I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
36
,
No
.
3
,
Dec
em
b
er
20
24
,
p
p
.
1
5
32
~1
5
42
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
36.
i
3
.
p
p
1
5
32
-
1
5
42
1532
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Enha
nced driving
as
sis
tance: auto
ma
ted
da
y
and ni
g
ht
v
ehicle
detec
tion sy
stem
utilizing
co
nv
o
lut
io
na
l neural net
w
o
rks
Abdelm
o
g
hit
Z
a
a
ra
ne,
I
btiss
a
m
Sli
m
a
ni,
M
o
ura
d E
l
h
a
bchi,
I
s
s
a
m
At
o
uf
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
s
B
e
n
M
’
si
c
k
,
H
a
ss
a
n
I
I
U
n
i
v
e
r
s
i
t
y
o
f
C
a
sa
b
l
a
n
c
a
,
C
a
sa
b
l
a
n
c
a
,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ap
r
29
,
2
0
2
4
R
ev
is
ed
Au
g
24
,
2
0
2
4
Acc
ep
ted
A
ug
31
,
2
0
2
4
Th
is
p
a
p
e
r
p
re
se
n
ts
a
n
e
n
h
a
n
c
e
d
re
a
l
-
ti
m
e
v
e
h
icle
d
e
tec
ti
o
n
s
y
s
tem
u
sin
g
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
s
(
CNN
s
)
fo
r
b
o
t
h
d
a
y
t
ime
a
n
d
n
ig
h
t
-
t
ime
c
o
n
d
i
ti
o
n
s.
I
n
it
iall
y
,
th
e
s
y
ste
m
d
e
term
in
e
s
th
e
ti
m
e
o
f
c
a
p
t
u
re
b
y
a
n
a
ly
z
in
g
th
e
u
p
p
e
r
p
a
rt
o
f
i
n
p
u
t
ima
g
e
s.
F
o
r
d
a
y
ti
m
e
d
e
tec
ti
o
n
,
it
u
se
s
n
o
rm
a
li
z
e
d
c
ro
ss
-
c
o
rre
latio
n
a
n
d
two
-
d
ime
n
sio
n
a
l
d
isc
re
te
wa
v
e
let
tran
sfo
rm
(2
D
-
D
WT
)
tec
h
n
iq
u
e
s.
Nig
h
t
-
ti
m
e
d
e
tec
ti
o
n
in
v
o
l
v
e
s
id
e
n
ti
f
y
i
n
g
v
e
h
i
c
le
lam
p
s
th
ro
u
g
h
c
o
l
o
r
t
h
re
sh
o
l
d
in
g
a
n
d
c
o
n
n
e
c
ted
c
o
m
p
o
n
e
n
t
tec
h
n
i
q
u
e
s,
fo
ll
o
we
d
b
y
sy
m
m
e
try
a
n
a
l
y
sis
a
n
d
CN
N
c
las
sifica
ti
o
n
.
Th
e
d
a
tas
e
t
fo
r
train
i
n
g
in
c
lu
d
e
s
ima
g
e
s
fro
m
th
e
Ca
lt
e
c
h
Ca
rs,
AO
L
P
,
KIT
TI
Visi
o
n
,
a
n
d
n
i
g
h
t
-
ti
m
e
v
e
h
icle
d
e
tec
ti
o
n
d
a
tas
e
ts,
e
n
su
rin
g
ro
b
u
st
p
e
rf
o
rm
a
n
c
e
a
c
ro
ss
v
a
rio
u
s
li
g
h
t
in
g
c
o
n
d
it
i
o
n
s.
Ex
p
e
rime
n
ts
d
e
m
o
n
stra
te
th
e
sy
ste
m
'
s
h
ig
h
a
c
c
u
ra
c
y
,
a
c
h
iev
in
g
9
9
.
2
%
d
u
rin
g
t
h
e
d
a
y
a
n
d
9
8
.
2
7
%
a
t
n
ig
h
t,
m
e
e
ti
n
g
re
a
l
-
ti
m
e
re
q
u
irem
e
n
ts an
d
e
n
h
a
n
c
in
g
d
ri
v
i
n
g
a
ss
istan
c
e
sy
ste
m
s'
re
li
a
b
il
it
y
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Day
-
tim
e
d
etec
tio
n
Dr
iv
in
g
ass
is
tan
ce
s
y
s
tem
s
I
m
ag
e
p
r
o
ce
s
s
in
g
Nig
h
t
-
tim
e
d
etec
tio
n
T
wo
-
d
im
en
s
io
n
al
d
is
cr
ete
wav
elet
tr
an
s
f
o
r
m
Veh
icle
d
etec
tio
n
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
:
Ab
d
elm
o
g
h
it Z
aa
r
an
e
Facu
lty
o
f
Scien
ce
s
B
en
M’
s
ick
,
Hass
an
I
I
Un
iv
er
s
ity
o
f
C
asab
lan
ca
Av
Dr
is
s
E
l
Har
ti Sid
i O
th
m
an
e
C
asab
lan
ca
B
.
P 7
9
5
5
,
Mo
r
o
cc
o
E
m
ail: z
.
ab
d
elm
o
g
h
it@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
R
ec
en
tly
,
th
e
d
r
iv
in
g
ass
is
ta
n
ce
s
y
s
tem
s
(
DAS
)
h
as
g
ain
ed
a
g
r
ea
t
co
n
s
id
er
atio
n
in
t
h
e
r
esear
ch
f
ield
,
in
lig
h
t
o
f
th
e
u
r
g
en
t
n
ee
d
to
a
u
to
m
ate
d
r
iv
in
g
s
y
s
tem
s
d
u
e
to
th
e
s
tr
o
n
g
g
r
o
wth
in
th
e
ac
cid
e
n
t
n
u
m
b
er
s
wh
ile
m
o
s
t
o
f
th
ese
ac
cid
en
ts
ar
e
ca
u
s
ed
b
y
h
u
m
a
n
ca
u
s
es.
Veh
icle
d
etec
tio
n
[
1
]
is
o
n
e
o
f
th
e
m
o
s
t
ch
allen
g
in
g
task
s
in
th
e
DAS
wh
er
e
it
ca
n
b
e
u
s
ed
f
o
r
s
ev
er
al
p
u
r
p
o
s
es,
s
u
ch
as
ca
lcu
latin
g
th
e
d
is
tan
ce
s
an
d
s
p
ee
d
s
o
f
o
th
er
v
e
h
icles
[
2
]
,
[
3
]
an
d
th
e
p
o
s
s
ib
ilit
y
o
f
o
v
er
t
ak
in
g
,
in
o
r
d
er
to
war
n
th
e
d
r
iv
er
to
s
lo
w
d
o
wn
an
d
av
o
i
d
co
llis
io
n
s
.
Ho
wev
e
r
,
th
er
e
ar
e
s
ev
er
al
d
if
f
ic
u
lties
an
d
co
n
s
tr
ain
ts
in
v
eh
icle
d
et
ec
tio
n
s
y
s
tem
s
d
u
e
to
th
e
d
y
n
a
m
ic
en
v
ir
o
n
m
e
n
t
a
n
d
th
e
co
m
p
lex
a
n
d
r
ap
id
m
o
v
em
en
t
o
f
o
b
jects
in
ter
f
e
r
in
g
in
th
e
r
o
ad
.
Sev
er
al
r
esear
ch
es
ar
e
d
o
n
e
to
d
ev
elo
p
ef
f
icien
t
v
eh
icle
d
etec
tio
n
s
y
s
tem
s
[
4
]
.
Ho
wev
er
,
m
o
s
t
o
f
r
esear
ch
er
s
ca
r
r
ied
o
u
t
th
eir
r
esear
ch
es
to
d
ea
l
with
v
e
h
icle
d
etec
tio
n
o
n
ly
in
d
ay
-
tim
e
in
s
u
itab
le
co
n
d
itio
n
s
,
wh
er
e
th
e
ass
is
tan
ce
is
m
o
r
e
n
ee
d
ed
in
n
ig
h
t
-
tim
e.
C
o
n
ce
r
n
in
g
th
e
wo
r
k
r
elate
d
t
o
th
e
d
etec
tio
n
d
u
r
i
n
g
d
a
y
-
ti
m
e.
R
u
an
et
a
l.
[
5
]
p
r
esen
t
a
s
y
s
tem
f
o
r
v
eh
icle
d
etec
tio
n
u
s
in
g
wh
ee
l
p
ar
t
id
en
tific
atio
n
.
T
h
eir
p
r
o
p
o
s
ed
m
eth
o
d
c
o
n
s
is
ts
o
f
two
m
ain
p
ar
ts
,
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
v
eh
icle
d
etec
ti
o
n
.
T
h
e
n
th
ey
d
etec
t
an
d
lo
ca
te
v
eh
icles
u
s
in
g
th
r
ee
d
if
f
er
e
n
t
s
u
b
m
o
d
u
les:
th
e
r
eg
io
n
o
f
in
ter
est
(
R
OI
)
s
eg
m
en
tatio
n
m
o
d
u
le,
th
e
w
h
ee
l
id
en
tific
atio
n
m
o
d
u
le,
a
n
d
t
h
e
v
eh
icle
d
etec
tio
n
m
o
d
u
le
.
W
ei
et
a
l
.
[
6
]
s
u
g
g
est
a
v
eh
icle
r
ec
o
g
n
itio
n
m
e
th
o
d
b
ased
o
n
Haa
r
an
d
h
is
to
g
r
am
o
f
o
r
ien
ted
g
r
ad
ien
ts
(
HOG
)
f
ea
tu
r
es,
wh
er
e
th
ey
s
eg
m
en
t
t
h
e
r
e
g
io
n
s
o
f
in
ter
est
b
ased
o
n
a
co
m
b
i
n
atio
n
b
etwe
en
Haa
r
an
d
HOG
f
ea
tu
r
es th
en
th
ey
cl
ass
if
y
th
em
u
s
in
g
Ad
aBo
o
s
t a
n
d
s
u
p
p
o
r
t v
ec
to
r
m
ac
h
i
n
e
s
(
SVMs
)
class
if
ier
.
Ho
wev
er
,
th
e
m
o
s
t
r
ec
en
t
w
o
r
k
h
as
alwa
y
s
b
ee
n
f
o
cu
s
e
d
o
n
th
e
u
s
e
o
f
YOL
O
alg
o
r
ith
m
s
f
o
r
d
etec
tio
n
,
y
o
u
o
n
ly
lo
o
k
o
n
c
e,
wh
ich
is
a
d
ee
p
lear
n
in
g
m
eth
o
d
.
T
h
is
YOL
O
alg
o
r
ith
m
was
p
r
o
p
o
s
ed
b
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
n
h
a
n
ce
d
d
r
ivin
g
a
s
s
is
ta
n
ce
:
a
u
to
ma
ted
d
a
y
a
n
d
n
ig
h
t v
e
h
i
cle
d
etec
tio
n
s
ystem
… (
A
b
d
el
mo
g
h
it Za
a
r
a
n
e
)
1533
R
ed
m
o
n
et
a
l.
[
7
]
.
I
t
is
a
s
in
g
le
-
s
tep
ap
p
r
o
ac
h
t
h
at
tr
an
s
lates
an
o
b
ject'
s
b
o
u
n
d
ar
y
b
o
x
p
o
s
itio
n
p
r
o
b
lem
d
ir
ec
tly
in
to
a
n
o
b
ject'
s
r
eg
r
ess
io
n
p
r
o
b
lem
with
o
u
t g
en
er
ati
n
g
ca
n
d
id
ate
b
o
x
es.
I
n
th
e
YO
L
O
f
r
am
ewo
r
k
,
th
e
im
ag
e
is
d
iv
id
e
d
in
t
o
a
ce
r
tain
n
u
m
b
er
o
f
g
r
id
s
.
E
ac
h
g
r
id
h
a
s
th
e
r
esp
o
n
s
ib
ilit
y
to
esti
m
ate
th
e
o
b
jects
with
in
th
e
g
r
i
d
th
at
h
av
e
a
ce
n
ter
.
So
m
e
o
f
th
e
au
th
o
r
s
w
h
o
h
a
v
e
u
s
ed
YOL
O,
C
h
en
an
d
L
i
[
8
]
u
s
e
o
n
t
h
e
YOL
Ov
3
alg
o
r
ith
m
a
n
d
t
h
e
s
in
g
le
-
s
h
o
t
d
etec
to
r
(
SSD
)
alg
o
r
ith
m
f
o
r
v
eh
icle
d
etec
tio
n
.
T
h
ey
f
ir
s
t
p
r
o
ce
s
s
th
e
in
p
u
t
im
ag
e
f
o
r
tr
ain
in
g
b
y
u
s
in
g
k
-
m
ea
n
s
f
o
r
clu
s
ter
in
g
th
e
b
o
u
n
d
in
g
b
o
x
o
f
th
e
ca
n
d
id
ates.
T
h
en
,
th
ey
tr
ain
ed
t
h
e
v
eh
icle
d
etec
tio
n
m
o
d
els
b
y
u
s
in
g
YOL
Ov
3
an
d
SS
D
a
lg
o
r
ith
m
s
.
Als
o
[
9
]
ar
e
d
ev
e
lo
p
ed
a
YOL
OV3
alg
o
r
ith
m
th
at
h
as
o
p
tim
al
p
er
f
o
r
m
an
ce
r
elativ
e
to
th
e
p
er
f
o
r
m
a
n
ce
o
f
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
,
f
o
r
au
to
n
o
m
o
u
s
v
eh
icle
im
a
g
e
r
ec
o
g
n
itio
n
.
C
o
n
ce
r
n
in
g
th
e
wo
r
k
r
elate
d
to
th
e
d
etec
tio
n
d
u
r
in
g
n
ig
h
t
-
tim
e.
Mo
s
t
ex
is
tin
g
v
e
h
icle
d
etec
tio
n
s
y
s
tem
s
an
d
m
eth
o
d
s
f
o
c
u
s
m
ain
ly
o
n
d
etec
tin
g
v
e
h
icles
in
d
ay
lig
h
t
c
o
n
d
itio
n
s
.
Ho
wev
er
,
s
tatis
tic
s
s
h
o
w
th
at
m
o
r
e
th
an
h
alf
o
f
all
d
a
n
g
er
o
u
s
r
o
ad
ac
cid
en
ts
o
cc
u
r
at
n
ig
h
t
.
Hen
ce
,
m
o
r
e
atten
tio
n
n
ee
d
s
to
b
e
p
aid
to
wo
r
k
o
n
v
eh
icle
d
etec
tio
n
at
th
is
ti
m
e
o
f
d
a
y
[
10
]
.
A
m
o
n
g
t
h
e
wo
r
k
s
f
o
c
u
s
in
g
o
n
n
ig
h
t
d
etec
tio
n
,
we
f
in
d
[
1
1
]
.
T
h
ey
p
r
o
p
o
s
e
a
f
r
o
n
t
v
eh
icles
d
etec
tio
n
s
y
s
tem
d
u
r
in
g
th
e
n
ig
h
t
-
tim
e
o
n
th
e
b
asis
o
f
b
o
th
laser
a
n
d
v
i
d
eo
i
n
f
o
r
m
atio
n
.
Firstl
y
,
th
ey
p
r
e
-
p
r
o
ce
s
s
ed
b
o
th
laser
in
f
o
r
m
atio
n
a
n
d
v
i
d
eo
f
r
am
es
u
s
in
g
th
e
r
eg
i
o
n
g
r
o
wth
an
d
th
r
esh
o
l
d
ar
ea
ex
p
u
ls
io
n
alg
o
r
ith
m
.
Nex
t,
th
ey
ex
tr
ac
ted
th
e
ch
ar
a
cter
is
tics
o
f
th
e
f
r
o
n
tal
v
eh
icles
u
s
in
g
th
e
Gab
o
r
f
ilter
wh
ich
u
s
es
th
e
p
r
in
cip
le
o
f
u
n
ce
r
tain
ty
,
an
d
th
ey
o
b
tai
n
ed
th
e
f
r
o
n
tal
v
eh
icle
d
is
tan
ce
s
v
ia
th
e
laser
p
o
in
t
clo
u
d
.
L
astl
y
,
th
ey
class
if
y
th
e
d
etec
ted
f
r
o
n
t v
e
h
icles b
ased
o
n
t
h
e
SVM
alg
o
r
ith
m
.
K
u
an
g
et
a
l.
[1
2
]
a
n
i
g
h
t
-
tim
e
m
eth
o
d
is
p
r
o
p
o
s
ed
f
o
r
d
et
ec
tin
g
v
eh
icles
u
s
in
g
th
e
co
m
b
in
atio
n
o
f
two
d
if
f
er
en
t
m
eth
o
d
s
to
in
cr
ea
s
e
th
e
ac
cu
r
ac
y
o
f
th
e
s
y
s
t
em
.
A
r
eg
io
n
o
f
in
ter
est
ex
tr
ac
tio
n
ap
p
r
o
ac
h
to
d
etec
t
v
eh
icle
lig
h
t
an
d
a
n
ig
h
t
-
tim
e
im
ag
e
im
p
r
o
v
em
e
n
t
ap
p
r
o
x
im
atio
n
o
n
th
e
b
asis
o
f
en
h
an
ce
d
m
u
lti
-
s
ca
le
r
etin
ex
(
MSR
)
f
o
r
ex
tr
ac
tin
g
a
p
r
ec
is
e
r
eg
io
n
o
f
in
ter
est
an
d
im
p
r
o
v
in
g
th
e
im
a
g
es
to
ac
cu
r
ately
d
etec
t
v
eh
icles
at
n
ig
h
t.
Hem
m
ati
et
a
l.
[1
3
]
p
r
esen
t
a
n
o
v
el
m
eth
o
d
f
o
r
n
ig
h
t
-
tim
e
v
eh
icle
d
etec
tio
n
b
ased
o
n
d
ee
p
b
elief
n
etwo
r
k
s
(
DB
N)
f
o
r
d
etec
tin
g
b
ac
k
lig
h
ts
in
im
ag
e
th
r
esh
o
ld
s
.
An
d
th
en
s
o
m
e
d
e
tecte
d
taillig
h
ts
ar
e
s
elec
ted
an
d
f
ed
to
an
SVM
class
if
ier
o
n
th
e
b
asis
o
f
th
e
s
ize
ch
ar
ac
ter
is
tics
r
etr
iev
ed
an
d
th
eir
d
is
tan
ce
to
lo
ca
lize
th
e
v
eh
icle.
C
o
m
b
in
in
g
th
e
Mo
b
ileNet
v
2
an
d
YOL
O
v
3
,
Hu
an
g
et
a
l.
[1
4
]
p
r
o
p
o
s
e
M
-
YOL
O
wh
ich
is
a
n
o
v
el
d
ee
p
n
eu
r
al
n
etwo
r
k
m
o
d
el.
I
n
itially
,
th
ey
ex
tr
ac
t
t
h
e
ch
a
r
ac
ter
is
tics
o
f
M
-
YOL
O
b
y
u
s
i
n
g
th
e
lig
h
tweig
h
t
n
etwo
r
k
M
o
b
ileNet
v
2
.
Su
b
s
e
q
u
en
tly
,
th
ey
u
s
e
th
e
alg
o
r
ith
m
“K
-
m
ea
n
s
”
t
o
r
e
o
r
g
a
n
ize
t
h
e
d
atab
ase
a
n
d
g
et
s
u
itab
le
an
d
ap
p
r
o
p
r
iate
a
n
c
h
o
r
b
o
x
es.
T
h
ir
d
,
th
ey
u
s
e
th
e
E
I
o
U
lo
s
s
f
u
n
ctio
n
t
o
o
p
tim
ize
th
e
m
o
d
el
co
n
tin
u
o
u
s
ly
.
Desp
ite
ad
v
an
ce
m
en
t
s
in
p
r
ev
io
u
s
wo
r
k
s
,
s
ev
er
al
p
r
o
b
lem
s
r
em
ain
u
n
r
eso
lv
e
d
,
th
e
m
ain
u
n
r
eso
lv
ed
p
r
o
b
lem
is
th
e
in
co
n
s
is
ten
cy
i
n
d
etec
tio
n
p
e
r
f
o
r
m
an
ce
ac
r
o
s
s
d
if
f
er
en
t
lig
h
tin
g
s
ce
n
ar
i
o
s
[
1
5
]
.
Ou
r
wo
r
k
aim
s
to
f
ill
th
is
g
a
p
b
y
d
e
v
elo
p
in
g
a
s
y
s
tem
th
at
m
ai
n
tain
s
h
ig
h
d
etec
tio
n
ac
c
u
r
ac
y
r
eg
a
r
d
less
o
f
am
b
ien
t
lig
h
t.
Sp
ec
if
ically
,
we
f
o
cu
s
o
n
im
p
r
o
v
in
g
d
etec
tio
n
r
ates
an
d
r
e
d
u
cin
g
f
alse
p
o
s
itiv
es,
th
er
eb
y
en
h
an
cin
g
o
v
e
r
all
s
y
s
tem
r
eliab
ilit
y
.
W
ith
in
th
is
r
eg
ar
d
,
th
e
g
o
al
o
f
th
is
p
a
p
er
is
to
p
r
o
p
o
s
e
an
en
h
an
ce
d
d
r
i
v
in
g
ass
is
tan
ce
f
o
r
v
e
h
icle
d
etec
tio
n
d
u
r
in
g
d
ay
a
n
d
n
ig
h
t.
T
h
e
wo
r
k
p
r
esen
ted
in
th
is
p
ap
er
is
b
ased
o
n
a
d
if
f
er
e
n
t
ef
f
icien
t
alg
o
r
ith
m
s
u
ch
as
n
o
r
m
alize
d
c
r
o
s
s
-
co
r
r
elatio
n
,
two
-
d
im
en
s
io
n
al
d
is
cr
ete
wav
elet
tr
an
s
f
o
r
m
(
2
D
-
DW
T
)
,
an
d
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN
s
)
.
T
h
e
co
n
tr
ib
u
tio
n
s
o
f
th
is
p
ap
er
ar
e
m
u
ltif
ac
eted
,
r
esid
es
in
ad
d
r
ess
in
g
s
ev
er
al
g
a
p
s
in
th
e
ex
is
tin
g
liter
atu
r
e
o
n
au
t
o
m
ated
v
eh
icle
d
etec
tio
n
s
y
s
tem
s
.
I
n
th
is
p
ap
e
r
,
we
p
r
esen
t
an
e
n
h
an
ce
d
d
r
i
v
in
g
ass
is
tan
ce
s
y
s
tem
th
at
lev
er
ag
es
a
C
NN
to
d
etec
t
v
eh
icles
wit
h
h
ig
h
p
r
ec
is
io
n
an
d
r
ec
all
in
b
o
th
d
ay
an
d
n
i
g
h
t
co
n
d
itio
n
s
.
Un
lik
e
p
r
ev
io
u
s
wo
r
k
s
th
at
o
f
ten
f
o
c
u
s
o
n
eith
er
d
ay
tim
e
o
r
n
ig
h
t
-
tim
e
d
etec
tio
n
,
o
u
r
ap
p
r
o
ac
h
p
r
o
v
id
es
a
co
m
p
r
eh
e
n
s
iv
e
s
o
lu
tio
n
th
at
ex
ce
ls
in
v
ar
y
i
n
g
lig
h
tin
g
co
n
d
itio
n
s
.
Fo
r
clar
ity
o
f
p
r
esen
tatio
n
,
t
h
e
p
ap
e
r
f
o
llo
ws
a
s
tr
u
ctu
r
e
d
o
r
g
an
izatio
n
.
Sectio
n
1
p
r
o
v
id
es
th
e
in
tr
o
d
u
ctio
n
.
Sectio
n
2
d
escr
ib
es
th
e
p
r
o
p
o
s
ed
v
e
h
icle
d
ete
ctio
n
m
eth
o
d
.
Sectio
n
3
p
r
o
v
i
d
es
th
e
r
esu
lts
o
f
th
e
ex
p
er
im
en
ts
,
wh
ile
s
ec
tio
n
4
o
f
f
er
s
th
e
co
n
clu
d
in
g
r
em
ar
k
s
f
o
r
th
e
p
a
p
er
.
2.
M
E
T
H
O
D
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
is
a
p
u
r
e
im
ag
e
p
r
o
ce
s
s
in
g
s
y
s
tem
b
ased
o
n
m
an
y
ex
p
e
r
im
en
ts
an
d
r
esear
ch
es
in
th
is
f
ield
.
I
n
th
is
a
r
ticle,
we
p
r
esen
t o
u
r
en
h
an
ce
d
s
y
s
tem
f
o
r
d
etec
tin
g
v
eh
icles
in
d
ay
-
tim
e
an
d
n
i
g
h
t
-
tim
e.
T
h
e
g
o
al
is
to
o
b
tain
an
ef
f
icien
t
v
eh
icle
d
ete
ctio
n
s
y
s
tem
d
u
r
in
g
d
a
y
an
d
n
ig
h
t
with
h
ig
h
ly
r
eliab
le
r
esu
lts
.
T
h
e
p
r
in
cip
le
o
f
th
is
s
y
s
tem
is
d
escr
ib
ed
in
th
e
o
v
er
all
f
lo
w
d
iag
r
am
p
r
esen
ted
i
n
th
e
Fig
u
r
e
1
.
T
o
d
etec
t
v
eh
icles,
we
n
ee
d
t
o
d
eter
m
in
e
wh
en
th
e
in
p
u
t
im
ag
es
a
r
e
ca
p
tu
r
ed
.
T
h
e
r
ef
o
r
e
,
we
f
ir
s
t
p
er
f
o
r
m
p
r
o
ce
s
s
in
g
o
n
th
e
u
p
p
e
r
p
ar
t
o
f
t
h
e
in
p
u
t
i
m
ag
es,
wh
ich
r
e
p
r
esen
ts
th
e
s
k
y
,
in
o
r
d
e
r
to
d
etec
t
wh
eth
er
th
ey
wer
e
ca
p
t
u
r
ed
d
u
r
in
g
th
e
d
ay
o
r
at
n
ig
h
t,
n
o
te
th
at
th
e
ele
v
atio
n
a
n
g
le
o
f
th
e
ca
m
er
a
e
q
u
als
ze
r
o
.
T
h
e
id
ea
b
eh
in
d
th
is
p
r
o
ce
s
s
in
g
is
to
c
o
m
p
u
te
th
e
av
er
ag
e
v
alu
es
o
f
th
e
in
te
n
s
ities
o
f
th
e
u
p
p
e
r
th
ir
d
o
f
th
e
i
n
p
u
t
im
ag
es
f
o
r
ea
c
h
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
36
,
No
.
3
,
Dec
em
b
er
20
24
:
1
5
32
-
1
5
42
1534
ad
ju
s
tab
le
n
u
m
b
er
o
f
im
ag
es.
W
h
er
e
th
e
in
p
u
t
im
ag
es
ar
e
co
n
s
id
er
ed
as
d
ay
tim
e
s
ce
n
es,
if
th
ei
r
av
e
r
ag
e
is
g
r
ea
ter
th
an
a
p
r
ed
e
f
in
ed
th
r
esh
o
ld
T
1
,
we
p
er
f
o
r
m
th
e
d
a
y
tim
e
d
etec
tio
n
alg
o
r
ith
m
.
C
o
n
v
er
s
ely
,
if
th
eir
av
er
ag
e
is
less
th
an
o
t
h
er
p
r
e
d
ef
in
ed
th
r
esh
o
ld
T
2
,
th
e
y
ar
e
co
n
s
id
er
e
d
as
n
ig
h
t
s
ce
n
es,
s
o
we
p
er
f
o
r
m
t
h
e
n
ig
h
t
d
etec
tio
n
alg
o
r
ith
m
(
wh
er
e
T
1
>T
2
)
.
An
y
o
t
h
er
s
ce
n
ar
io
in
d
icate
s
th
at
th
e
n
atu
r
e
o
f
th
e
im
ag
es
r
em
ai
n
s
u
n
ch
an
g
ed
.
Fig
u
r
e
1
.
Ov
e
r
all
f
lo
w
d
iag
r
a
m
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
2
.
1
.
Da
y
-
t
i
m
e
v
ehicle
det
ec
t
i
o
n m
et
ho
d
T
h
e
f
ir
s
t
p
ar
t
o
f
th
is
s
ec
tio
n
f
o
cu
s
es
o
n
v
eh
icle
d
etec
tio
n
d
u
r
in
g
th
e
d
ay
,
h
ig
h
lig
h
tin
g
th
e
p
r
o
ce
s
s
's
two
m
ain
s
tag
es:
h
y
p
o
th
esis
g
en
er
a
tio
n
an
d
h
y
p
o
th
esis
v
er
if
icatio
n
.
T
h
is
two
-
s
tep
ap
p
r
o
a
ch
en
s
u
r
es
th
at
th
e
v
eh
icle
d
etec
tio
n
s
y
s
tem
is
b
o
th
ef
f
icien
t
a
n
d
r
eliab
le,
p
r
o
v
id
in
g
ac
c
u
r
ate
r
esu
lts
u
n
d
e
r
d
ay
lig
h
t
co
n
d
itio
n
s
.
B
y
b
r
ea
k
in
g
d
o
wn
t
h
e
task
in
to
th
ese
s
tag
es,
th
e
s
y
s
tem
ca
n
m
o
r
e
e
f
f
ec
tiv
ely
m
a
n
ag
e
th
e
co
m
p
lex
ities
o
f
d
etec
tin
g
v
eh
icles in
v
ar
ied
en
v
ir
o
n
m
e
n
ts
.
2
.
1
.
1
.
H
y
po
t
hes
is
g
ener
a
t
io
n
s
t
a
g
e
T
h
e
p
e
r
f
o
r
m
an
ce
o
f
th
is
s
tag
e
is
th
e
k
ey
to
ac
h
iev
e
th
e
e
f
f
icien
cy
o
f
t
h
e
e
n
tire
s
y
s
tem
,
wh
er
e
th
e
g
o
o
d
g
en
er
atio
n
o
f
h
y
p
o
th
esis
h
as
a
d
ir
ec
t
im
p
ac
t
o
n
th
e
r
e
s
u
lt
o
f
th
e
v
e
r
if
icatio
n
s
tag
e.
Fo
r
th
is
r
ea
s
o
n
,
w
e
h
av
e
p
er
f
o
r
m
e
d
s
ev
er
al
alg
o
r
i
th
m
s
in
o
r
d
er
to
co
r
r
ec
tly
g
en
er
ate
all
th
e
h
y
p
o
t
h
eses
.
I
n
th
i
s
s
tag
e
we
b
ased
o
n
p
atter
n
d
etec
tio
n
i
n
o
r
d
er
to
l
o
ca
te
th
e
p
o
te
n
tial
v
eh
icles
(
h
y
p
o
th
eses
)
.
I
n
itially
,
p
r
ep
r
o
ce
s
s
in
g
is
co
n
d
u
cted
o
n
in
p
u
t
im
ag
es
b
y
co
n
v
er
tin
g
th
em
f
r
o
m
r
ed
,
g
r
ee
n
an
d
b
lu
e
(
R
GB
)
s
p
ac
e
to
h
u
e
s
atu
r
atio
n
v
alu
e
(
HSV
)
s
p
ac
e
as
s
h
o
wn
in
Fig
u
r
e
2
to
s
eg
r
eg
ate
lu
m
in
o
s
ity
in
f
o
r
m
atio
n
f
r
o
m
c
o
lo
r
in
f
o
r
m
a
tio
n
.
Su
b
s
eq
u
en
tl
y
,
h
is
to
g
r
am
e
q
u
a
lizatio
n
is
ap
p
l
ied
to
th
e
lu
m
in
o
s
ity
in
f
o
r
m
at
io
n
(
V
ch
a
n
n
el)
to
e
n
h
an
c
e
c
o
n
tr
ast,
as
d
ep
icted
in
Fig
u
r
e
3
.
Fig
u
r
e
4
p
r
esen
ts
th
e
r
esu
ltin
g
im
ag
e
af
ter
h
is
t
o
g
r
am
eq
u
aliza
tio
n
ju
x
tap
o
s
e
d
with
th
e
o
r
ig
in
al
im
ag
e.
As
ev
i
d
en
t
f
r
o
m
b
o
th
Fig
u
r
es
3
an
d
4
,
a
s
u
b
s
t
an
ti
al
en
h
an
ce
m
en
t
i
n
c
o
n
tr
ast
is
o
b
s
er
v
e
d
,
wh
ich
is
ex
p
ec
ted
to
f
ac
ilit
ate
s
u
b
s
eq
u
e
n
t p
r
o
ce
s
s
es.
Fig
u
r
e
2
.
R
GB
to
HSV
co
n
v
er
s
io
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
n
h
a
n
ce
d
d
r
ivin
g
a
s
s
is
ta
n
ce
:
a
u
to
ma
ted
d
a
y
a
n
d
n
ig
h
t v
e
h
i
cle
d
etec
tio
n
s
ystem
… (
A
b
d
el
mo
g
h
it Za
a
r
a
n
e
)
1535
Fig
u
r
e
3
.
His
to
g
r
am
eq
u
aliza
ti
o
n
Fig
u
r
e
4
.
I
m
ag
e
r
esu
lted
af
ter
ap
p
ly
in
g
h
is
to
g
r
am
e
q
u
aliza
tio
n
Fo
llo
win
g
th
is
,
k
ey
f
ea
tu
r
es,
s
p
ec
if
ically
ed
g
es,
ar
e
ex
tr
ac
te
d
u
s
in
g
th
e
C
an
n
y
ed
g
e
d
etec
t
o
r
[1
6
]
to
ex
p
ed
ite
p
r
o
ce
s
s
in
g
tim
e
an
d
en
h
an
ce
p
atter
n
d
etec
tio
n
ac
cu
r
ac
y
.
Su
b
s
eq
u
e
n
t
to
ed
g
e
ex
tr
ac
tio
n
,
t
h
e
tech
n
iq
u
e
o
f
f
ast
n
o
r
m
alize
d
c
r
o
s
s
-
co
r
r
elatio
n
is
em
p
lo
y
ed
t
o
id
en
tif
y
v
eh
icle
p
atter
n
s
wit
h
in
th
e
im
ag
e.
T
h
is
tech
n
iq
u
e
in
v
o
lv
es
ap
p
ly
in
g
f
ast
n
o
r
m
alize
d
cr
o
s
s
-
co
r
r
elatio
n
to
th
e
f
ilter
ed
in
p
u
t
im
ag
e
an
d
m
o
d
el
im
ag
es,
af
ter
s
u
b
jectin
g
th
em
to
th
e
C
an
n
y
ed
g
e
d
etec
to
r
.
No
r
m
a
lized
cr
o
s
s
-
co
r
r
elatio
n
[1
7
]
is
a
wid
ely
u
tili
ze
d
m
eth
o
d
f
o
r
q
u
an
tify
in
g
th
e
r
es
em
b
lan
ce
b
etwe
en
two
im
ag
es
,
ex
h
i
b
itin
g
n
o
tab
le
ef
f
icac
y
in
p
atter
n
d
etec
tio
n
,
p
ar
ticu
lar
ly
wh
en
c
o
m
b
in
e
d
with
th
e
C
an
n
y
ed
g
e
d
etec
to
r
.
C
o
n
s
eq
u
en
tly
,
p
o
ten
t
ial
v
eh
i
cles
with
in
th
e
in
p
u
t
im
ag
es
ar
e
id
e
n
tifie
d
b
ased
o
n
th
e
d
e
g
r
ee
o
f
s
im
ilar
ity
b
etwe
en
m
o
d
el
im
ag
es
an
d
test
im
a
g
es.
An
illu
s
tr
ativ
e
ex
am
p
le
is
d
ep
icted
in
Fig
u
r
e
5
.
T
h
e
b
est
m
atch
o
cc
u
r
s
wh
en
th
e
m
o
d
el
an
d
test
im
ag
es
h
av
e
a
m
ax
im
u
m
co
r
r
elatio
n
v
al
u
e.
E
m
p
lo
y
in
g
t
h
is
tech
n
iq
u
e
en
a
b
les th
e
d
is
co
v
er
y
o
f
m
u
ltip
le
p
o
ten
tial v
e
h
icle
lo
ca
tio
n
s
.
F
ig
u
r
e
5
.
I
m
ag
e
m
o
d
el
ex
a
m
p
les
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
36
,
No
.
3
,
Dec
em
b
er
20
24
:
1
5
32
-
1
5
42
1536
2
.
1
.
2
.
H
y
po
t
hes
is
v
er
if
ica
t
io
n sta
g
e
Hy
p
o
th
esis
v
er
if
icatio
n
s
tag
e
is
n
o
t
less
im
p
o
r
tan
t
th
an
t
h
e
f
ir
s
t
s
tag
e
wh
ich
is
r
esp
o
n
s
i
b
le
f
o
r
t
h
e
class
if
icatio
n
o
f
th
e
h
y
p
o
th
es
es
g
en
er
ated
in
t
h
e
p
r
e
v
io
u
s
s
tag
e.
T
h
is
s
tag
e
is
co
m
p
o
s
ed
o
f
two
p
r
in
cip
al
p
ar
ts
,
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
c
lass
if
icatio
n
.
I
n
th
e
f
ir
s
t
p
ar
t,
we
ex
tr
ac
t
th
e
i
m
p
o
r
ta
n
t
ch
ar
ac
ter
is
tics
u
s
in
g
th
e
th
ir
d
lev
el
o
f
th
e
two
-
d
im
en
s
io
n
al
d
is
cr
ete
wav
elet
tr
an
s
f
o
r
m
atio
n
o
n
th
e
g
en
e
r
ated
ca
n
d
id
ates,
wh
er
e
we
h
av
e
u
s
ed
Haa
r
wav
elet
tr
a
n
s
f
o
r
m
.
T
h
is
tr
an
s
f
o
r
m
atio
n
is
a
v
er
y
im
p
o
r
tan
t
an
d
p
o
wer
f
u
l
tech
n
iq
u
e
to
r
ep
r
esen
t
in
p
u
t
d
ata
at
d
i
f
f
er
e
n
t
s
ca
les
an
d
f
r
eq
u
en
cies.
T
h
e
2D
-
DW
T
[1
8
]
as
a
m
ath
em
at
ical
to
o
l
is
u
tili
ze
d
to
d
ec
o
m
p
o
s
e
a
n
im
a
g
e
i
n
to
f
o
u
r
elem
e
n
ts
an
d
ex
tr
ac
t
t
h
e
ch
a
r
ac
ter
is
tics
o
f
th
e
im
a
g
e
u
s
in
g
th
e
m
o
s
t
s
ig
n
if
ican
t
elem
en
t
o
f
th
ese
f
o
u
r
elem
en
ts
.
I
t
s
p
lits
th
e
im
a
g
e
to
f
o
u
r
p
ar
ts
,
th
r
ee
p
a
r
ts
with
h
ig
h
f
r
eq
u
e
n
cies
an
d
o
n
e
p
a
r
t
with
lo
w
f
r
eq
u
en
cy
.
I
n
th
is
p
ar
t,
we
r
elied
o
n
th
e
lo
w
f
r
eq
u
e
n
cy
s
u
b
-
i
m
ag
es
to
f
ee
d
th
e
class
if
icatio
n
p
ar
t.
Fo
r
th
e
cla
s
s
if
icatio
n
p
ar
t,
we
b
ased
o
n
t
h
e
C
NN
t
o
v
er
if
y
th
e
g
en
er
ate
d
h
y
p
o
th
eses
b
ased
o
n
th
eir
e
x
tr
ac
ted
f
ea
t
u
r
es f
o
r
class
if
y
in
g
th
em
in
to
tr
u
e
o
r
f
alse h
y
p
o
th
eses
(
v
eh
icles o
r
n
o
n
-
v
e
h
icles).
C
NN
s
[
19
]
f
o
r
class
if
icatio
n
ar
e
th
e
cu
r
r
en
tly
b
est
p
er
f
o
r
m
in
g
m
o
d
els
an
d
co
n
s
is
t
o
f
tw
o
d
if
f
er
e
n
t
p
ar
ts
.
T
h
e
in
p
u
t
is
an
im
a
g
e
g
iv
en
as
a
m
atr
ix
o
f
p
ix
els.
T
h
is
m
atr
ix
co
n
tain
s
two
elem
en
ts
f
o
r
a
g
r
ay
s
ca
le
im
ag
e.
T
h
e
th
ir
d
elem
en
t
is
t
h
e
co
lo
r
.
T
h
e
in
itial
p
ar
t
o
f
a
C
NN
is
th
e
co
n
v
o
lu
tio
n
al
p
ar
t.
I
t
wo
r
k
s
a
f
ea
tu
r
e
ex
tr
ac
tio
n
s
y
s
tem
o
f
im
ag
es.
T
h
e
r
esu
lt
o
f
th
e
co
n
v
o
lu
tio
n
a
l
p
ar
t
is
f
ed
in
to
th
e
s
ec
o
n
d
p
a
r
t,
wh
ich
co
n
s
is
ts
o
f
f
u
lly
co
n
n
ec
ted
lay
er
s
.
T
h
is
p
ar
t’
s
task
co
n
s
is
ts
o
f
co
n
n
ec
tin
g
th
e
f
ea
tu
r
es
o
b
tain
ed
p
r
ev
i
o
u
s
ly
in
o
r
d
er
f
o
r
class
if
y
in
g
th
e
im
ag
es.
T
h
e
r
esu
lt
is
a
la
s
t
lay
er
with
o
n
ly
a
n
eu
r
o
n
f
o
r
ea
ch
class
.
T
h
e
o
b
tain
ed
r
esu
lts
ar
e
ty
p
ically
n
o
r
m
alize
d
to
0
a
n
d
1
,
with
th
e
s
u
m
o
f
th
ese
v
al
u
e
eq
u
al
to
1
,
in
o
r
d
er
to
g
en
er
ate
a
p
r
o
b
ab
ilit
y
d
is
tr
ib
u
tio
n
o
v
er
th
e
class
es.
B
u
ild
in
g
a
n
o
v
el
C
NN
is
v
er
y
in
ten
s
iv
e
in
ter
m
s
o
f
k
n
o
wl
ed
g
e,
h
ar
d
war
e,
an
d
v
o
lu
m
e
o
f
lab
eled
d
atasets
.
Fo
r
th
is
r
ea
s
o
n
,
th
er
e
ar
e
s
p
ec
ialis
t
r
esear
ch
er
s
w
h
o
f
o
c
u
s
o
n
en
h
a
n
cin
g
C
NNs.
T
h
ey
m
ak
e
th
ei
r
tech
n
o
lo
g
ical
im
p
r
o
v
e
m
en
ts
p
u
b
lic,
to
g
eth
er
with
in
f
o
r
m
a
tio
n
ab
o
u
t
th
e
tr
ain
ed
n
etwo
r
k
s
in
th
e
d
atab
ase,
wh
ich
ca
n
b
e
e
x
p
lo
ited
f
o
r
tr
a
n
s
f
er
tr
ain
in
g
.
T
h
er
e
a
r
e
s
ev
er
al
p
o
wer
f
u
l m
o
d
els th
at
co
u
l
d
b
e
u
s
ed
f
o
r
tr
an
s
f
er
lear
n
in
g
s
u
c
h
as
th
e
m
o
d
el
u
s
ed
in
th
is
wo
r
k
,
I
n
ce
p
tio
n
v
4
[
20
]
.
T
h
is
m
o
d
el
tr
ain
e
d
b
y
g
o
o
g
le
u
s
in
g
1
.
2
m
illi
o
n
im
ag
es
class
if
ied
in
1
,
0
0
0
ca
teg
o
r
ies.
W
e
ch
o
s
e
to
u
s
e
th
is
m
o
d
el
b
ec
au
s
e
o
f
its
lo
w
co
n
s
u
m
p
tio
n
o
f
tim
e
an
d
r
eso
u
r
ce
s
,
an
d
also
b
ec
au
s
e
o
f
its
h
ig
h
ac
c
u
r
ac
y
a
n
d
h
ig
h
p
er
f
o
r
m
an
ce
o
f
class
if
icatio
n
.
2
.
2
.
Nig
ht
-
t
im
e
v
ehicle
det
ec
t
io
n m
et
ho
d
T
h
e
s
ec
o
n
d
p
a
r
t
o
f
th
is
s
ec
tio
n
f
o
cu
s
es
o
n
v
eh
icle
d
etec
ti
o
n
at
n
ig
h
t,
wh
ich
is
d
o
n
e
in
two
m
ain
s
tag
es:
lam
p
ca
n
d
id
ate
g
en
er
atio
n
an
d
lam
p
ca
n
d
id
ate
v
e
r
if
icatio
n
.
First,
th
e
s
y
s
tem
i
d
en
tifie
s
p
o
s
s
ib
le
v
eh
icle
lig
h
ts
i
n
th
e
im
ag
e
b
y
d
etec
tin
g
b
r
ig
h
t
s
p
o
ts
.
Nex
t,
it
v
er
if
ies
t
h
ese
p
o
te
n
tial
lig
h
ts
to
co
n
f
ir
m
th
ey
ar
e
ac
tu
al
v
eh
icle
lam
p
s
.
T
h
is
p
r
o
ce
s
s
,
s
h
o
wn
in
Fig
u
r
e
6
,
h
el
p
s
im
p
r
o
v
e
th
e
ac
cu
r
ac
y
o
f
d
etec
tin
g
v
eh
icles
in
lo
w
-
lig
h
t c
o
n
d
itio
n
s
.
Fig
u
r
e
6
.
Gen
e
r
al
f
lo
w
ch
a
r
t o
f
th
e
v
eh
icle
d
etec
tio
n
s
y
s
tem
at
n
ig
h
t
2
.
2
.
1
.
L
a
m
p c
a
nd
ida
t
e
g
ener
a
t
io
n sta
g
e
T
o
g
en
e
r
ate
ca
n
d
i
d
ate
lam
p
s
,
two
ess
en
tial
tech
n
iq
u
es
ar
e
em
p
lo
y
ed
:
th
e
c
o
lo
r
t
h
r
esh
o
ld
in
g
tech
n
iq
u
e
an
d
th
e
c
o
n
n
ec
te
d
co
m
p
o
n
en
ts
tech
n
iq
u
e.
I
n
itially
,
s
ig
n
if
ican
t
f
ea
tu
r
es,
n
am
ely
co
lo
r
s
,
ar
e
ex
tr
ac
ted
u
s
in
g
a
c
o
lo
r
th
r
esh
o
ld
in
g
m
eth
o
d
.
T
h
is
m
et
h
o
d
in
v
o
lv
es
th
r
esh
o
ld
in
g
th
e
i
n
p
u
t
im
ag
es
b
ased
o
n
th
e
co
lo
r
in
ten
s
ities
o
f
im
a
g
e
p
ix
els
u
s
in
g
p
r
ed
ef
i
n
ed
th
r
esh
o
ld
s
.
Sp
ec
if
ically
,
b
r
ig
h
t
wh
ite
r
e
g
io
n
s
r
ep
r
esen
tin
g
p
o
ten
tial
h
ea
d
la
m
p
s
an
d
b
r
ig
h
t
r
ed
r
eg
io
n
s
r
ep
r
esen
tin
g
p
o
ten
tial
r
ea
r
la
m
p
s
ar
e
p
r
eser
v
ed
,
wh
ile
all
o
th
er
r
eg
io
n
s
ar
e
d
is
r
eg
ar
d
ed
.
T
h
e
p
r
o
ce
s
s
b
eg
in
s
b
y
co
n
v
er
tin
g
th
e
in
p
u
t
im
ag
es
f
r
o
m
th
e
R
GB
co
lo
r
s
p
ac
e
to
th
e
HSV
co
lo
r
s
p
ac
e,
wh
ich
is
o
p
tim
al
f
o
r
co
l
o
r
th
r
esh
o
ld
in
g
.
Su
b
s
eq
u
e
n
tly
,
co
lo
r
th
r
esh
o
ld
in
g
is
ap
p
lied
p
ix
el
b
y
p
ix
el
in
th
e
HSV
co
lo
r
s
p
ac
e
to
r
etain
o
n
ly
th
e
r
elev
an
t
r
eg
io
n
s
.
Pix
e
ls
ar
e
class
if
ied
as
p
o
ten
tial
p
ar
ts
o
f
r
ed
r
eg
io
n
s
i
f
th
e
h
u
e
(
H)
f
alls
with
in
th
e
r
an
g
e
o
f
3
4
5
to
1
0
,
s
atu
r
atio
n
(
S)
f
alls
with
in
5
5
to
1
0
0
,
a
n
d
v
alu
e
(
V)
f
alls
with
in
5
0
to
1
0
0
.
C
o
n
v
er
s
ely
,
p
ix
els
ar
e
id
en
tifie
d
as
p
o
te
n
tial
p
ar
ts
o
f
wh
ite
r
eg
io
n
s
if
S
f
alls
with
in
0
to
2
0
an
d
V
f
alls
with
in
9
0
to
1
0
0
,
r
eg
a
r
d
less
o
f
th
e
H
v
alu
e
.
So
m
e
r
eg
io
n
s
m
a
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
n
h
a
n
ce
d
d
r
ivin
g
a
s
s
is
ta
n
ce
:
a
u
to
ma
ted
d
a
y
a
n
d
n
ig
h
t v
e
h
i
cle
d
etec
tio
n
s
ystem
… (
A
b
d
el
mo
g
h
it Za
a
r
a
n
e
)
1537
co
n
tain
g
ap
s
th
at
c
o
u
ld
f
r
a
g
m
en
t
th
em
in
to
m
u
ltip
le
r
e
g
io
n
s
.
T
o
ad
d
r
ess
th
is
is
s
u
e,
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
ar
e
p
er
f
o
r
m
ed
to
f
ill
an
d
r
est
o
r
e
m
is
s
in
g
p
ix
els
with
in
th
ese
g
ap
s
.
Sp
ec
if
ically
,
d
ilatio
n
with
a
3
x
3
k
er
n
el
f
o
llo
wed
b
y
er
o
s
io
n
with
a
3
×
3
k
er
n
el
is
em
p
lo
y
ed
.
Su
b
s
eq
u
en
tly
,
ea
c
h
r
e
m
ain
in
g
r
eg
io
n
is
ex
tr
ac
ted
as
a
co
m
p
o
n
en
t
(
ca
n
d
i
d
ate)
u
s
in
g
th
e
co
n
n
ec
te
d
co
m
p
o
n
en
ts
tech
n
iq
u
e
[2
1
]
.
T
h
is
tech
n
iq
u
e
in
v
o
lv
es
s
ca
n
n
i
n
g
th
e
b
in
ar
y
im
ag
e
co
n
tain
in
g
th
e
r
eg
io
n
s
p
ix
el
b
y
p
ix
el
to
g
r
o
u
p
all
ad
jace
n
t
p
ix
els
o
f
ea
ch
b
r
ig
h
t
r
eg
i
o
n
in
to
o
n
e
co
n
n
ec
ted
co
m
p
o
n
e
n
t
an
d
p
r
o
v
id
e
its
in
f
o
r
m
atio
n
,
in
clu
d
in
g
s
ize
an
d
co
o
r
d
in
ates.
Fin
ally
,
th
e
ex
t
r
ac
ted
co
m
p
o
n
e
n
ts
u
n
d
e
r
g
o
f
il
tr
atio
n
to
elim
in
ate
ex
ce
s
s
iv
ely
lar
g
e
o
r
s
m
all
co
m
p
o
n
e
n
ts
th
at
ar
e
u
n
lik
el
y
to
r
ep
r
esen
t h
ea
d
lam
p
s
o
r
r
ea
r
la
m
p
s
.
2
.
2
.
2
.
L
a
m
p c
a
nd
id
a
t
e
v
er
if
i
ca
t
io
n sta
g
e
I
n
o
r
d
er
to
v
er
if
y
th
e
g
e
n
er
a
ted
ca
n
d
i
d
ates,
we
f
i
r
s
t
g
r
o
u
p
th
em
in
to
s
ets
o
f
p
air
s
ac
co
r
d
in
g
to
s
p
ec
if
ic
cr
iter
ia,
wh
er
e
p
o
ten
ti
al
lam
p
s
,
wh
ich
h
av
e
th
e
s
am
e
co
lo
r
an
d
a
r
e
r
elativ
ely
clo
s
e
to
ea
ch
o
th
er
an
d
alig
n
ed
h
o
r
izo
n
tally
,
ar
e
g
ath
er
ed
in
th
e
s
am
e
s
et.
Ve
h
icle
lam
p
s
h
av
e
d
if
f
er
en
t
s
h
a
p
es
th
at
v
ar
y
f
r
o
m
o
n
e
v
eh
icle
to
a
n
o
th
er
,
m
a
k
in
g
t
h
em
q
u
ite
d
if
f
icu
lt
t
o
class
if
y
.
Ho
wev
er
,
th
e
p
ai
r
o
f
lam
p
s
in
ea
ch
v
eh
icle
is
s
y
m
m
etr
ically
id
en
tical.
T
h
er
ef
o
r
e,
t
o
p
r
e
-
class
if
y
ca
n
d
i
d
ate
lam
p
s
as
p
o
te
n
tial
v
eh
icles,
we
p
air
t
h
e
m
o
s
t
s
y
m
m
etr
ically
id
en
tical
lam
p
ca
n
d
id
ates
b
y
p
er
f
o
r
m
in
g
a
n
o
r
m
alize
d
cr
o
s
s
-
co
r
r
elatio
n
tech
n
iq
u
e
b
etwe
en
a
m
ir
r
o
r
ed
lam
p
ca
n
d
id
ate
an
d
th
e
o
th
e
r
lam
p
s
ca
n
d
id
ates
in
th
e
s
am
e
s
et.
Pair
s
th
at
ar
e
h
i
g
h
ly
co
r
r
elate
d
ar
e
c
o
n
s
id
er
ed
as p
o
ten
tial v
eh
icl
e.
Po
ten
tial
v
eh
icles
ar
e
class
if
ied
as
eith
er
v
e
h
icles
o
r
n
o
n
-
v
eh
icles
u
s
in
g
th
e
C
NN
class
if
ier
.
T
h
e
C
NN
is
r
etr
ain
ed
b
y
a
d
ataset
th
at
co
n
tain
s
p
o
s
itiv
e
s
am
p
les
as
well
as
n
eg
ativ
e
s
am
p
les.
T
h
e
p
o
s
itiv
e
s
am
p
les
co
n
tain
im
ag
es
o
f
v
eh
icles
tak
en
f
r
o
m
d
if
f
er
e
n
t
v
id
eo
s
in
d
if
f
e
r
en
t
lo
ca
tio
n
s
,
an
d
th
e
n
e
g
ativ
e
s
am
p
les co
n
tain
d
if
f
er
e
n
t im
a
g
es o
f
v
ar
i
o
u
s
elem
en
ts
o
r
a
r
e
as o
f
th
e
n
atu
r
al
s
ce
n
e.
2
.
3
.
Ret
r
a
ini
ng
CNN
m
o
del
2
.
3
.
1
.
Da
t
a
co
llect
io
n a
nd
prepro
ce
s
s
ing
T
h
e
d
ataset
u
s
ed
f
o
r
r
etr
ain
i
n
g
th
e
u
s
ed
C
NN
m
o
d
el
is
a
cu
s
to
m
-
b
u
ilt
v
e
h
icle
d
etec
tio
n
d
ataset
th
at
in
clu
d
es
d
iv
er
s
e
d
a
y
an
d
n
ig
h
t
co
n
d
itio
n
s
.
T
h
e
d
ataset
was
co
m
p
iled
f
r
o
m
s
ev
er
al
p
u
b
licly
av
ailab
le
d
atasets
,
en
s
u
r
in
g
a
c
o
m
p
r
e
h
en
s
iv
e
co
llectio
n
o
f
v
e
h
icle
im
ag
es.
Sp
ec
if
ically
,
it
in
clu
d
es
im
ag
e
s
f
r
o
m
th
e
C
altec
h
C
ar
s
d
ataset
[
2
2
]
,
th
e
AOL
P d
ataset
[
2
3
]
,
th
e
KI
T
T
I
Vis
io
n
B
en
ch
m
ar
k
Su
ite
[
2
4
]
,
an
d
th
e
n
ig
h
t
-
tim
e
v
eh
icle
d
etec
tio
n
d
ataset
[
2
5
]
.
I
n
to
tal
,
th
e
d
ataset
co
m
p
r
is
es
1
0
,
0
0
0
im
ag
es,
with
eq
u
al
d
is
tr
ib
u
tio
n
b
etwe
en
d
ay
tim
e
an
d
n
ig
h
t
-
tim
e
s
ce
n
es,
th
at
en
s
u
r
es a
r
o
b
u
s
t m
o
d
el
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
d
if
f
er
e
n
t lig
h
tin
g
s
ce
n
ar
io
s
.
Data
p
r
ep
r
o
ce
s
s
in
g
was
cr
iti
ca
l
to
en
s
u
r
e
h
i
g
h
-
q
u
ality
in
p
u
ts
f
o
r
th
e
I
n
ce
p
tio
n
V
4
m
o
d
el.
T
h
e
f
o
llo
win
g
s
tep
s
wer
e
u
n
d
er
tak
en
:
−
I
m
ag
e
r
esizin
g
:
all
im
ag
es
w
er
e
r
esized
to
2
9
9
×
2
9
9
p
i
x
els
to
m
atch
th
e
i
n
p
u
t
d
im
en
s
io
n
s
r
eq
u
ir
e
d
b
y
th
e
I
n
ce
p
tio
n
V4
m
o
d
el.
−
No
r
m
aliza
tio
n
:
p
ix
e
l
v
alu
es
wer
e
n
o
r
m
alize
d
to
th
e
r
a
n
g
e
[
0
,
1
]
to
f
ac
ilit
ate
f
aster
co
n
v
er
g
en
ce
d
u
r
in
g
tr
ain
in
g
.
−
Data
au
g
m
en
tatio
n
:
tech
n
iq
u
e
s
s
u
ch
as
h
o
r
izo
n
tal
f
lip
p
i
n
g
,
r
o
tatio
n
,
an
d
b
r
i
g
h
tn
ess
ad
ju
s
tm
en
t
wer
e
ap
p
lied
to
a
u
g
m
en
t t
h
e
d
ataset
an
d
im
p
r
o
v
e
m
o
d
el
r
o
b
u
s
tn
ess
.
2
.
3
.
2
.
I
ncept
io
n v
4
a
rc
hite
ct
ure
W
e
em
p
lo
y
ed
th
e
I
n
ce
p
tio
n
V
4
m
o
d
el,
u
tili
zin
g
tr
an
s
f
er
lear
n
in
g
to
le
v
er
ag
e
p
r
e
-
tr
ain
e
d
w
eig
h
ts
o
n
th
e
I
m
ag
eNe
t
d
ataset.
T
h
e
I
n
ce
p
tio
n
V4
ar
c
h
itectu
r
e
is
k
n
o
wn
f
o
r
its
ef
f
icien
c
y
an
d
h
i
g
h
p
er
f
o
r
m
a
n
ce
in
im
ag
e
class
if
icatio
n
task
s
.
T
h
e
m
o
d
el
was f
in
e
-
tu
n
ed
o
n
o
u
r
cu
s
to
m
v
eh
icle
d
etec
tio
n
d
ataset.
−
I
n
p
u
t
lay
e
r
: a
cc
ep
ts
2
9
9
×
2
9
9
R
GB
im
ag
e
s.
−
Pre
-
tr
ain
ed
la
y
er
s
:
all
lay
er
s
u
p
to
th
e
f
in
al
in
ce
p
tio
n
m
o
d
u
l
e
wer
e
f
r
o
ze
n
to
r
etain
th
e
lea
r
n
ed
f
ea
tu
r
es
f
r
o
m
I
m
ag
eNe
t.
−
Fu
lly
co
n
n
ec
te
d
lay
er
s
:
ad
d
ed
cu
s
to
m
f
u
lly
co
n
n
ec
ted
lay
er
s
with
1
,
0
2
4
an
d
5
1
2
n
eu
r
o
n
s
r
esp
ec
tiv
ely
,
f
o
llo
wed
b
y
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U
)
ac
tiv
atio
n
.
−
Ou
tp
u
t
lay
er
: a
s
o
f
tm
a
x
lay
er
with
two
n
eu
r
o
n
s
f
o
r
b
in
ar
y
cl
ass
if
icatio
n
(
v
eh
icle
p
r
esen
t o
r
n
o
t)
.
T
h
is
m
o
d
el
was
tr
ain
ed
u
s
in
g
th
e
Ad
am
o
p
tim
izer
with
a
le
ar
n
in
g
r
ate
o
f
0
.
0
0
0
1
.
T
h
e
l
o
s
s
f
u
n
ctio
n
u
s
ed
was
b
in
ar
y
cr
o
s
s
-
en
tr
o
p
y
,
an
d
t
h
e
m
o
d
el
was
tr
ain
ed
f
o
r
3
0
ep
o
c
h
s
with
a
b
atch
s
i
ze
o
f
3
2
.
T
o
r
etr
ain
th
is
m
o
d
el,
we
s
p
lit
th
e
d
atase
t
in
to
tr
ain
in
g
(
7
0
%),
v
alid
ati
o
n
(
1
5
%),
an
d
test
(
1
5
%)
s
ets
.
T
r
ain
in
g
in
v
o
l
v
ed
iter
ativ
ely
u
p
d
atin
g
th
e
m
o
d
el
weig
h
ts
u
s
in
g
b
ac
k
p
r
o
p
a
g
atio
n
.
E
ar
ly
s
to
p
p
i
n
g
was
em
p
lo
y
ed
to
p
r
e
v
en
t
o
v
er
f
itti
n
g
,
b
ased
o
n
th
e
v
alid
atio
n
lo
s
s
.
T
h
e
ch
o
ice
o
f
I
n
ce
p
tio
n
V
4
was
m
o
tiv
ated
b
y
its
p
r
o
v
en
ef
f
icac
y
an
d
ef
f
icien
c
y
in
im
ag
e
class
if
icatio
n
an
d
o
b
ject
d
etec
tio
n
task
s
.
T
r
an
s
f
er
lear
n
in
g
was
em
p
lo
y
ed
to
lev
er
ag
e
p
r
e
-
tr
ain
ed
weig
h
ts
,
s
ig
n
if
ican
tly
r
ed
u
cin
g
tr
ain
in
g
tim
e
an
d
en
h
an
cin
g
p
e
r
f
o
r
m
a
n
ce
.
Data
au
g
m
en
tatio
n
was
u
s
ed
to
ad
d
r
ess
th
e
v
ar
iab
ilit
y
in
r
ea
l
-
wo
r
ld
d
r
iv
in
g
co
n
d
itio
n
s
,
th
er
eb
y
en
h
an
cin
g
th
e
m
o
d
el'
s
g
en
er
aliza
tio
n
ca
p
ab
ilit
ies.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
36
,
No
.
3
,
Dec
em
b
er
20
24
:
1
5
32
-
1
5
42
1538
3.
E
XP
E
R
I
M
E
N
T
R
E
SU
L
T
S
3
.
1
.
P
er
f
o
r
m
a
nce
m
e
t
rics
I
n
o
r
d
er
to
e
v
alu
ate
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
,
we
g
ath
e
r
ed
s
ix
v
id
eo
s
eq
u
en
ce
s
o
f
r
ea
l
tr
af
f
ic
s
ce
n
es
in
d
if
f
er
en
t
en
v
i
r
o
n
m
e
n
ts
with
a
ca
m
er
a
m
o
u
n
ted
o
n
t
h
e
r
ea
r
-
v
iew
m
ir
r
o
r
o
f
th
e
h
o
s
tin
g
ca
r
.
T
h
r
ee
v
id
eo
s
wer
e
co
llected
in
d
a
y
tim
e
an
d
th
e
r
em
ain
in
g
th
r
ee
in
n
i
g
h
t
-
tim
e.
T
h
e
e
x
p
er
im
e
n
ts
wer
e
p
er
f
o
r
m
ed
u
s
in
g
C
++
,
Op
en
C
V
an
d
T
en
s
o
r
Flo
w
o
n
a
1
.
2
GHz
d
u
al
-
c
o
r
e
h
a
r
d
war
e
p
r
o
ce
s
s
o
r
s
y
s
tem
AR
M
C
o
r
tex
-
A9
(
HPS)
o
p
er
atin
g
u
n
d
er
an
L
XDE
d
e
s
k
to
p
with
1
.
0
GB
DDR3
m
em
o
r
y
.
T
h
is
h
a
r
d
war
e
p
r
o
ce
s
s
o
r
s
y
s
tem
is
h
o
u
s
ed
with
in
a
VE
E
K
-
MT
2
S
wh
ic
h
in
clu
d
es
th
e
MT
L
C
2
m
o
d
u
le
an
d
th
e
DE
1
0
s
tan
d
a
r
d
FP
GA
s
u
p
p
lied
b
y
T
E
R
ASI
C
.
I
n
o
r
d
er
t
o
ev
al
u
ate
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
,
we
u
s
ed
p
r
ec
is
io
n
a
n
d
r
ec
all
as
th
e
p
r
im
ar
y
m
etr
ics.
Pre
cisi
o
n
m
ea
s
u
r
es
th
e
r
atio
o
f
tr
u
e
p
o
s
itiv
e
d
etec
tio
n
s
t
o
th
e
to
tal
d
ete
ctio
n
s
,
wh
ile
r
ec
all
m
ea
s
u
r
es
th
e
r
atio
o
f
tr
u
e
p
o
s
itiv
e
d
etec
tio
n
s
to
th
e
to
tal
ac
t
u
al
p
o
s
itiv
es.
T
he
se
m
ea
s
u
r
es
wer
e
co
m
p
u
ted
as
s
h
o
wn
(
1
)
an
d
(
2
)
:
=
+
(
1
)
=
+
(
2
)
w
ith
FNs
is
f
al
s
e
n
eg
ativ
es,
FP
s
is
f
alse p
o
s
itiv
e
s
an
d
T
Ps
is
tr
u
e
p
o
s
itiv
es.
T
h
ese
m
etr
ics
wer
e
ca
lcu
lated
f
o
r
b
o
th
d
ay
tim
e
an
d
n
i
g
h
t
-
tim
e
d
atasets
,
p
r
o
v
i
d
in
g
a
co
m
p
r
e
h
en
s
iv
e
ev
alu
atio
n
o
f
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
.
3
.
2
.
E
v
a
lua
t
i
o
n r
esu
lt
s
Ou
r
ex
p
er
im
en
ts
d
em
o
n
s
tr
ate
d
th
at
th
e
p
r
o
p
o
s
ed
s
y
s
tem
s
ig
n
if
ican
tly
im
p
r
o
v
es
v
e
h
icle
d
etec
tio
n
p
r
ec
is
io
n
u
n
d
er
v
ar
y
i
n
g
lig
h
tin
g
co
n
d
itio
n
s
.
Fo
r
in
s
tan
ce
,
as
s
h
o
wn
in
T
ab
les
1
a
n
d
2
,
th
e
s
y
s
tem
ac
h
iev
ed
an
p
r
ec
is
io
n
r
ate
o
f
u
p
t
o
9
9
.
2
%
d
u
r
in
g
d
ay
tim
e
an
d
u
p
t
o
9
7
.
5
7
%
at
n
ig
h
t.
T
h
ese
r
esu
lts
in
d
icate
th
at
th
e
co
m
b
in
atio
n
o
f
n
o
r
m
alize
d
cr
o
s
s
-
co
r
r
elatio
n
,
2
D
-
DW
T
,
an
d
C
NN
p
r
o
v
id
es
r
o
b
u
s
t
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
d
if
f
er
en
t scen
ar
i
o
s
.
W
h
er
e
s
o
m
e
o
f
its
r
esu
lts
ar
e
s
h
o
wn
in
Fig
u
r
es 7
an
d
8
.
T
ab
le
1
.
E
v
alu
atio
n
r
esu
lts
o
f
th
e
th
r
ee
v
e
h
icle
d
etec
tio
n
m
et
h
o
d
s
d
u
r
in
g
d
ay
V
i
d
e
o
s
1
2
3
M
e
t
h
o
d
s
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
R
u
a
n
e
t
a
l
.
[
5
]
9
5
.
1
1
9
4
.
8
6
9
6
.
2
3
9
6
.
0
3
96
9
5
.
8
9
W
e
i
e
t
a
l
.
[
6
]
9
7
.
5
8
9
7
.
2
7
9
7
.
9
7
9
8
.
0
1
9
7
.
6
8
9
7
.
1
1
O
u
r
p
r
o
p
o
s
e
d
w
o
r
k
9
8
.
8
6
9
8
.
9
3
9
9
.
2
9
8
.
7
5
9
9
.
0
5
9
8
.
8
2
T
ab
le
2
.
E
v
alu
atio
n
r
esu
lts
o
f
th
e
th
r
ee
v
e
h
icle
d
etec
tio
n
m
et
h
o
d
s
d
u
r
in
g
n
ig
h
t
V
i
d
e
o
s
1
2
3
M
e
t
h
o
d
s
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
K
u
a
n
g
e
t
a
l
.
[
1
2
]
9
4
.
3
9
4
.
0
5
9
4
.
8
3
9
4
.
1
7
9
5
.
1
1
9
4
.
0
6
H
e
mm
a
t
i
e
t
a
l
.
[
1
3
]
9
6
.
8
1
9
5
.
9
9
6
.
3
2
9
5
.
9
3
9
5
.
7
5
9
6
.
1
O
u
r
p
r
o
p
o
s
e
d
w
o
r
k
9
7
.
5
7
9
7
.
0
1
97
9
6
.
8
6
9
6
.
1
9
9
6
.
5
T
o
p
r
o
p
er
l
y
e
v
alu
ate
th
e
s
y
s
tem
th
at
we
h
av
e
p
r
o
p
o
s
ed
,
we
u
s
ed
f
o
u
r
d
if
f
er
en
t
wo
r
k
s
f
r
o
m
th
e
liter
atu
r
e
to
ev
alu
ate
th
e
p
r
o
p
o
s
ed
s
y
s
tem
ag
ain
s
t.
W
e
u
s
ed
two
wo
r
k
s
to
ev
alu
ate
th
e
d
ay
-
tim
e
d
etec
tio
n
m
eth
o
d
.
T
h
e
f
ir
s
t
wo
r
k
,
p
r
ese
n
ted
b
y
R
u
an
et
a
l.
[
5
]
,
in
tr
o
d
u
ce
s
a
v
eh
icle
d
etec
tio
n
s
y
s
te
m
tar
g
etin
g
v
eh
icle
wh
ee
l
id
en
tific
atio
n
,
co
m
p
r
i
s
in
g
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
v
eh
icle
d
etec
tio
n
s
tag
es.
I
t
in
co
r
p
o
r
ates
R
OI
s
eg
m
en
tatio
n
,
wh
ee
l
id
e
n
tific
atio
n
,
an
d
v
eh
icle
d
etec
tio
n
.
A
k
ey
d
r
awb
ac
k
is
its
d
e
p
en
d
en
ce
o
n
wh
ee
l
v
is
ib
ilit
y
,
p
r
o
n
e
to
o
b
s
tr
u
ctio
n
o
r
o
cc
lu
s
io
n
in
r
ea
l
-
wo
r
ld
s
ce
n
ar
io
s
,
im
p
air
in
g
ac
cu
r
ac
y
.
Fu
r
th
er
m
o
r
e,
th
e
m
eth
o
d
m
a
y
s
tr
u
g
g
le
with
co
m
p
lex
b
ac
k
g
r
o
u
n
d
s
an
d
v
ar
y
i
n
g
lig
h
tin
g
c
o
n
d
itio
n
s
d
u
r
in
g
th
e
d
ay
.
T
h
e
s
ec
o
n
d
wo
r
k
,
b
y
W
ei
et
a
l.
[
6
]
,
p
r
o
p
o
s
es
a
v
eh
icle
r
ec
o
g
n
itio
n
m
eth
o
d
u
s
in
g
Haa
r
an
d
HOG
f
ea
tu
r
es
f
o
r
R
OI
s
eg
m
en
tatio
n
an
d
Ad
aBo
o
s
t/S
VM
clas
s
if
ier
s
f
o
r
clas
s
if
ica
tio
n
.
T
h
e
p
er
f
o
r
m
an
ce
o
f
th
is
m
eth
o
d
s
u
f
f
er
s
in
r
ea
l
-
wo
r
ld
s
ettin
g
s
d
u
e
to
d
iv
er
s
e
v
eh
icle
s
h
ap
es,
s
izes,
an
d
co
lo
r
s
.
C
o
m
b
in
in
g
m
u
ltip
le
f
ea
tu
r
es
an
d
class
if
ier
s
al
s
o
h
in
d
er
s
r
ea
l
-
ti
m
e
im
p
lem
en
tatio
n
i
n
DAS.
T
h
e
co
m
p
a
r
is
o
n
b
etwe
en
th
e
p
r
o
p
o
s
ed
m
eth
o
d
a
n
d
th
o
s
e
b
y
R
u
an
et
a
l.
[
5
]
a
n
d
W
ei
et
a
l.
[
6
]
,
as
d
etailed
in
T
ab
le
1
,
illu
s
tr
ates
s
ig
n
if
ican
t
im
p
r
o
v
em
en
ts
in
b
o
th
p
r
ec
is
io
n
a
n
d
r
ec
all.
As
m
en
tio
n
e
d
ea
r
lier
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
n
h
a
n
ce
d
d
r
ivin
g
a
s
s
is
ta
n
ce
:
a
u
to
ma
ted
d
a
y
a
n
d
n
ig
h
t v
e
h
i
cle
d
etec
tio
n
s
ystem
… (
A
b
d
el
mo
g
h
it Za
a
r
a
n
e
)
1539
o
u
r
p
r
o
p
o
s
e
d
m
e
t
h
o
d
a
c
h
i
e
v
e
s
p
r
e
c
i
s
i
o
n
o
f
u
p
t
o
9
9
.
2
%
a
n
d
r
e
c
a
l
l
o
f
9
8
.
9
3
%
.
I
n
c
o
n
t
r
a
s
t
,
R
u
an
et
a
l.
[
5
]
ac
h
iev
e
a
p
r
ec
is
io
n
o
f
u
p
to
9
6
.
2
3
%
an
d
a
r
ec
all
o
f
u
p
to
9
6
.
0
3
%,
wh
ile
W
ei
et
a
l.
[
6
]
ac
h
iev
e
a
p
r
ec
is
io
n
o
f
u
p
to
9
7
.
9
7
%
an
d
a
r
ec
all
o
f
u
p
to
9
8
.
0
1
%.
Alth
o
u
g
h
W
ei
et
a
l.
[
6
]
m
et
h
o
d
p
er
f
o
r
m
s
b
et
ter
th
an
R
u
an
et
a
l.
[
5
]
b
o
th
f
all
s
h
o
r
t
o
f
th
e
p
r
ec
is
io
n
a
n
d
r
ec
all
ac
h
iev
e
d
b
y
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
.
T
h
ese
h
ig
h
er
v
alu
es
d
em
o
n
s
tr
ate
th
e
s
u
p
er
i
o
r
ab
il
ity
o
f
o
u
r
m
eth
o
d
to
ac
cu
r
ately
d
etec
t
an
d
class
if
y
v
eh
icl
es
d
u
r
in
g
d
a
y
tim
e,
u
n
d
er
s
co
r
i
n
g
its
r
o
b
u
s
tn
ess
an
d
ef
f
ec
tiv
en
ess
ac
r
o
s
s
v
ar
io
u
s
d
ay
tim
e
s
ce
n
ar
io
s
.
Fi
g
u
r
e
7
.
So
m
e
r
esu
lts
o
f
d
a
y
-
tim
e
v
eh
icle
Fig
u
r
e
8
.
So
m
e
r
esu
lts
o
f
n
ig
h
t
-
tim
e
v
eh
icle
de
t
e
c
t
i
o
n
T
o
ev
alu
ate
th
e
n
ig
h
t
-
tim
e
d
etec
tio
n
m
eth
o
d
,
we
u
s
ed
th
e
o
th
er
two
wo
r
k
s
.
T
h
e
f
ir
s
t
wo
r
k
,
b
y
Ku
an
g
et
a
l.
[
1
2
]
,
p
r
esen
tes
a
n
ig
h
t
-
tim
e
v
eh
icle
d
etec
tio
n
s
y
s
tem
in
teg
r
atin
g
R
OI
e
x
tr
ac
tio
n
with
im
ag
e
en
h
an
ce
m
e
n
t
tech
n
iq
u
es
to
im
p
r
o
v
e
v
is
ib
ilit
y
u
n
d
er
lo
w
-
li
g
h
t
co
n
d
itio
n
s
.
Ho
wev
e
r
,
its
ef
f
icac
y
r
elies
h
ea
v
ily
o
n
p
r
ep
r
o
ce
s
s
in
g
q
u
ality
,
p
o
t
en
tially
lead
in
g
to
f
alse
p
o
s
it
iv
es
o
r
m
is
s
ed
d
etec
tio
n
s
.
Ad
ap
tab
ilit
y
to
v
ar
ied
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
36
,
No
.
3
,
Dec
em
b
er
20
24
:
1
5
32
-
1
5
42
1540
n
ig
h
t
-
tim
e
co
n
d
itio
n
s
an
d
ca
m
er
a
s
en
s
o
r
s
is
also
lim
ited
b
y
its
r
elian
ce
o
n
s
p
ec
if
ic
en
h
a
n
ce
m
en
t
tech
n
iq
u
es.
T
h
e
s
ec
o
n
d
wo
r
k
,
b
y
Hem
m
a
ti
et
a
l.
[
1
3
]
,
p
r
o
p
o
s
es
an
ad
ap
tiv
e
s
y
s
tem
f
o
r
r
ea
l
-
tim
e
au
to
n
o
m
o
u
s
d
r
iv
in
g
,
f
o
cu
s
in
g
o
n
d
etec
tin
g
v
e
h
icle
h
ea
d
lig
h
ts
at
n
ig
h
t.
T
h
is
ap
p
r
o
ac
h
id
en
tifie
s
b
r
ig
h
t
r
e
g
io
n
s
as
h
ea
d
lig
h
t,
ef
f
ec
tiv
e
f
o
r
v
e
h
icles
with
v
is
i
b
le
h
ea
d
lig
h
ts
b
u
t
less
s
o
f
o
r
o
th
er
s
.
I
t
m
ay
p
r
o
d
u
ce
f
alse
p
o
s
itiv
es
f
r
o
m
n
o
n
-
v
eh
icle
b
r
ig
h
t o
b
jects lik
e
s
tr
e
etlig
h
ts
,
im
p
ac
tin
g
d
etec
tio
n
a
cc
u
r
ac
y
.
T
h
e
r
esu
lts
f
o
r
n
i
g
h
t
-
tim
e
v
e
h
icle
d
etec
tio
n
,
p
r
esen
ted
in
T
ab
le
2
,
also
h
ig
h
lig
h
t
th
e
s
u
p
er
io
r
ity
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
co
m
p
ar
ed
to
th
e
wo
r
k
s
b
y
Ku
an
g
et
a
l.
[
1
2
]
an
d
Hem
m
ati
et
a
l.
[
1
3
]
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
ac
h
ie
v
es
p
r
ec
is
io
n
o
f
u
p
to
9
7
.
5
7
%
an
d
r
ec
all
o
f
u
p
t
o
9
7
.
0
1
%
ac
r
o
s
s
th
e
th
r
ee
v
id
eo
s
.
I
n
co
m
p
ar
is
o
n
,
K
u
an
g
et
a
l
.
[
1
2
]
m
eth
o
d
s
h
o
ws
lo
wer
p
e
r
f
o
r
m
an
ce
with
p
r
ec
is
io
n
o
f
u
p
to
9
5
.
1
1
%
an
d
r
ec
all
o
f
u
p
to
9
4
.
1
7
%.
Hem
m
ati
et
a
l.
[
1
3
]
a
p
p
r
o
ac
h
p
er
f
o
r
m
s
b
ette
r
th
an
Ku
a
n
g
et
a
l.
[
1
2
]
b
u
t
s
till
lag
s
b
eh
in
d
th
e
p
r
o
p
o
s
ed
m
eth
o
d
,
with
p
r
ec
is
io
n
o
f
u
p
to
9
6
.
8
1
%
an
d
r
ec
al
l
o
f
u
p
to
9
6
.
1
%.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
'
s
h
ig
h
er
p
r
ec
is
io
n
an
d
r
ec
all
r
ates
r
ef
lect
its
en
h
an
ce
d
ca
p
ab
ilit
y
to
ac
cu
r
ately
d
etec
t
v
e
h
icles
u
n
d
er
lo
w
-
lig
h
t
co
n
d
itio
n
s
,
th
u
s
o
f
f
er
in
g
m
o
r
e
r
eliab
le
p
er
f
o
r
m
an
ce
in
n
ig
h
t
-
tim
e
s
ce
n
ar
io
s
an
d
d
e
m
o
n
s
tr
atin
g
s
ig
n
if
ican
t
ad
v
an
ce
m
e
n
ts
o
v
er
e
x
is
tin
g
m
eth
o
d
s
.
Ou
r
p
u
r
p
o
s
e
is
to
d
esig
n
a
s
y
s
tem
th
at
n
o
t
o
n
ly
h
as
a
b
ette
r
s
u
cc
ess
f
u
l
d
etec
tio
n
r
ate
,
b
u
t
it
s
h
o
u
ld
also
b
e
as
f
ast
as
p
o
s
s
ib
le.
Fig
u
r
e
9
s
h
o
ws
th
e
s
p
ee
d
o
f
o
u
r
s
y
s
tem
in
th
e
th
r
ee
ex
p
e
r
im
en
t
s
.
W
h
er
e
th
e
C
NN
-
b
ased
d
ay
tim
e
v
eh
icle
d
etec
tio
n
s
y
s
tem
co
u
ld
p
r
o
ce
s
s
u
p
to
2
1
.
6
4
f
r
am
es
p
e
r
s
ec
o
n
d
,
at
th
e
s
am
e
tim
e
th
e
av
er
ag
e
f
r
am
es
p
er
s
ec
o
n
d
p
r
o
ce
s
s
ed
f
o
r
th
e
th
r
ee
v
id
eo
s
i
s
2
0
.
7
2
f
r
am
es
p
er
s
ec
o
n
d
.
I
n
ad
d
itio
n
,
th
e
n
ig
h
t
-
tim
e
v
eh
icle
d
etec
tio
n
s
y
s
tem
co
u
ld
p
r
o
ce
s
s
u
p
to
2
3
.
0
9
f
r
a
m
es
p
er
s
ec
o
n
d
,
wh
ile
2
2
.
6
5
f
r
am
es
p
er
s
ec
o
n
d
is
th
e
av
er
ag
e
f
r
am
es
p
er
s
ec
o
n
d
p
r
o
ce
s
s
ed
f
o
r
th
e
th
r
ee
v
id
eo
s
.
T
h
ese
r
esu
lts
ar
e
s
u
f
f
i
cien
t
f
o
r
r
ea
l
-
tim
e
p
r
o
ce
s
s
in
g
.
Fig
u
r
e
9
.
Sp
ee
d
o
f
o
u
r
s
y
s
tem
in
th
e
th
r
ee
e
x
p
er
im
e
n
ts
On
e
o
f
th
e
m
ain
s
tr
en
g
th
s
o
f
o
u
r
m
e
th
o
d
is
its
ad
ap
tab
ilit
y
to
d
if
f
er
e
n
t
lig
h
tin
g
co
n
d
itio
n
s
with
o
u
t
r
eq
u
ir
in
g
ad
d
itio
n
al
h
ar
d
war
e
,
s
u
ch
as
laser
s
.
T
h
is
is
a
s
ig
n
i
f
ican
t
im
p
r
o
v
em
en
t
o
v
er
m
et
h
o
d
s
th
at
d
ep
e
n
d
o
n
laser
an
d
v
id
eo
in
f
o
r
m
atio
n
[
1
1
]
.
Ho
wev
er
,
o
u
r
s
y
s
tem
'
s
p
er
f
o
r
m
a
n
ce
s
lig
h
tly
d
r
o
p
s
in
ex
tr
em
ely
l
o
w
-
lig
h
t
co
n
d
itio
n
s
,
wh
ic
h
we
p
la
n
to
ad
d
r
ess
in
f
u
tu
r
e
wo
r
k
.
Un
ex
p
ec
ted
ly
,
th
e
s
y
s
tem
p
er
f
o
r
m
ed
b
etter
in
d
etec
tin
g
s
m
aller
v
eh
icles a
t n
ig
h
t c
o
m
p
ar
ed
to
lar
g
er
o
n
es,
wh
ich
we
attr
ib
u
te
to
th
e
s
p
ec
if
ic
tr
ain
i
n
g
d
ata
u
s
ed
.
4.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
p
r
o
p
o
s
es
an
au
to
m
ated
v
eh
icle
d
etec
tio
n
s
y
s
tem
f
o
r
d
r
i
v
er
ass
is
tan
ce
s
y
s
te
m
s
(
DAS)
,
wh
er
e
v
eh
icles
ar
e
d
etec
ted
i
n
r
ea
l
tim
e
d
u
r
in
g
d
a
y
an
d
n
ig
h
t.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
f
ir
s
t
p
r
o
ce
s
s
es
th
e
u
p
p
er
p
ar
t
o
f
th
e
in
p
u
t
v
i
d
eo
s
at
ea
c
h
p
er
io
d
t
o
d
etec
t
wh
eth
e
r
th
e
v
id
eo
s
wer
e
r
ec
o
r
d
ed
d
u
r
in
g
d
ay
o
r
n
ig
h
t.
T
h
e
n
,
if
d
ay
-
tim
e
is
d
etec
ted
,
we
p
er
f
o
r
m
th
e
d
a
y
-
tim
e
v
eh
icle
d
et
ec
tio
n
m
eth
o
d
,
wh
ic
h
co
n
s
is
ts
o
f
two
m
ajo
r
p
ar
ts
:
th
e
h
y
p
o
th
esis
g
en
er
atio
n
p
a
r
t,
an
d
th
e
h
y
p
o
th
esis
v
er
if
icatio
n
p
ar
t.
At
th
e
f
ir
s
t
p
ar
t,
th
e
h
y
p
o
th
eses
ar
e
g
en
er
ated
b
y
d
etec
tin
g
th
e
v
e
h
icle
p
atter
n
s
u
s
in
g
a
n
o
r
m
al
ized
cr
o
s
s
-
co
r
r
elatio
n
a
f
ter
c
o
n
v
er
tin
g
th
e
in
p
u
t
im
ag
es
to
HSV
co
lo
r
s
p
ac
e,
p
er
f
o
r
m
in
g
h
is
to
g
r
am
eq
u
aliza
tio
n
an
d
r
u
n
n
in
g
th
e
C
an
n
y
ed
g
e
d
etec
to
r
.
At
th
e
0
5
10
15
20
25
D
a
y
t
i
me
p
r
o
p
o
se
d
sy
st
e
m
N
i
g
h
t
t
i
me
p
r
o
p
o
se
d
sy
st
e
m
Fr
a
m
e
p
e
r
sec
o
n
d
Pr
o
p
o
sed
sy
st
e
m
s
V
i
d
e
o
1
V
i
d
e
o
2
V
i
d
e
o
3
A
v
e
r
a
g
e
V
a
l
u
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
E
n
h
a
n
ce
d
d
r
ivin
g
a
s
s
is
ta
n
ce
:
a
u
to
ma
ted
d
a
y
a
n
d
n
ig
h
t v
e
h
i
cle
d
etec
tio
n
s
ystem
… (
A
b
d
el
mo
g
h
it Za
a
r
a
n
e
)
1541
s
ec
o
n
d
p
ar
t,
we
v
er
if
y
t
h
e
h
y
p
o
th
eses
g
en
er
ated
an
d
class
if
y
th
em
as
v
eh
icles
o
r
n
o
n
-
v
eh
ic
les
u
s
in
g
th
e
C
NN
class
if
ier
af
ter
ex
tr
ac
tin
g
th
ei
r
f
ea
tu
r
es
with
th
e
th
ir
d
lev
el
o
f
2
D
-
DW
T
.
An
d
if
n
ig
h
t
-
tim
e
is
d
etec
ted
,
we
ap
p
ly
th
e
n
ig
h
t
-
tim
e
v
eh
icle
d
etec
tio
n
m
eth
o
d
,
wh
ic
h
is
d
i
v
i
d
ed
in
to
two
m
ain
s
tag
es:
th
e
g
en
er
atio
n
o
f
lam
p
ca
n
d
id
ates
an
d
th
e
v
er
if
icati
o
n
lam
p
ca
n
d
id
ates.
At
th
e
f
ir
s
t
s
tag
e,
we
g
en
er
ate
th
e
ca
n
d
id
ates
b
y
f
ir
s
t
p
er
f
o
r
m
in
g
th
e
co
lo
r
th
r
esh
o
l
d
tech
n
iq
u
e
o
n
in
p
u
t
im
ag
es
in
HSV
co
lo
r
s
p
ac
e
to
ex
tr
ac
t
th
e
r
e
d
an
d
wh
ite
b
r
ig
h
t
r
e
g
io
n
s
(
r
ea
r
-
lam
p
ca
n
d
id
ates
an
d
h
ea
d
lam
p
ca
n
d
i
d
ates)
,
th
en
we
ex
tr
ac
t
ea
ch
b
r
ig
h
t
r
e
g
io
n
as
a
co
m
p
o
n
en
t
u
s
in
g
th
e
c
o
n
n
ec
te
d
co
m
p
o
n
e
n
t
tech
n
iq
u
e
af
te
r
p
er
f
o
r
m
in
g
s
o
m
e
m
o
r
p
h
o
l
o
g
ica
l
f
u
n
ctio
n
s
.
At
th
e
s
ec
o
n
d
s
tag
e,
we
v
er
if
y
th
e
g
en
er
ated
ca
n
d
id
ates
b
y
f
ir
s
t
p
air
in
g
th
em
in
to
s
ets.
T
h
en
,
we
k
ee
p
o
n
ly
s
y
m
m
etr
ically
id
en
tical
p
air
s
u
s
in
g
th
e
n
o
r
m
alize
d
c
r
o
s
s
-
co
r
r
elatio
n
a
n
d
class
if
y
th
em
as
v
eh
icles
o
r
n
o
n
-
v
eh
icles
u
s
in
g
th
e
C
NN
clas
s
i
f
ier
.
E
x
p
er
im
e
n
ts
h
av
e
s
h
o
wn
th
at
th
e
p
r
o
p
o
s
ed
s
y
s
tem
wo
r
k
s
well
an
d
d
etec
ts
v
eh
icles
with
a
h
ig
h
ac
cu
r
ac
y
.
I
n
a
d
d
itio
n
,
a
co
m
p
ar
ativ
e
s
tu
d
y
was
ca
r
r
ie
d
o
u
t
b
etwe
en
t
h
e
p
r
o
p
o
s
ed
s
y
s
tem
an
d
o
th
er
w
o
r
k
s
in
th
e
liter
at
u
r
e
in
o
r
d
er
to
o
b
tain
a
g
o
o
d
ev
alu
atio
n
.
T
h
e
o
b
tain
e
d
r
esu
lts
p
r
o
v
e
th
e
h
i
g
h
ac
cu
r
ac
y
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
T
h
e
im
p
licatio
n
s
o
f
th
ese
f
in
d
in
g
s
s
u
g
g
est
th
at
o
u
r
m
eth
o
d
ca
n
b
e
in
teg
r
ated
in
to
v
ar
io
u
s
d
r
iv
in
g
ass
i
s
tan
ce
s
y
s
tem
s
,
im
p
r
o
v
in
g
r
o
ad
s
af
ety
an
d
r
ed
u
cin
g
ac
cid
en
t
r
ates.
Fu
tu
r
e
r
esear
ch
will
f
o
cu
s
o
n
r
ef
in
i
n
g
th
e
s
y
s
tem
'
s
p
er
f
o
r
m
an
ce
in
lo
w
-
lig
h
t
co
n
d
itio
n
s
an
d
e
x
p
a
n
d
in
g
t
h
e
d
ataset
to
in
clu
d
e
a
wid
er
v
ar
iety
o
f
v
e
h
icle
ty
p
es a
n
d
en
v
ir
o
n
m
en
tal
s
ce
n
ar
io
s
.
RE
F
E
R
E
NC
E
S
[
1]
I
.
S
l
i
ma
n
i
,
A
.
Za
a
r
a
n
e
,
W
.
A
.
O
k
a
i
s
h
i
,
I
.
A
t
o
u
f
,
a
n
d
A
.
H
a
m
d
o
u
n
,
“
A
n
a
u
t
o
ma
t
e
d
l
i
c
e
n
se
p
l
a
t
e
d
e
t
e
c
t
i
o
n
a
n
d
r
e
c
o
g
n
i
t
i
o
n
sy
s
t
e
m
b
a
s
e
d
o
n
w
a
v
e
l
e
t
d
e
c
o
m
p
o
si
t
i
o
n
a
n
d
C
N
N
,
”
Arra
y
,
v
o
l
.
8
,
p
.
1
0
0
0
4
0
,
D
e
c
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
J.AR
R
A
Y
.
2
0
2
0
.
1
0
0
0
4
0
.
[
2
]
A
.
Za
a
r
a
n
e
,
I
.
S
l
i
m
a
n
i
,
W
.
A
.
O
k
a
i
s
h
i
,
I
.
A
t
o
u
f
,
a
n
d
A
.
H
a
md
o
u
n
,
“
D
i
st
a
n
c
e
me
a
su
r
e
me
n
t
sy
st
e
m
f
o
r
a
u
t
o
n
o
m
o
u
s
v
e
h
i
c
l
e
s
u
s
i
n
g
st
e
r
e
o
c
a
m
e
r
a
,
”
Arr
a
y
,
v
o
l
.
5
,
p
.
1
0
0
0
1
6
,
M
a
r
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
J
.
A
R
R
A
Y
.
2
0
2
0
.
1
0
0
0
1
6
.
[
3
]
A
.
Za
a
r
a
n
e
,
I
.
S
l
i
ma
n
i
,
A
.
H
a
m
d
o
u
n
,
a
n
d
I
.
A
t
o
u
f
,
“
V
e
h
i
c
l
e
t
o
v
e
h
i
c
l
e
d
i
st
a
n
c
e
m
e
a
s
u
r
e
m
e
n
t
f
o
r
se
l
f
-
dr
i
v
i
n
g
s
y
st
e
ms,
”
2
0
1
9
6
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
n
t
r
o
l
,
D
e
c
i
s
i
o
n
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
i
e
s
,
C
o
D
I
T
2
0
1
9
,
p
p
.
1
5
8
7
–
1
5
9
1
,
A
p
r
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
C
O
D
I
T.
2
0
1
9
.
8
8
2
0
5
7
2
.
[
4
]
C
.
M
e
n
g
,
H
.
B
a
o
,
a
n
d
Y
.
M
a
,
“
V
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
:
a
r
e
v
i
e
w
,
”
J
o
u
rn
a
l
o
f
P
h
y
si
c
s:
C
o
n
f
e
r
e
n
c
e
S
e
ri
e
s
,
v
o
l
.
1
6
3
4
,
n
o
.
1
,
p
.
0
1
2
1
0
7
,
S
e
p
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
2
-
6
5
9
6
/
1
6
3
4
/
1
/
0
1
2
1
0
7
.
[
5
]
Y
.
S
.
R
u
a
n
,
I
.
C
.
C
h
a
n
g
,
a
n
d
H
.
Y
.
Y
e
h
,
“
V
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
b
a
s
e
d
o
n
w
h
e
e
l
p
a
r
t
d
e
t
e
c
t
i
o
n
,
”
2
0
1
7
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
n
su
m
e
r E
l
e
c
t
r
o
n
i
c
s
-
T
a
i
w
a
n
,
I
C
C
E
-
T
W
2
0
1
7
,
p
p
.
1
8
7
–
1
8
8
,
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
E
-
C
h
i
n
a
.
2
0
1
7
.
7
9
9
1
0
5
8
.
[
6
]
Y
.
W
e
i
,
Q
.
Ti
a
n
,
J
.
G
u
o
,
W
.
H
u
a
n
g
,
a
n
d
J
.
C
a
o
,
“
M
u
l
t
i
-
v
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
a
l
g
o
r
i
t
h
m
t
h
r
o
u
g
h
c
o
mb
i
n
i
n
g
H
a
r
r
a
n
d
H
O
G
f
e
a
t
u
r
e
s
,
”
Ma
t
h
e
m
a
t
i
c
s
a
n
d
C
o
m
p
u
t
e
rs i
n
S
i
m
u
l
a
t
i
o
n
,
v
o
l
.
1
5
5
,
p
p
.
1
3
0
–
1
4
5
,
J
a
n
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
J.
M
A
TC
O
M
.
2
0
1
7
.
1
2
.
0
1
1
.
[
7
]
J.
R
e
d
m
o
n
,
S
.
D
i
v
v
a
l
a
,
R.
G
i
r
sh
i
c
k
,
a
n
d
A
.
F
a
r
h
a
d
i
,
“
(
Y
O
LO
)
y
o
u
o
n
l
y
l
o
o
k
o
n
c
e
,”
I
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
c
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
p
p
.
7
7
9
–
7
8
8
,
D
e
c
.
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
9
1
.
[
8
]
Y
.
C
h
e
n
a
n
d
Z
.
L
i
,
“
A
n
e
f
f
e
c
t
i
v
e
a
p
p
r
o
a
c
h
o
f
v
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
”
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
N
e
u
ro
s
c
i
e
n
c
e
,
v
o
l
.
2
0
2
2
,
n
o
.
1
,
p
.
2
0
1
9
2
5
7
,
J
a
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
2
/
2
0
1
9
2
5
7
.
[
9
]
S
.
K
a
r
l
,
Y
.
D
o
n
z
i
a
,
Y
.
-
P
.
G
e
u
m,
a
n
d
H
.
-
K
.
K
i
m
,
“
A
s
t
u
d
y
o
n
a
u
t
o
n
o
m
o
u
s d
r
i
v
i
n
g
a
d
a
p
t
i
v
e
si
mu
l
a
t
i
o
n
s
y
st
e
m
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
Y
O
LO
V
3
,
”
C
o
v
e
n
a
n
t
J
o
u
r
n
a
l
o
f
I
n
f
o
rm
a
t
i
c
s
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
9
,
n
o
.
2
,
p
p
.
1
–
1
5
,
D
e
c
.
2
0
2
1
,
A
c
c
e
ss
e
d
:
A
u
g
.
0
7
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
j
o
u
r
n
a
l
s.c
o
v
e
n
a
n
t
u
n
i
v
e
r
si
t
y
.
e
d
u
.
n
g
/
i
n
d
e
x
.
p
h
p
/
c
j
i
c
t
/
a
r
t
i
c
l
e
/
v
i
e
w
/
2
7
9
4
[
1
0
]
N
.
A
r
o
r
a
a
n
d
Y
.
K
u
mar,
“
A
u
t
o
ma
t
i
c
v
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
sy
s
t
e
m
i
n
d
a
y
a
n
d
n
i
g
h
t
m
o
d
e
:
c
h
a
l
l
e
n
g
e
s
,
a
p
p
l
i
c
a
t
i
o
n
s
a
n
d
p
a
n
o
r
a
mi
c
r
e
v
i
e
w
,
”
Ev
o
l
u
t
i
o
n
a
r
y
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
6
,
n
o
.
4
,
p
p
.
1
0
7
7
–
1
0
9
5
,
A
u
g
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
S
1
2
0
6
5
-
0
2
2
-
0
0
7
2
3
-
0.
[
1
1
]
R
.
H
.
Z
h
a
n
g
,
F
.
Y
o
u
,
F
.
C
h
e
n
,
a
n
d
W
.
Q
.
H
e
,
“
V
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
me
t
h
o
d
f
o
r
i
n
t
e
l
l
i
g
e
n
t
v
e
h
i
c
l
e
a
t
n
i
g
h
t
t
i
me
b
a
se
d
o
n
v
i
d
e
o
a
n
d
l
a
ser
i
n
f
o
r
m
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
P
a
t
t
e
rn
R
e
c
o
g
n
i
t
i
o
n
a
n
d
A
rt
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
2
,
n
o
.
4
,
A
p
r
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
4
2
/
S
0
2
1
8
0
0
1
4
1
8
5
0
0
0
9
X
.
[
1
2
]
H
.
K
u
a
n
g
,
L.
C
h
e
n
,
F
.
G
u
,
J.
C
h
e
n
,
L.
C
h
a
n
,
a
n
d
H
.
Y
a
n
,
“
C
o
m
b
i
n
i
n
g
r
e
g
i
o
n
-
of
-
i
n
t
e
r
e
st
e
x
t
r
a
c
t
i
o
n
a
n
d
i
m
a
g
e
e
n
h
a
n
c
e
me
n
t
f
o
r
n
i
g
h
t
t
i
me
v
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
,
”
I
E
EE
I
n
t
e
l
l
i
g
e
n
t
S
y
st
e
m
s
,
v
o
l
.
3
1
,
n
o
.
3
,
p
p
.
5
7
–
6
5
,
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
M
I
S
.
2
0
1
6
.
1
7
.
[
1
3
]
M
.
H
e
mm
a
t
i
,
M
.
B
i
g
l
a
r
i
-
A
b
h
a
r
i
,
a
n
d
S
.
N
i
a
r
,
“
A
d
a
p
t
i
v
e
v
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
f
o
r
r
e
a
l
-
t
i
me
a
u
t
o
n
o
mo
u
s
d
r
i
v
i
n
g
sy
s
t
e
m
,”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
9
D
e
s
i
g
n
,
Au
t
o
m
a
t
i
o
n
a
n
d
T
e
st
i
n
Eu
r
o
p
e
C
o
n
f
e
r
e
n
c
e
a
n
d
Ex
h
i
b
i
t
i
o
n
,
D
ATE
2
0
1
9
,
p
p
.
1
0
3
4
–
1
0
3
9
,
M
a
y
2
0
1
9
,
d
o
i
:
1
0
.
2
3
9
1
9
/
D
A
T
E.
2
0
1
9
.
8
7
1
4
8
1
8
.
[
1
4
]
S
.
H
u
a
n
g
e
t
a
l
.
,
“M
-
Y
O
LO
:
a
n
i
g
h
t
t
i
me
v
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
me
t
h
o
d
c
o
mb
i
n
i
n
g
m
o
b
i
l
e
n
e
t
v
2
a
n
d
Y
O
LO
v
3
,
”
I
n
J
o
u
r
n
a
l
o
f
P
h
y
s
i
c
s:
C
o
n
f
e
re
n
c
e
S
e
r
i
e
s,
v
o
l
.
1
8
8
3
,
n
o
.
1
,
p
.
0
1
2
0
9
4
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
2
-
6
5
9
6
/
1
8
8
3
/
1
/
0
1
2
0
9
4
.
[
1
5
]
Z.
Y
a
n
g
a
n
d
L
.
S
.
C
.
P
u
n
-
C
h
e
n
g
,
“
V
e
h
i
c
l
e
d
e
t
e
c
t
i
o
n
i
n
i
n
t
e
l
l
i
g
e
n
t
t
r
a
n
s
p
o
r
t
a
t
i
o
n
s
y
s
t
e
ms
a
n
d
i
t
s
a
p
p
l
i
c
a
t
i
o
n
s
u
n
d
e
r
v
a
r
y
i
n
g
e
n
v
i
r
o
n
m
e
n
t
s
:
a
r
e
v
i
e
w
,
”
I
m
a
g
e
a
n
d
Vi
si
o
n
C
o
m
p
u
t
i
n
g
,
v
o
l
.
6
9
,
p
p
.
1
4
3
–
1
5
4
,
Ja
n
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
J.I
M
A
V
I
S
.
2
0
1
7
.
0
9
.
0
0
8
.
[
1
6
]
J.
C
a
n
n
y
,
“
A
c
o
mp
u
t
a
t
i
o
n
a
l
a
p
p
r
o
a
c
h
t
o
e
d
g
e
d
e
t
e
c
t
i
o
n
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
P
a
t
t
e
r
n
A
n
a
l
y
s
i
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
P
A
M
I
-
8
,
n
o
.
6
,
p
p
.
6
7
9
–
6
9
8
,
1
9
8
6
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
1
9
8
6
.
4
7
6
7
8
5
1
.
[
1
7
]
S
.
D
.
W
e
i
a
n
d
S
.
H
.
La
i
,
“
F
a
st
t
e
mp
l
a
t
e
ma
t
c
h
i
n
g
b
a
s
e
d
o
n
n
o
r
m
a
l
i
z
e
d
c
r
o
s
s
c
o
r
r
e
l
a
t
i
o
n
w
i
t
h
a
d
a
p
t
i
v
e
mu
l
t
i
l
e
v
e
l
w
i
n
n
e
r
u
p
d
a
t
e
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
I
m
a
g
e
Pro
c
e
ss
i
n
g
,
v
o
l
.
1
7
,
n
o
.
1
1
,
p
p
.
2
2
2
7
–
2
2
3
5
,
2
0
0
8
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
P
.
2
0
0
8
.
2
0
0
4
6
1
5
.
[
1
8
]
I
.
S
l
i
ma
n
i
,
A
.
Za
a
r
a
n
e
,
a
n
d
A
.
H
a
m
d
o
u
n
,
“
C
o
n
v
o
l
u
t
i
o
n
a
l
g
o
r
i
t
h
m
f
o
r
i
m
p
l
e
men
t
i
n
g
2
D
d
i
scre
t
e
w
a
v
e
l
e
t
t
r
a
n
sf
o
r
m
o
n
t
h
e
F
P
G
A
,
”
Pro
c
e
e
d
i
n
g
s
o
f
I
EEE
/
A
C
S
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
S
y
st
e
m
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s,
A
I
C
C
S
A
,
v
o
l
.
0
,
Ju
l
.
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
A
I
C
C
S
A
.
2
0
1
6
.
7
9
4
5
8
3
1
.
[
1
9
]
K
.
N
o
g
u
e
i
r
a
,
O
.
A
.
B
.
P
e
n
a
t
t
i
,
a
n
d
J.
A
.
d
o
s
S
a
n
t
o
s
,
“
T
o
w
a
r
d
s
b
e
t
t
e
r
e
x
p
l
o
i
t
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
r
e
mo
t
e
se
n
s
i
n
g
sce
n
e
c
l
a
ssi
f
i
c
a
t
i
o
n
,
”
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
,
v
o
l
.
6
1
,
p
p
.
5
3
9
–
5
5
6
,
Ja
n
.
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
J.PA
T
C
O
G
.
2
0
1
6
.
0
7
.
0
0
1
.
[
2
0
]
C
.
S
z
e
g
e
d
y
,
S
.
I
o
f
f
e
,
V
.
V
a
n
h
o
u
c
k
e
,
a
n
d
A
.
A
.
A
l
e
mi
,
“
I
n
c
e
p
t
i
o
n
-
v
4
,
I
n
c
e
p
t
i
o
n
-
R
e
sN
e
t
a
n
d
t
h
e
i
m
p
a
c
t
o
f
r
e
si
d
u
a
l
c
o
n
n
e
c
t
i
o
n
s o
n
l
e
a
r
n
i
n
g
,
”
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
AA
AI
C
o
n
f
e
re
n
c
e
o
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
1
,
n
o
.
1
,
p
p
.
4
2
7
8
–
4
2
8
4
,
F
e
b
.
2
0
1
7
,
d
o
i
:
1
0
.
1
6
0
9
/
A
A
A
I
.
V
3
1
I
1
.
1
1
2
3
1
.
[
2
1
]
R
.
C
.
G
o
n
z
a
l
e
s a
n
d
P
.
W
i
n
t
z
,
D
i
g
i
t
a
l
i
m
a
g
e
p
r
o
c
e
ssi
n
g
,
2
n
d
e
d
.
A
d
d
i
s
o
n
-
W
e
sl
e
y
L
o
n
g
m
a
n
P
u
b
l
i
s
h
i
n
g
C
o
.
,
1
9
8
7
.
A
c
c
e
sse
d
:
A
u
g
.
0
7
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
d
l
.
a
c
m
.
o
r
g
/
d
o
i
/
a
b
s/
1
0
.
5
5
5
5
/
2
2
8
8
1
.
[
2
2
]
“
C
a
l
t
e
c
h
c
a
r
s
d
a
t
a
s
e
t
,
”
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
:
/
/
w
w
w
.
v
i
s
i
o
n
.
c
a
l
t
e
c
h
.
e
d
u
/
a
r
c
h
i
v
e
.
h
t
m
l
[
2
3
]
G
.
H
su
,
J.
C
h
e
n
,
Y
.
C
.
-
I
.
t
r
a
n
s
a
c
t
i
o
n
s
o
n
v
e
h
i
c
u
l
a
r
,
a
n
d
u
n
d
e
f
i
n
e
d
2
0
1
2
,
“
A
p
p
l
i
c
a
t
i
o
n
-
o
r
i
e
n
t
e
d
l
i
c
e
n
s
e
p
l
a
t
e
r
e
c
o
g
n
i
t
i
o
n
,
”
I
EE
E
t
ra
n
s
a
c
t
i
o
n
s
o
n
v
e
h
i
c
u
l
a
r t
e
c
h
n
o
l
o
g
y
,
v
o
l
.
6
2
,
n
o
.
2
,
2
0
1
3
,
d
o
i
:
1
0
.
1
1
0
9
/
TV
T.
2
0
1
2
.
2
2
2
6
2
1
8
.
Evaluation Warning : The document was created with Spire.PDF for Python.