
u
m
a
n
d
e
tec
to
r
w
it
h
p
a
rti
a
l
o
c
c
lu
sio
n
h
a
n
d
l
in
g
,
"
2
0
0
9
IE
EE
1
2
t
h
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
C
o
mp
u
ter
Vi
si
o
n
,
Ky
o
to
,
Ja
p
a
n
,
2
0
0
9
,
p
p
.
3
2
-
3
9
,
d
o
i
:
1
0
.
1
1
0
9
/ICC
V
.
2
0
0
9
.
5
4
5
9
2
0
7
.
[6
]
Y.
Zh
a
o
,
Y.
Zh
a
n
g
,
R.
Ch
e
n
g
,
D.
W
e
i
a
n
d
G
.
L
i,
"
A
n
En
h
a
n
c
e
d
Histo
g
ra
m
o
f
Orie
n
ted
G
ra
d
ien
ts
f
o
r
P
e
d
e
strian
De
tec
ti
o
n
,
"
in
IEE
E
I
n
telli
g
e
n
t
T
ra
n
sp
o
rta
ti
o
n
S
y
ste
ms
M
a
g
a
zin
e
,
v
o
l.
7
,
n
o
.
3
,
p
p
.
2
9
-
3
8
,
F
a
ll
2
0
1
5
,
d
o
i:
1
0
.
1
1
0
9
/M
IT
S
.
2
0
1
5
.
2
4
2
7
3
6
6
.
[7
]
Y.
Ch
o
i,
S
.
Je
o
n
g
,
a
n
d
M
.
L
e
e
,
“
F
e
a
tu
re
S
e
lec
ti
o
n
f
o
r
HO
G
De
s
c
rip
to
r
Ba
se
d
o
n
G
re
e
d
y
A
lg
o
rit
h
m
,
”
in
Ne
u
ra
l
In
fo
rm
a
t
io
n
Pro
c
e
ss
in
g
.
ICONI
P
2
0
1
3
.
L
e
c
tu
re
No
tes
in
C
o
m
p
u
ter
S
c
ien
c
e
,
v
o
l.
8
2
2
8
,
p
p
.
4
1
7
–
4
2
4
,
2
0
1
3
,
d
o
i:
1
0
.
1
0
0
7
/9
7
8
-
3
-
6
4
2
-
4
2
0
5
1
-
1
_
5
2
.
[8
]
R.
P
.
Y
a
d
a
v
,
V
.
S
e
n
th
a
m
il
a
ra
su
,
K.
Ku
tt
y
,
a
n
d
S
.
P
.
Ug
a
le,
“
I
m
p
le
m
e
n
tatio
n
o
f
ro
b
u
st HOG
-
S
V
M
b
a
se
d
p
e
d
e
strian
c
las
si
f
ica
ti
o
n
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Co
mp
u
ter
Ap
p
li
c
a
t
io
n
s
,
v
o
l.
1
1
4
,
n
o
.
1
9
,
p
p
.
1
0
-
1
6
,
2
0
1
5
,
d
o
i:
1
0
.
5
1
2
0
/2
0
0
8
4
-
2
0
2
6
.
[9
]
Y.
L
iu
,
J.
Ya
o
,
R.
X
ie
a
n
d
S
.
Zh
u
,
"
P
e
d
e
strian
d
e
tec
ti
o
n
f
ro
m
st
il
l
im
a
g
e
s
b
a
se
d
o
n
m
u
lt
i
-
f
e
a
tu
re
c
o
v
a
rian
c
e
s,"
2
0
1
3
I
EE
E
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
In
f
o
rm
a
ti
o
n
a
n
d
Au
to
ma
ti
o
n
(
ICIA)
,
Yi
n
c
h
u
a
n
,
C
h
i
n
a
,
2
0
1
3
,
p
p
.
6
1
4
-
6
1
9
,
d
o
i:
1
0
.
1
1
0
9
/ICIn
f
A
.
2
0
1
3
.
6
7
2
0
3
7
0
.
[1
0
]
W
.
Ru
d
in
,
“
F
u
n
c
ti
o
n
a
l
a
n
a
ly
sis.
1
9
9
1
,
”
In
ter
n
a
t.
S
e
r.
P
u
re
Ap
p
l.
M
a
th
,
1
9
9
1
.
[1
1
]
I.
S
in
g
e
r,
“
Be
st ap
p
ro
x
im
a
ti
o
n
in
n
o
rm
e
d
li
n
e
a
r
sp
a
c
e
s b
y
e
le
m
e
n
ts o
f
li
n
e
a
r
su
b
s
p
a
c
e
s,”
Ne
w
Yo
rk
,
1
9
7
0
.
[1
2
]
G
.
Be
li
tsk
ii
a
n
d
o
th
e
rs,
“
M
a
tri
x
n
o
rm
s a
n
d
th
e
ir
a
p
p
li
c
a
ti
o
n
s
,
”
Birk
h
ä
u
se
r
v
e
rlag
b
a
se
l,
1
9
8
8
.
[1
3
]
S
.
S
ra
,
S
.
No
w
o
z
in
,
a
n
d
S
.
J.
W
rig
h
t,
“
Op
ti
m
iza
ti
o
n
f
o
r
m
a
c
h
in
e
lea
rn
in
g
,”
Ca
m
b
rid
g
e
,
M
it
P
re
ss
,
2
0
1
2
.
[1
4
]
S
.
B.
Ko
tsian
ti
s,
I.
D.
Za
h
a
ra
k
is,
a
n
d
P
.
E
.
P
i
n
tela
s,
“
M
a
c
h
in
e
lea
rn
in
g
:
a
re
v
iew
o
f
c
la
ss
i
f
ica
ti
o
n
a
n
d
c
o
m
b
in
i
n
g
tec
h
n
iq
u
e
s,”
Arti
fi
c
ia
l
In
telli
g
e
n
c
e
Rev
iew
.
,
v
o
l.
2
6
,
n
o
.
3
,
p
p
.
1
5
9
-
1
9
0
,
2
0
0
6
,
d
o
i:
1
0
.
1
0
0
7
/s
1
0
4
6
2
-
0
0
7
-
9
0
5
2
-
3.
[1
5
]
I.
G
o
o
d
f
e
ll
o
w
,
Y.
Be
n
g
io
,
a
n
d
A
.
Co
u
rv
il
le
,
“
De
e
p
L
e
a
rn
in
g
,
”
M
IT
P
re
ss
,
2
0
1
6
.
[1
6
]
F
.
Nie
,
H.
Hu
a
n
g
,
X.
Ca
i,
a
n
d
C.
H.
Di
n
g
,
“
Ef
f
i
c
ien
t
a
n
d
ro
b
u
st
f
e
a
tu
re
se
lec
ti
o
n
v
ia
jo
i
n
t
ℓ2
,
1
-
n
o
rm
s
m
in
i
m
iza
ti
o
n
,
”
in
A
d
v
a
n
c
e
s in
n
e
u
ra
l
i
n
f
o
rm
a
ti
o
n
p
ro
c
e
ss
in
g
sy
ste
ms
,
v
o
l.
2
3
,
p
p
.
1
8
1
3
-
1
8
2
1
,
2
0
1
0
.
[1
7
]
P
.
S
.
Bra
d
ley
a
n
d
O.
L
.
M
a
n
g
a
sa
rian
,
“
F
e
a
tu
re
se
lec
ti
o
n
v
ia
c
o
n
c
a
v
e
m
in
im
i
z
a
t
io
n
a
n
d
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
s
,
”
in
ICM
L
,
v
o
l
.
9
8
,
p
p
.
8
2
-
9
0
,
1
9
9
8
.
[1
8
]
J.
-
Y.
T
a
n
,
C.
-
H.
Zh
a
n
g
,
a
n
d
N.
-
Y.
De
n
g
,
“
Ca
n
c
e
r
re
late
d
g
e
n
e
id
e
n
ti
f
ica
ti
o
n
v
ia
p
-
n
o
rm
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
,
”
in
T
h
e
4
t
h
i
n
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
c
o
mp
u
t
a
ti
o
n
a
l
sy
ste
ms
b
io
l
o
g
y
,
v
o
l
.
1
,
p
p
.
1
0
1
-
1
0
8
,
2
0
1
0
.
[1
9
]
W
.
H.
Org
a
n
iza
ti
o
n
a
n
d
o
t
h
e
rs,
“
Na
m
in
g
th
e
c
o
ro
n
a
v
iru
s
d
ise
a
se
(COV
ID
-
1
9
)
a
n
d
th
e
v
ir
u
s th
a
t
c
a
u
se
s it
,
”
2
0
2
0
.
[2
0
]
Y.
Zh
o
u
e
t
a
l.
,
“
Cli
n
ica
l
re
p
o
r
t
s
o
n
e
a
rly
d
iag
n
o
sis
o
f
n
o
v
e
l
c
o
ro
n
a
v
iru
s
(2
0
1
9
-
n
Co
V
)
p
n
e
u
m
o
n
ia
in
ste
a
lt
h
in
f
e
c
ted
p
a
ti
e
n
ts,”
Pre
p
rin
ts
,
v
o
l.
1
,
p
p
.
1
-
8
,
2
0
2
0
,
d
o
i:
1
0
.
2
0
9
4
4
/p
r
e
p
rin
ts2
0
2
0
0
2
.
0
1
5
6
.
v
1
.
[2
1
]
C.
S
o
h
ra
b
i
e
t
a
l.
,
“
W
o
rl
d
He
a
lt
h
Org
a
n
iza
ti
o
n
d
e
c
lare
s
g
lo
b
a
l
e
m
e
rg
e
n
c
y
:
A
re
v
ie
w
o
f
th
e
2
0
1
9
n
o
v
e
l
c
o
ro
n
a
v
iru
s
(COV
ID
-
1
9
),
”
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
S
u
r
g
e
ry
,
v
o
l.
7
6
,
p
p
.
7
1
-
7
6
,
A
p
r.
2
0
2
0
,
d
o
i:
1
0
.
1
0
1
6
/j
.
ij
s
u
.
2
0
2
0
.
0
2
.
0
3
4
.
[2
2
]
W
.
H.
O.
W
HO
,
“
W
HO
Dire
c
t
o
r
-
G
e
n
e
ra
l’s
o
p
e
n
in
g
re
m
a
rk
s
a
t
th
e
m
e
d
ia
b
rief
in
g
o
n
C
OV
ID
-
19
-
1
1
M
a
rc
h
2
0
2
0
,
”
2
0
2
0
,
[
On
li
n
e
]
.
A
v
a
ib
le:
h
tt
p
s://
w
ww
.
w
h
o
.
in
t/
d
irec
to
r
-
g
e
n
e
ra
l/
sp
e
e
c
h
e
s/d
e
tail/
w
h
o
-
d
ire
c
to
r
-
g
e
n
e
ra
l
-
s
-
o
p
e
n
i
n
g
-
re
m
a
rk
s
-
at
-
th
e
-
m
e
d
ia
-
b
ri
e
f
in
g
-
on
-
c
o
v
id
-
19
---
11
-
m
a
rc
h
-
2
0
2
0
.
[2
3
]
“
COV
ID
-
1
9
Co
r
o
n
a
v
iru
s
P
a
n
d
e
m
ic,”
wo
rld
o
me
ter
s
,
2
0
2
0
.
[
O
n
li
n
e
]
.
A
v
a
ib
le:
h
tt
p
s:/
/www
.
w
o
rld
o
m
e
ters
.
in
f
o
/c
o
ro
n
a
v
iru
s/.
[2
4
]
Q.
Ha
n
,
Q.
L
in
,
S
.
Ji
n
,
a
n
d
L
.
Yo
u
,
“
Co
ro
n
a
v
iru
s
2
0
1
9
-
n
Co
V
:
A
b
rief
p
e
rsp
e
c
ti
v
e
f
ro
m
th
e
f
ro
n
t
li
n
e
,
”
J
o
u
r
n
a
l
o
f
In
fec
ti
o
n
,
v
o
l.
8
0
,
n
o
.
4
,
p
p
.
3
7
3
-
3
7
7
,
A
p
r.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/j
.
ji
n
f
.
2
0
2
0
.
0
2
.
0
1
0
.
[2
5
]
J.
Co
h
e
n
.
,
“
Co
v
i
d
c
h
e
st
x
-
ra
y
d
a
t
a
se
t,
”
2
0
2
0
.
[
O
n
li
n
e
]
.
A
v
a
ib
le:
h
tt
p
s://
g
it
h
u
b
.
c
o
m
/i
e
e
e
8
0
2
3
/co
v
id
-
%
0
A
c
h
e
st
x
ra
y
-
d
a
ta
se
t.
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.
11
,
No
.
5
,
Octo
b
e
r
2
0
2
1
:
4
4
2
3
-
4430
4430
[2
6
]
S
.
F
e
n
g
,
H.
Z
h
o
u
,
a
n
d
H.
D
o
n
g
,
“
Us
in
g
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
w
it
h
sm
a
ll
d
a
tas
e
t
to
p
re
d
ict
m
a
teria
l
d
e
f
e
c
ts,”
M
a
ter
ia
ls
&
De
sig
n
,
v
o
l.
1
6
2
,
p
p
.
3
0
0
-
3
1
0
,
Ja
n
.
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
m
a
td
e
s.2
0
1
8
.
1
1
.
0
6
0
.
[2
7
]
P
.
M
o
o
n
e
y
,
“
Ch
e
st x
-
ra
y
i
m
a
g
e
s (
p
n
e
u
m
o
n
ia),”
k
a
g
g
le
,
2
0
1
8
.
[2
8
]
A
.
Ro
se
b
ro
c
k
,
“
D
e
tec
ti
n
g
COV
ID
-
1
9
in
X
-
ra
y
i
m
a
g
e
s
w
it
h
Ke
r
a
s,
Ten
so
rF
lo
w
,
a
n
d
De
e
p
L
e
a
rn
in
g
,
”
[
O
n
li
n
e
]
.
Av
a
ib
le:
h
tt
p
s//www
.
p
y
i
m
a
g
e
s
e
a
rc
h
.
c
o
m
/2
0
2
0
/0
3
/1
6
/d
e
tec
ti
n
g
-
c
o
v
id
-
19
-
in
-
x
-
ra
y
i
m
a
g
e
s
-
w
it
h
-
k
e
ra
s
-
ten
so
rf
lo
w
-
a
n
d
-
d
e
e
p
-
lea
rn
i
n
g
.
[2
9
]
G
.
L
o
w
e
,
“
S
IF
T
-
th
e
sc
a
le i
n
v
a
ria
n
t
f
e
a
tu
re
tran
sf
o
r
m
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
,
v
o
l.
2
,
p
p
.
9
1
–
1
1
0
,
2
0
0
4
.
[3
0
]
J.
Ko
ss
a
if
i,
“
V
e
c
to
rise
d
im
p
lem
e
n
tatio
n
o
f
th
e
h
ist
o
g
ra
m
o
f
o
rien
ted
g
ra
d
ien
ts,”
2
0
2
0
.
[
O
n
li
n
e
]
.
A
v
a
ib
le:
h
tt
p
s:/
/g
it
h
u
b
.
c
o
m
/Je
a
n
Ko
ss
a
if
i/
p
y
th
o
n
-
ho
g.
[3
1
]
N.
V
.
Ch
a
w
la,
“
Da
ta
M
in
in
g
f
o
r
Im
b
a
lan
c
e
d
D
a
tas
e
ts:
A
n
O
v
e
r
v
i
e
w
,
”
in
Da
ta
M
in
in
g
a
n
d
Kn
o
wl
e
d
g
e
Disc
o
v
e
ry
Ha
n
d
b
o
o
k
,
Bo
st
o
n
,
M
A
:
S
p
ri
n
g
e
r
US,
2
0
0
9
,
p
p
.
8
7
5
–
8
8
6
,
d
o
i:
1
0
.
1
0
0
7
/9
7
8
-
0
-
3
8
7
-
0
9
8
2
3
-
4
_
4
5
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
Nuh
a
H
.
H
a
m
a
d
a
is
a
n
a
ss
istan
t
p
ro
f
e
ss
o
r
w
o
rk
in
g
a
t
A
l
A
in
Un
iv
e
rsit
y
,
A
b
u
Dh
a
b
i,
UA
E.
S
h
e
h
a
s c
o
m
p
lete
d
h
e
r
P
h
.
D.
o
n
F
u
n
c
t
io
n
a
l
A
n
a
ly
sis
-
Hilb
e
rt
sp
a
c
e
s
f
ro
m
Un
iv
e
rsit
y
o
f
B
a
g
h
d
a
d
.
He
r
P
h
.
D.
th
e
sis
a
d
d
re
ss
e
d
th
e
Jo
rd
a
n
*
-
d
e
riv
a
ti
o
n
o
n
th
e
a
lg
e
b
ra
o
f
a
ll
b
o
u
n
d
e
d
li
n
e
a
r
o
p
e
ra
to
rs
o
n
se
p
a
ra
b
le
in
f
in
it
e
d
ime
n
sio
n
a
l
c
o
m
p
lex
Hilb
e
rt
sp
a
c
e
.
He
r
r
e
se
a
r
c
h
in
tere
sts
in
c
lu
d
e
su
p
e
rc
y
c
li
c
o
p
e
ra
to
rs,
c
y
c
li
c
p
h
e
n
o
m
e
n
a
a
n
d
Ch
a
o
s T
h
e
o
ry
.
In
a
d
d
it
io
n
,
sh
e
h
a
s a
c
o
n
tri
b
u
ti
o
n
o
n
th
e
a
re
a
o
f
q
u
a
n
t
it
a
ti
v
e
a
n
a
l
y
sis
in
m
a
n
a
g
e
m
e
n
t,
c
h
a
o
s
th
e
o
ry
f
ro
m
d
e
c
isio
n
-
m
a
k
in
g
p
e
rsp
e
c
ti
v
e
,
a
p
p
ly
in
g
sta
ti
stica
l
tec
h
n
iq
u
e
to
in
v
e
stig
a
te
d
iff
icu
lt
ies
in
lea
rn
in
g
a
n
d
f
i
n
d
so
m
e
c
u
lt
u
ra
l
f
a
c
to
rs
th
a
t
a
ffe
c
t
th
e
p
ro
c
e
ss
o
f
te
a
c
h
in
g
a
n
d
lea
rn
i
n
g
a
n
d
stu
d
y
in
g
h
o
w
LM
S
c
a
n
su
p
p
o
rt
tea
c
h
i
n
g
a
n
d
lea
rn
in
g
.
Fa
te
n
F.
K
h
a
r
b
a
t
h
o
ld
s
P
h
D
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
:
Da
ta
M
in
in
g
a
n
d
Kn
o
w
led
g
e
Disc
o
v
e
r
y
f
ro
m
th
e
Un
iv
e
rsit
y
o
f
th
e
W
e
st o
f
En
g
lan
d
,
UK
.
He
r
M
S
c
a
n
d
BS
c
w
e
re
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Un
iv
e
rsit
y
o
f
Jo
rd
a
n
.
He
r
m
a
in
re
se
a
r
c
h
in
tere
st
is
C
a
n
c
e
r
c
a
re
In
f
o
r
m
a
ti
c
s,
h
e
a
lt
h
in
f
o
rm
a
ti
c
s,
ima
g
e
p
ro
c
e
ss
in
g
,
c
o
n
ten
t
a
n
a
ly
sis,
a
p
p
li
e
d
d
a
ta
m
in
in
g
,
a
n
d
re
c
e
n
tl
y
w
a
s
in
v
o
lv
e
d
in
q
u
a
li
ty
o
f
h
ig
h
e
r
e
d
u
c
a
ti
o
n
.
S
h
e
is
a
n
A
ss
o
c
iate
P
ro
f
e
ss
o
r
in
Co
m
p
u
ter
S
c
ien
c
e
a
t
A
l
A
in
Un
iv
e
rsit
y
,
A
b
u
Dh
a
b
i
Ca
m
p
u
s,
UA
E.
A
t
th
e
m
o
m
e
n
t,
sh
e
is
th
e
d
e
p
u
ty
d
e
a
n
f
o
r
th
e
Co
ll
e
g
e
o
f
En
g
in
e
e
rin
g
in
A
b
u
Dh
a
b
i
Ca
m
p
u
s.
Evaluation Warning : The document was created with Spire.PDF for Python.