I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
24
,
p
p
.
6
3
1
3
~
6
3
2
1
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
14
i
6
.
pp
6
3
1
3
-
6
3
2
1
6313
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Tra
ff
ic sig
ns d
ete
ction a
nd proh
i
bi
tor sig
ns reco
g
niti
o
n in
M
o
ro
cco ro
a
d scene
I
m
a
ne
T
a
o
uq
i
1
,
Abde
s
s
a
m
a
d K
lilo
u
1
,
K
ebir
Cha
j
i
1
,
Ass
i
a
Arsa
la
ne
2
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
M
i
S
ET
Te
a
m,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
s
a
n
d
T
e
c
h
n
o
l
o
g
i
e
s,
S
u
l
t
a
n
M
o
u
l
a
y
S
l
i
m
a
n
e
U
n
i
v
e
r
s
i
t
y
,
B
e
n
i
M
e
l
l
a
l
,
M
o
r
o
c
c
o
2
La
b
o
r
a
t
o
r
y
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
A
p
p
l
i
e
d
Te
c
h
n
o
l
o
g
i
e
s
,
H
i
g
h
e
r
S
c
h
o
o
l
o
f
T
e
c
h
n
o
l
o
g
y
,
S
u
l
t
a
n
M
o
u
l
a
y
S
l
i
m
a
n
e
U
n
i
v
e
r
si
t
y
,
B
e
n
i
M
e
l
l
a
l
,
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
Feb
1
,
2
0
2
4
R
ev
is
ed
J
u
l 1
1
,
2
0
2
4
Acc
ep
ted
J
u
l 1
7
,
2
0
2
4
Traffic
sig
n
d
e
tec
ti
o
n
is
a
c
ru
c
ial
a
sp
e
c
t
o
f
a
d
v
a
n
c
e
d
d
r
iv
e
r
a
ss
istan
c
e
sy
ste
m
s
(AD
AS)
fo
r
a
c
a
d
e
m
ic
re
se
a
rc
h
a
n
d
t
h
e
a
u
to
m
o
ti
v
e
i
n
d
u
st
ry
.
se
e
in
g
th
a
t
a
c
c
u
ra
te
a
n
d
ti
m
e
ly
d
e
tec
ti
o
n
o
f
traffic
sig
n
s
(TS
)
is
e
ss
e
n
ti
a
l
fo
r
e
n
su
rin
g
t
h
e
sa
fe
ty
o
f
d
riv
in
g
.
Ho
we
v
e
r,
TS
d
e
tec
ti
o
n
m
e
th
o
d
s
e
n
c
o
u
n
ter
c
h
a
ll
e
n
g
e
s
li
k
e
slo
w
d
e
tec
ti
o
n
s
p
e
e
d
a
n
d
a
lac
k
o
f
ro
b
u
st
n
e
ss
in
c
o
m
p
lex
e
n
v
iro
n
m
e
n
ts.
Th
is
p
a
p
e
r
s
u
g
g
e
sts
a
d
d
re
ss
in
g
th
e
se
li
m
it
a
ti
o
n
s
b
y
p
ro
p
o
sin
g
th
e
u
se
o
f
th
e
y
o
u
o
n
ly
lo
o
k
o
n
e
v
e
rsio
n
7
(
YO
LOv
7
)
n
e
two
rk
to
d
e
tec
t
a
n
d
re
c
o
g
n
ize
TS
i
n
r
o
a
d
sc
e
n
e
s.
F
u
r
th
e
rm
o
re
,
th
e
k
-
m
e
a
n
s+
+
a
lg
o
rit
h
m
is
u
se
d
to
a
c
q
u
ire
a
n
c
h
o
r
b
o
x
e
s.
Ad
d
it
i
o
n
a
ll
y
,
a
ti
n
y
v
e
rsio
n
o
f
YO
LO
v
7
is
u
se
d
to
tak
e
a
d
v
a
n
tag
e
o
f
it
s
re
a
l
-
ti
m
e
a
n
d
lo
w
m
o
d
e
l
siz
e
,
w
h
ich
a
re
re
q
u
ired
f
o
r
re
a
l
-
ti
m
e
h
a
rd
wa
re
imp
lem
e
n
tatio
n
.
S
o
,
we
c
o
n
d
u
c
ted
a
n
e
x
p
e
ri
m
e
n
t
u
sin
g
o
u
r
p
r
o
p
r
ieta
ry
M
o
r
o
c
c
o
d
a
tas
e
t.
Ac
c
o
rd
in
g
to
t
h
e
e
x
p
e
rime
n
t
a
l
re
su
lt
s,
YO
LOv
7
a
c
h
iev
e
s
8
5
%
i
n
term
s
o
f
m
e
a
n
a
v
e
ra
g
e
p
re
c
isio
n
(m
AP)
a
t
0
.
5
f
o
r
a
ll
c
las
se
s.
An
d
YO
LOv
7
-
t
in
y
o
b
tain
s
9
0
%
in
t
h
e
sa
m
e
term
.
Af
terw
a
rd
,
a
re
c
o
g
n
it
i
o
n
sy
ste
m
fo
r
t
h
e
p
r
o
h
i
b
it
iv
e
c
las
s
u
si
n
g
th
e
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
(CNN
)
is
trai
n
e
d
a
n
d
i
n
teg
ra
ted
in
si
d
e
th
e
YO
LOv
7
a
lg
o
rit
h
m
;
i
ts
m
o
d
e
l
a
c
h
iev
e
s
a
n
a
c
c
u
ra
c
y
o
f
9
9
%
,
wh
ich
lea
d
s
to
a
g
o
o
d
sp
e
c
if
ica
ti
o
n
o
f
th
e
p
r
o
h
i
b
it
i
v
e
sig
n
m
e
a
n
in
g
.
K
ey
w
o
r
d
s
:
Ad
v
an
ce
d
d
r
iv
er
ass
is
tan
ce
s
y
s
tem
I
m
ag
e
p
r
o
ce
s
s
in
g
K
-
m
ea
n
s
++
R
ec
o
g
n
itio
n
T
r
af
f
ic
s
ig
n
d
etec
tio
n
YOL
Ov
7
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
:
I
m
an
e
T
ao
u
q
i
Dep
ar
tm
en
t o
f
E
lectr
ical
E
n
g
i
n
ee
r
in
g
MiSE
T
T
ea
m
,
Facu
lty
o
f
Scien
ce
s
an
d
T
ec
h
n
o
l
o
g
ies,
Su
ltan
Mo
u
lay
Sli
m
an
e
Un
iv
er
s
ity
5
2
3
,
B
en
i M
ellal,
Mo
r
o
cc
o
E
m
ail: im
an
e.
tao
u
q
i@
u
s
m
s
.
m
a
1.
I
NT
RO
D
UCT
I
O
N
T
r
af
f
ic
s
ig
n
d
etec
tio
n
s
y
s
tem
s
h
av
e
a
cr
u
cial
r
o
le
in
r
e
d
u
cin
g
ac
cid
en
ts
a
n
d
i
n
cid
en
ts
,
h
elp
in
g
d
r
iv
er
s
an
ticip
ate
a
n
d
r
ea
ct
to
p
o
te
n
tially
d
an
g
er
o
u
s
s
i
tu
atio
n
s
.
Ho
wev
er
,
tr
af
f
ic
s
i
g
n
s
(
T
S)
d
etec
tio
n
alg
o
r
ith
m
s
ar
e
ca
teg
o
r
ized
in
to
two
g
r
o
u
p
s
:
d
etec
tio
n
u
s
in
g
tr
ad
itio
n
al
ap
p
r
o
ac
h
es
an
d
d
etec
tio
n
u
tili
zin
g
d
ee
p
lear
n
in
g
.
T
r
a
d
itio
n
al
d
e
tectio
n
alg
o
r
ith
m
s
f
o
r
T
S
ar
e
s
tr
u
ctu
r
ed
in
to
th
r
ee
s
tep
s
:
s
elec
t
th
e
r
e
g
io
n
,
ex
tr
ac
t
th
e
f
ea
tu
r
e
,
an
d
class
if
y
th
e
f
ea
tu
r
e
[
1
]
.
T
h
e
m
ain
g
o
al
o
f
th
e
f
ir
s
t
s
tep
is
t
o
g
en
e
r
ate
a
d
iv
er
s
e
s
et
o
f
p
r
o
p
o
s
ed
r
e
g
io
n
s
wh
e
r
e
th
e
o
b
ject
m
ay
lo
ca
lize
in
th
e
f
ee
d
im
ag
e.
T
h
is
s
tep
ca
n
b
e
p
er
f
o
r
m
ed
b
y
u
s
in
g
th
e
s
elec
tiv
e
s
ea
r
ch
alg
o
r
ith
m
[
2
]
.
Af
ter
war
d
,
alg
o
r
ith
m
s
f
o
r
f
e
atu
r
e
ex
tr
ac
tio
n
,
s
u
ch
as
h
is
to
g
r
am
s
o
f
o
r
ien
ted
g
r
ad
ien
ts
[
3
]
,
ar
e
em
p
lo
y
e
d
t
o
ex
tr
ac
t
f
ea
tu
r
es
o
f
T
S
f
r
o
m
p
r
o
p
o
s
ed
r
eg
i
o
n
s
.
T
h
o
s
e
e
x
tr
ac
ted
f
ea
tu
r
es
ar
e
class
if
ied
u
s
in
g
s
u
p
er
v
is
ed
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
,
f
o
r
ex
am
p
le,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
[
4
]
a
n
d
r
an
d
o
m
f
o
r
ests
[
5
]
.
Nev
e
r
th
ele
s
s
,
th
e
tr
ad
itio
n
al
o
b
ject
d
etec
tio
n
m
eth
o
d
h
as so
m
e
lim
its
in
ter
m
s
o
f
r
ea
l
-
tim
e
p
r
o
ce
s
s
in
g
.
I
t
ca
n
also
r
eq
u
ir
e
f
r
e
q
u
en
t
r
etr
ain
in
g
if
d
atas
ets
o
r
s
ce
n
a
r
io
s
ch
a
n
g
e,
wh
i
ch
ca
n
b
e
co
s
tly
in
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
3
1
3
-
6
3
2
1
6314
ter
m
s
o
f
tim
e
an
d
h
ar
d
war
e
r
eso
u
r
ce
s
.
No
wad
ay
s
,
th
e
em
er
g
en
ce
o
f
th
e
ar
tific
ial
in
tellig
en
ce
er
a
o
n
th
e
co
m
p
u
ter
v
is
io
n
s
id
e
an
d
tec
h
n
o
lo
g
ical
ad
v
an
ce
m
e
n
ts
[
6
]
s
u
ch
as
s
o
p
h
is
ticated
s
en
s
o
r
s
an
d
h
ig
h
-
r
eso
lu
tio
n
ca
m
er
as
h
av
e
co
n
s
id
er
ab
ly
f
ac
ilit
ated
r
ea
l
-
tim
e
in
ter
ac
tio
n
with
t
h
e
e
n
v
ir
o
n
m
en
t,
en
a
b
lin
g
f
ast
an
d
m
o
r
e
ac
cu
r
ate
r
o
a
d
d
ec
is
io
n
s
.
T
h
e
o
n
ly
n
ee
d
is
to
b
e
tr
ain
e
d
with
a
lar
g
e
am
o
u
n
t
o
f
co
m
p
lex
en
v
ir
o
n
m
en
t
im
ag
e
d
ata
to
o
b
tain
ac
cu
r
ate
d
ete
ctio
n
an
d
r
ec
o
g
n
itio
n
r
esu
lts
.
I
n
th
is
ter
m
,
o
b
ject
d
etec
t
io
n
m
eth
o
d
s
h
a
v
e
in
tr
o
d
u
ce
d
m
o
r
e
in
teg
r
ate
d
a
p
p
r
o
ac
h
es
th
at
co
m
b
i
n
e
r
eg
i
o
n
s
elec
tio
n
,
f
ea
tu
r
e
ex
tr
ac
tio
n
,
an
d
class
if
icatio
n
in
to
a
s
in
g
le
p
ip
elin
e
p
r
o
ce
s
s
;
s
o
m
e
o
f
th
o
s
e
alg
o
r
ith
m
s
ar
e
co
m
p
o
s
ed
o
f
two
s
tag
es,
an
d
o
th
er
s
o
n
ly
o
n
e
s
tag
e
[
7
]
.
T
h
e
m
ain
d
if
f
e
r
en
ce
h
er
e
is
th
at
o
n
e
-
s
tag
e
d
et
ec
to
r
s
d
ir
ec
tly
p
r
ed
ict
b
o
u
n
d
i
n
g
b
o
x
es
a
n
d
class
lab
els
in
a
s
in
g
le
s
tep
with
o
u
t
a
p
r
io
r
r
eg
io
n
p
r
o
p
o
s
al
s
tep
,
wh
ich
in
v
o
lv
es
lo
wer
c
o
m
p
u
tatio
n
an
d
f
aster
s
p
ee
d
.
W
h
ile
th
e
two
-
s
tag
e
d
etec
tio
n
alg
o
r
it
h
m
s
f
o
llo
w
a
t
wo
-
s
tep
p
r
o
ce
s
s
,
T
h
e
f
ir
s
t
s
tag
e
g
e
n
er
ates
r
eg
io
n
p
r
o
p
o
s
als,
an
d
th
e
s
ec
o
n
d
s
ta
g
e
r
ef
in
es
th
ese
p
r
o
p
o
s
als
b
y
p
r
ed
ictin
g
th
e
f
in
al
b
o
u
n
d
i
n
g
b
o
x
an
d
class
lab
el,
wh
ich
lead
s
to
a
h
ig
h
er
co
m
p
u
tatio
n
co
s
t
an
d
lo
w
r
ap
i
d
ity
.
wh
ich
is
u
n
d
esira
b
le
in
th
e
T
S
d
etec
tio
n
co
n
te
x
t
[
8
]
.
T
h
e
r
ef
o
r
e
,
th
e
p
r
im
a
r
y
ch
allen
g
e
f
o
r
a
tr
af
f
ic
s
ig
n
d
ete
ctin
g
s
y
s
tem
is
to
u
n
d
er
s
tan
d
an
d
r
ea
ct
to
tr
af
f
ic
s
ig
n
s
co
r
r
ec
tly
an
d
in
s
tan
tly
,
ev
en
in
th
e
p
r
esen
ce
o
f
f
ac
to
r
s
lik
e
o
cc
lu
s
io
n
,
co
m
p
lex
b
ac
k
g
r
o
u
n
d
s
,
an
d
v
ar
iab
le
lig
h
tin
g
co
n
d
itio
n
s
.
J
ag
ad
ee
s
h
an
d
Sre
e
[
5
]
s
p
lits
th
e
T
S
r
ec
o
g
n
itio
n
p
r
o
ce
s
s
in
to
two
s
tag
es:
th
e
f
ir
s
t
in
v
o
lv
es
p
r
ep
r
o
ce
s
s
in
g
th
e
in
p
u
t
im
ag
e
(
f
ilter
in
g
,
co
lo
r
s
p
ac
e
co
n
v
er
s
io
n
)
an
d
co
lo
r
th
r
esh
o
ld
in
g
;
th
e
s
ec
o
n
d
s
tag
e
in
v
o
lv
es
f
ea
tu
r
e
ex
tr
ac
tio
n
u
s
in
g
h
is
to
g
r
am
o
f
o
r
ie
n
ted
g
r
ad
ien
ts
(
HOG)
an
d
class
if
ic
atio
n
u
s
in
g
r
an
d
o
m
f
o
r
e
s
t;
th
is
m
eth
o
d
a
ch
iev
es
an
ac
cu
r
ac
y
s
co
r
e
o
f
9
5
.
5
9
%;
h
o
wev
er
,
th
er
e
is
n
o
in
f
o
r
m
atio
n
r
eg
a
r
d
in
g
in
f
er
en
ce
,
d
esp
ite
th
e
f
ac
t
t
h
at
in
f
er
en
ce
is
a
cr
u
cial
m
etr
ic
to
ass
es
s
th
e
alg
o
r
ith
m
’
s
p
er
f
o
r
m
an
ce
;
ad
d
itio
n
ally
,
th
e
m
et
h
o
d
s
e
m
p
lo
y
ed
m
ay
h
a
v
e
s
o
m
e
l
im
itatio
n
s
in
ter
m
s
o
f
p
r
o
ce
s
s
in
g
ea
ch
s
tep
in
d
ep
en
d
en
tly
i
n
r
ea
l
-
tim
e,
o
n
th
e
o
th
er
h
an
d
,
en
c
o
u
n
te
r
v
a
r
io
u
s
p
r
o
b
lem
s
,
s
u
ch
as
o
cc
l
u
s
io
n
a
n
d
d
etec
tio
n
o
f
th
e
in
ap
p
r
o
p
r
iate
s
ig
n
as
tr
a
f
f
i
c
s
ig
n
(
e.
g
.
co
m
m
er
cial
s
ig
n
)
.
Sriv
astav
a
et
a
l.
[
9
]
u
s
ed
YO
L
Ov
4
b
y
ad
ju
s
tin
g
th
e
r
elev
an
t
p
a
r
am
eter
s
an
d
r
e
tr
ain
in
g
th
e
m
o
d
el
o
n
a
m
an
u
ally
an
n
o
tated
d
ata
s
et
to
d
ete
ct
T
S.
T
h
is
m
eth
o
d
attain
ed
an
ac
cu
r
ac
y
o
f
9
0
.
7
%.
T
ian
[
1
0
]
p
r
o
p
o
s
ed
a
n
ew
m
o
d
el
ca
lled
f
ea
tu
r
e
f
u
s
io
n
s
in
g
le
s
h
o
t
d
etec
to
r
(
FFSSD
)
,
wh
ich
im
p
r
o
v
es
th
e
SS
D
alg
o
r
ith
m
[
1
1
]
.
B
y
f
u
s
in
g
f
ea
tu
r
es
f
r
o
m
lo
w
d
im
en
s
io
n
s
in
to
f
ea
tu
r
es
th
at
ar
e
in
h
ig
h
d
im
e
n
s
io
n
s
,
in
s
er
tin
g
th
e
SE
m
o
d
el,
an
d
ad
o
p
ti
n
g
th
e
f
o
ca
l
lo
s
s
to
im
p
r
o
v
e
an
d
f
ix
th
e
o
r
i
g
in
al
lo
s
s
f
u
n
ctio
n
,
th
is
m
eth
o
d
r
e
ac
h
es
a
p
r
ec
is
io
n
o
f
9
2
.
7
9
%.
b
o
th
m
eth
o
d
s
p
r
o
v
id
e
d
n
o
in
f
o
r
m
atio
n
ab
o
u
t
in
f
er
en
ce
.
Selçu
k
et
a
l.
[
1
2
]
u
s
ed
th
e
YOL
Ov
4
n
etwo
r
k
to
d
etec
t
tr
af
f
ic
s
p
ee
d
lim
it
s
ig
n
s
.
I
t
is
ac
h
iev
in
g
an
ac
cu
r
ac
y
o
f
9
5
%
an
d
4
2
f
r
a
m
es
p
er
s
ec
o
n
d
(
FP
S)
.
No
n
et
h
eless
,
th
is
m
eth
o
d
co
v
er
s
o
n
ly
o
n
e
class
o
f
T
S.
Me
g
alin
g
am
et
a
l.
[
1
3
]
p
r
o
p
o
s
es
th
e
r
ef
in
ed
m
ask
r
e
g
io
n
-
b
ased
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
(
m
ask
R
-
C
NN
)
f
o
r
d
etec
tio
n
o
f
tr
af
f
ic
s
ig
n
s
with
an
attain
t
p
r
ec
is
io
n
o
f
9
7
.
0
8
%.
n
o
tin
g
th
at
th
e
ac
cu
r
ac
y
is
h
ig
h
er
u
s
in
g
t
h
e
s
er
ies
o
f
R
-
C
NN,
b
ec
au
s
e
th
ey
u
s
e
two
s
tag
es
f
o
r
d
etec
tio
n
to
g
et
m
o
r
e
ac
c
u
r
ate
r
esu
lts
;
o
th
er
wis
e,
th
e
in
f
er
en
ce
is
n
o
t
s
u
f
f
icien
t
t
o
g
et
r
ea
l
-
tim
e
r
esp
o
n
s
es
in
c
r
itical
s
itu
atio
n
s
.
Xia
et
a
l.
[
1
4
]
p
r
esen
ts
in
h
is
p
a
p
er
DSR
A
-
DE
T
R
,
a
n
o
v
el
ap
p
r
o
ac
h
f
o
c
u
s
ed
o
n
im
p
r
o
v
in
g
m
u
ltis
ca
le
d
etec
tio
n
p
er
f
o
r
m
an
ce
,
th
is
m
eth
o
d
r
ea
ch
es
an
av
er
a
g
e
p
r
ec
is
io
n
o
f
7
6
.
1
3
%
w
h
ich
s
till
b
e
i
m
p
r
o
v
e
d
esp
ec
ially
wh
en
we
ar
e
i
n
a
co
m
p
lex
en
v
ir
o
n
m
en
t o
r
th
e
s
ig
n
s
ar
e
o
cc
u
lted
,
th
e
ac
cu
r
ac
y
will d
ec
r
ea
s
e.
Desp
ite
t
h
e
im
p
r
o
v
em
en
t
m
e
n
ti
o
n
e
d
ab
o
v
e
,
d
ete
cti
n
g
,
a
n
d
r
e
c
o
g
n
i
zi
n
g
t
r
a
f
f
ic
s
i
g
n
s
u
s
i
n
g
o
b
jec
t
d
et
ec
ti
o
n
al
g
o
r
i
t
h
m
s
,
s
ti
ll
f
ac
e
ch
all
en
g
es
i
n
ac
h
i
e
v
i
n
g
a
n
e
f
f
e
cti
v
e
t
r
ad
e
-
o
f
f
b
etw
ee
n
d
et
ec
t
i
o
n
a
n
d
r
ec
o
g
n
iti
o
n
s
p
e
ed
,
an
d
ac
c
u
r
ac
y
,
b
ec
a
u
s
e
o
f
d
if
f
er
en
t
s
h
a
p
es
a
n
d
m
ea
n
i
n
g
s
,
t
o
t
ac
k
l
e
t
h
es
e
co
n
ce
r
n
s
,
w
e
wi
ll
i
n
v
es
ti
g
at
e
i
n
th
is
p
ap
er
h
o
w
to
d
ea
l
wit
h
t
h
e
m
e
n
t
io
n
ed
li
m
its
ab
o
v
e
,
es
p
e
cia
ll
y
w
h
e
n
w
e
a
r
e
in
a
co
m
p
le
x
e
n
v
i
r
o
n
m
e
n
t
,
also
k
e
p
t
t
h
e
r
ea
l
-
t
im
e
r
es
p
o
n
s
e
wh
ic
h
is
n
ee
d
e
d
in
c
r
iti
c
al
s
it
u
ati
o
n
s
a
n
d
d
e
p
l
o
y
m
en
t
in
to
a
n
e
m
b
e
d
d
e
d
p
r
o
ce
s
s
o
r
.
T
h
e
p
r
es
e
n
t
ar
tic
le
s
u
g
g
ests
th
e
u
s
e
o
f
YO
L
O
v
7
[
1
5
]
,
w
h
i
ch
is
b
o
t
h
p
r
e
cise
a
n
d
f
u
n
cti
o
n
s
in
r
ea
l
-
tim
e,
wi
th
an
ch
o
r
b
o
x
a
d
a
p
t
ati
o
n
u
s
i
n
g
k
-
m
ea
n
s
++
[
1
6
]
.
T
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
f
o
r
d
etec
t
io
n
is
tr
ain
ed
o
v
er
o
u
r
p
r
o
p
r
ietar
y
Mo
r
o
cc
o
d
ata
s
et;
an
d
s
p
lit
in
to
5
class
es
b
ased
o
n
th
e
s
h
ap
es
an
d
co
l
o
r
s
to
h
av
e
a
g
o
o
d
g
en
er
atio
n
ab
ilit
y
f
o
r
T
S
d
ete
ctio
n
,
in
ca
s
e
we
h
av
e
im
ag
es
th
at
d
o
n
o
t
b
elo
n
g
to
th
e
tr
ai
n
ed
d
atasets
.
T
h
e
n
,
f
o
r
th
e
p
r
o
h
ib
itiv
e
class
,
a
r
e
co
g
n
itio
n
m
o
d
u
le
b
ased
o
n
C
NN
is
in
teg
r
ated
in
to
t
h
e
YO
L
Ov
7
alg
o
r
ith
m
to
r
ec
o
g
n
ize
th
e
m
ea
n
in
g
o
f
th
e
cu
r
r
en
t
d
etec
ted
p
r
o
h
ib
it si
g
n
.
T
h
e
m
ajo
r
c
o
n
tr
ib
u
tio
n
s
ar
e:
−
Use
th
e
latest
v
er
s
io
n
o
f
th
e
YOL
O
alg
o
r
ith
m
to
tak
e
ad
v
an
tag
e
o
f
th
ei
r
am
elio
r
atio
n
as
a
o
n
e
-
s
tag
e
d
etec
to
r
.
−
T
r
ain
th
e
YOL
O
v
7
al
g
o
r
ith
m
o
n
Mo
r
o
cc
an
h
i
g
h
way
s
ce
n
e
d
atasets
.
−
Sp
lit th
e
d
atasets
in
to
5
class
e
s
d
er
iv
ed
f
r
o
m
th
eir
s
h
ap
es a
n
d
co
lo
r
s
f
o
r
p
r
ec
is
e
d
etec
tio
n
.
−
Ad
ap
t th
e
an
c
h
o
r
b
o
x
u
s
in
g
k
-
m
ea
n
s
++
alg
o
r
ith
m
.
−
Use
th
e
tin
y
v
er
s
io
n
o
f
YO
L
Ov
7
to
p
e
r
f
o
r
m
r
ea
l
-
tim
e
d
etec
tio
n
th
at
is
r
eliab
le
f
o
r
d
ep
lo
y
m
en
t
in
em
b
ed
d
e
d
h
ar
d
war
e.
−
E
x
p
lo
r
e
th
e
r
esu
lt
o
f
YOL
Ov
7
,
ad
d
a
r
ec
o
g
n
itio
n
m
o
d
u
le
f
o
r
th
e
p
r
o
h
ib
ito
r
y
s
ig
n
s
class
,
wh
ich
is
th
e
cr
itical
ca
teg
o
r
y
th
at
ca
n
lim
it
r
o
ad
s
ce
n
e
ac
cid
e
n
ts
.
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:
2088
-
8
7
0
8
Tr
a
ffic sig
n
s
d
etec
tio
n
a
n
d
p
r
o
h
ib
ito
r
s
ig
n
s
r
ec
o
g
n
itio
n
i
n
Mo
r
o
cc
o
r
o
a
d
s
ce
n
e
(
I
ma
n
e
T
a
o
u
q
i
)
6315
T
h
e
f
o
llo
win
g
s
ec
tio
n
s
o
f
th
i
s
p
ap
er
a
r
e
s
tr
u
ctu
r
ed
as
f
o
ll
o
ws:
First
s
ec
tio
n
is
in
tr
o
d
u
ct
io
n
.
I
n
th
e
s
ec
o
n
d
s
ec
tio
n
,
th
e
p
r
o
p
o
s
ed
m
eth
o
d
,
with
a
d
escr
ip
tio
n
o
f
th
e
u
s
ed
alg
o
r
ith
m
,
m
o
d
e
l
ar
ch
itectu
r
e,
a
n
d
m
eth
o
d
o
l
o
g
y
.
T
h
e
th
ir
d
s
ec
tio
n
p
r
esen
ts
a
n
ex
p
er
im
en
t
ev
al
u
atio
n
to
d
is
cu
s
s
d
atasets
,
ev
a
lu
atio
n
in
d
icato
r
s
,
an
d
test
r
esu
lts
u
s
in
g
d
if
f
er
en
t
p
ar
am
eter
s
an
d
en
v
i
r
o
n
m
e
n
t
co
m
p
lex
ity
.
A
t
th
e
en
d
,
a
co
n
clu
s
io
n
is
p
r
o
v
id
e
d
to
r
ec
ap
itu
late
th
is
p
ap
e
r
.
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
to
d
etec
t
an
d
r
ec
o
g
n
ize
tr
a
f
f
ic
s
ig
n
s
is
ac
co
m
p
lis
h
ed
in
th
r
ee
m
ajo
r
s
tep
s
:
f
ir
s
t,
th
e
TS
d
etec
tio
n
m
o
d
el
h
as
b
ee
n
b
u
ilt
u
s
in
g
th
e
p
r
o
p
o
s
ed
a
lg
o
r
ith
m
YOL
Ov
7
with
k
-
m
e
an
s
++
ad
ap
tatio
n
to
r
aise
th
e
ac
c
u
r
ac
y
of
th
e
al
g
o
r
ith
m
,
th
en
an
o
th
er
m
o
d
el
was
ad
d
e
d
,
so
it
will
be
r
esp
o
n
s
ib
le
f
o
r
TS
r
ec
o
g
n
itio
n
an
d
class
if
icatio
n
ac
co
r
d
i
n
g
to
t
h
e
m
ea
n
in
g
of
th
e
c
u
r
r
en
t
d
etec
ted
s
ig
n
,
an
d
f
in
ally
th
e
ass
o
ciatio
n
of
d
etec
tio
n
a
n
d
r
ec
o
g
n
itio
n
m
o
d
els.
Fig
u
r
e
1
d
escr
ib
es
th
e
wo
r
k
f
l
o
w
of
t
h
e
p
r
o
p
o
s
ed
s
y
s
tem
,
wh
er
e
th
e
in
p
u
t
im
ag
e
tak
en
f
r
o
m
th
e
r
o
ad
s
ce
n
e
en
v
ir
o
n
m
en
t
is
f
ed
to
t
h
e
YOL
Ov
7
n
et
wo
r
k
to
lo
ca
lize
t
h
e
r
elev
an
t
TS
by
a
b
o
u
n
d
in
g
box
.
A
d
if
f
e
r
en
t
s
ig
n
class
m
ay
be
d
etec
ted
d
e
p
en
d
i
n
g
on
th
e
r
o
ad
s
ce
n
e
s
ig
n
(
e
.
g
.
m
an
d
ato
r
y
an
d
war
n
in
g
)
,
if
th
e
d
etec
ted
s
ig
n
s
ar
e
ass
o
ciat
ed
with
p
r
o
h
ib
ito
r
y
class
,
it
will
be
cr
o
p
p
ed
an
d
f
ee
d
to
th
e
s
ec
o
n
d
tr
ain
e
d
n
et
wo
r
k
to
r
ec
o
g
n
ize
th
e
m
ea
n
i
n
g
of
c
u
r
r
en
t
d
etec
ted
cr
itical
p
r
o
h
ib
it
s
ig
n
.
Fig
u
r
e
1
.
W
o
r
k
f
lo
w
o
f
T
S d
et
ec
tio
n
an
d
r
ec
o
g
n
itio
n
s
y
s
tem
3.
M
E
T
H
O
D
3
.
1
.
YO
L
O
v
7
a
lg
o
rit
h
m
YOL
O
n
etwo
r
k
co
m
p
r
is
es
f
o
u
r
m
ain
s
ec
tio
n
s
:
in
p
u
t,
b
ac
k
b
o
n
e,
n
ec
k
,
an
d
h
ea
d
,
f
o
llo
w
ed
by
th
e
p
r
ed
ictio
n
lay
er
[
1
7
]
.
T
h
e
in
p
u
t
lay
er
in
YOL
Ov
7
is
r
es
p
o
n
s
ib
le
f
o
r
r
ec
eiv
in
g
th
e
im
ag
e
or
v
id
e
o
to
be
p
r
o
ce
s
s
ed
an
d
en
s
u
r
in
g
u
n
if
o
r
m
r
esizin
g
of
th
e
in
p
u
t
im
ag
e
to
640
×
6
4
0
×
3,
th
u
s
m
ee
tin
g
th
e
s
p
ec
if
ied
in
p
u
t
s
ize
cr
iter
ia
f
o
r
th
e
b
ac
k
b
o
n
e,
wh
ich
is
th
e
s
ec
o
n
d
p
ar
t
of
t
h
e
n
etwo
r
k
,
wh
ich
e
x
tr
ac
ts
cr
u
cial
f
ea
tu
r
es
f
r
o
m
th
e
in
p
u
t
im
ag
e
.
th
e
b
ac
k
b
o
n
e
u
s
ed
in
th
e
YOL
Ov
7
alg
o
r
ith
m
is
b
ased
on
Dar
k
n
et
-
5
3
,
wh
ich
is
a
d
ee
p
co
n
v
o
l
u
tio
n
n
etwo
r
k
with
53
lay
er
s
.
It
u
s
es
r
esid
u
al
b
l
o
ck
s
to
f
ac
ilit
ate
d
ee
p
lear
n
in
g
an
d
av
o
id
th
e
d
is
ap
p
ea
r
ed
g
r
ad
ien
t
p
r
o
b
lem
.
T
h
e
b
ac
k
b
o
n
e
tak
es
an
in
p
u
t
im
ag
e
an
d
p
ass
es
it
th
r
o
u
g
h
s
ev
er
al
co
n
v
o
lu
tio
n
lay
er
s
to
ex
tr
ac
t
f
ea
tu
r
es
f
r
o
m
d
if
f
er
en
t
s
ca
les.
T
h
en
YOL
Ov
7
u
s
es
a
n
ec
k
to
co
m
p
ile
th
e
r
etr
iev
ed
f
ea
tu
r
e
m
ap
s
an
d
b
u
ild
s
f
e
atu
r
e
p
y
r
a
m
id
s
to
m
er
g
e
th
e
ex
tr
ac
ted
f
ea
tu
r
es
at
d
if
f
e
r
en
t
s
ca
les.
T
h
e
n
ec
k
co
n
s
is
ts
of
s
ev
er
al
co
n
v
o
lu
tio
n
an
d
ag
g
r
eg
atio
n
lay
er
s
,
wh
ich
allo
w
co
n
tex
tu
al
in
f
o
r
m
atio
n
f
r
o
m
d
if
f
er
en
t
s
p
atial
r
eso
lu
tio
n
s
to
be
i
n
teg
r
ated
.
O
n
ce
th
e
f
ea
tu
r
es
h
av
e
b
ee
n
m
e
r
g
ed
by
th
e
n
ec
k
[
1
8
]
,
th
ey
ar
e
tr
an
s
m
itted
to
th
e
"h
ea
d
"
of
th
e
n
etwo
r
k
to
p
r
ed
ict
b
o
u
n
d
i
n
g
b
o
x
es
an
d
co
r
r
es
p
o
n
d
in
g
o
b
ject
class
es.
T
h
e
h
ea
d
is
co
m
p
o
s
ed
of
s
ev
er
al
co
n
v
o
l
u
tio
n
al
an
d
lin
ea
r
lay
er
s
th
at
g
en
er
ate
th
e
f
in
al
p
r
ed
ictio
n
s
[
1
9
]
.
T
h
is
last
is
p
er
f
o
r
m
ed
u
s
in
g
an
ch
o
r
s
an
d
g
r
id
s
.
An
c
h
o
r
s
ar
e
p
r
ed
ef
in
e
d
b
o
u
n
d
in
g
b
o
x
es
of
d
if
f
e
r
en
t
s
izes
an
d
s
h
ap
es
[
2
0
]
,
wh
ich
ar
e
u
s
ed
to
p
r
ed
ict
th
e
co
o
r
d
in
ates
of
b
o
u
n
d
in
g
b
o
x
es.
Gr
id
s
d
iv
id
e
t
h
e
im
ag
e
in
to
s
ev
er
al
r
e
g
io
n
s
,
an
d
each
r
e
g
io
n
is
r
esp
o
n
s
ib
le
f
o
r
p
r
ed
ictin
g
t
h
e
class
es
of
o
b
jects
p
r
esen
t
in
th
at
r
eg
io
n
.
Fo
r
each
g
r
id
,
th
e
n
etwo
r
k
g
e
n
er
ates
p
r
ed
ictio
n
s
f
o
r
s
ev
er
al
p
o
te
n
tial
b
o
u
n
d
in
g
b
o
x
es,
u
s
in
g
lin
ea
r
r
eg
r
ess
io
n
s
to
p
r
e
d
ict
th
e
co
o
r
d
in
ates
(
x
,
y,
wid
th
,
h
eig
h
t)
of
ea
c
h
b
o
u
n
d
in
g
b
o
x
r
elativ
e
to
th
e
co
r
r
esp
o
n
d
in
g
an
ch
o
r
.
T
h
en
,
a
So
f
tMa
x
ac
tiv
atio
n
f
u
n
ctio
n
[
2
1
]
is
ap
p
lied
to
t
h
e
n
etwo
r
k
o
u
tp
u
ts
to
g
et
th
e
p
r
o
b
a
b
ilit
ies
of
th
e
d
if
f
e
r
en
t
class
es
of
o
b
jects.
Fin
ally
,
a
co
n
f
id
en
ce
th
r
esh
o
ld
is
ap
p
lied
to
p
r
ed
ictio
n
s
to
f
ilter
d
etec
tio
n
s
with
co
n
f
id
en
ce
less
th
an
a
p
r
ed
ef
in
e
d
th
r
esh
o
ld
.
T
h
e
r
e
m
ain
in
g
d
etec
tio
n
s
ar
e
th
e
n
g
r
o
u
p
e
d
,
an
d
non
-
m
ax
im
u
m
s
u
p
p
r
ess
io
n
is
ap
p
lied
to
elim
in
ate
r
ed
u
n
d
an
t
or
o
v
er
lap
p
in
g
d
etec
tio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
3
1
3
-
6
3
2
1
6316
3
.
2
.
Ancho
rs
bo
x
a
da
pta
t
io
ns
us
i
ng
k
-
m
ea
ns
+
+
a
lg
o
rit
hm
E
x
is
tin
g
an
ch
o
r
b
o
x
es
th
at
a
r
e
u
s
ed
in
YOL
Ov
7
with
C
OC
O
d
ataset
s
ar
e
not
s
u
itab
le
f
o
r
o
u
r
d
ataset,
so
b
ef
o
r
e
tr
ai
n
in
g
o
u
r
d
atasets
u
s
in
g
th
e
YOL
Ov
7
n
etwo
r
k
,
it
i
s
r
elev
an
t
to
p
er
f
o
r
m
k
-
m
ea
n
s
++
clu
s
ter
in
g
,
to
an
aly
ze
a
n
d
s
elec
t
th
e
th
r
ee
m
o
s
t
s
u
itab
le
an
ch
o
r
b
o
x
es
f
o
r
ea
c
h
s
ca
le
an
d
f
o
r
m
at
to
en
h
an
ce
m
o
d
el
ac
cu
r
ac
y
.
T
h
e
f
o
llo
win
g
p
o
in
ts
d
escr
ib
e
t
h
e
p
er
f
o
r
m
e
d
s
tep
s
by
k
-
m
ea
n
s
++
alg
o
r
ith
m
:
−
I
n
itializatio
n
:
r
an
d
o
m
ly
s
elec
t
k
p
o
in
ts
in
th
e
d
ata
s
p
ac
e
as
th
e
in
itial
clu
s
ter
ce
n
ter
s
.
−
Po
in
ts
allo
ca
tio
n
:
each
d
ata
p
o
in
t
is
ass
ig
n
ed
to
th
e
clu
s
ter
with
th
e
clo
s
est
ce
n
ter
.
T
h
is
is
u
s
u
ally
b
ased
on
th
e
E
u
clid
ea
n
d
is
tan
ce
b
etwe
en
th
e
p
o
i
n
t
an
d
t
h
e
ce
n
ter
of
th
e
clu
s
ter
.
−
Up
d
ate
clu
s
ter
ce
n
ter
s
:
r
ec
al
cu
late
th
e
ce
n
ter
s
of
ea
c
h
c
lu
s
ter
by
tak
in
g
th
e
av
er
a
g
e
of
th
e
p
o
in
ts
b
elo
n
g
in
g
to
th
is
clu
s
ter
.
−
R
ep
ea
t
s
tep
s
2
an
d
3:
r
ep
ea
t
th
e
s
tep
s
f
o
r
ass
ig
n
in
g
p
o
in
ts
an
d
u
p
d
atin
g
ce
n
ter
s
u
n
til
th
e
clu
s
ter
ce
n
ter
s
co
n
v
er
g
e
to
a
s
tab
le
p
o
s
itio
n
.
−
C
o
n
v
er
g
en
ce
:
t
h
e
alg
o
r
ith
m
c
o
n
v
er
g
es
wh
en
clu
s
ter
ce
n
ter
s
no
lo
n
g
er
ch
a
n
g
e
or
ch
a
n
g
e
n
e
g
lig
ib
ly
[
2
2
]
.
At
th
e
en
d
of
th
e
alg
o
r
ith
m
,
each
d
ata
p
o
in
t
is
ass
ig
n
ed
to
a
s
p
ec
if
ic
clu
s
ter
,
f
o
r
m
in
g
s
ep
ar
ate
g
r
o
u
p
s
.
C
lu
s
ter
ce
n
ter
s
r
ep
r
es
en
t
th
e
ce
n
tr
al
ch
ar
ac
ter
is
tics
of
each
clu
s
ter
.
On
ce
we
h
av
e
th
e
ce
n
ter
s
of
each
p
ar
titi
o
n
in
g
,
we
class
if
y
th
em
to
g
r
o
u
p
th
em
in
to
s
m
all,
m
ed
iu
m
,
an
d
lar
g
e
s
ets
of
An
ch
o
r
.
T
h
is
g
iv
es
us
th
r
ee
an
ch
o
r
s
:
−
[
4
.
1
1
9
3
9
0
3
8
,
3
.
9
0
4
4
1
7
1
2
,
7
.
0
0
4
6
6
3
9
9
,
1
1
.
5
3
3
2
4
8
4
,
1
3
.
5
1
5
9
5
8
6
7
,
5
.
3
8
8
0
1
8
1
2
]
−
[
1
5
.
1
3
0
9
9
5
7
6
,
1
7
.
6
6
1
1
1
7
8
5
,
2
0
.
2
6
7
4
3
3
2
6
,
9
.
8
4
2
9
7
6
4
8
,
2
5
.
7
6
2
8
7
1
9
5
,
2
3
.
6
6
2
7
8
6
0
7
]
−
[
2
7
.
3
3
8
7
5
2
0
7
,
1
3
.
4
2
2
4
6
1
8
9
,
3
4
.
8
6
6
9
6
3
7
1
,
1
8
.
5
3
4
4
7
4
6
,
4
2
.
6
2
1
5
7
5
0
9
,
2
3
.
6
7
6
4
7
8
0
3
]
.
3
.
3
.
Sequ
ent
ia
l
m
o
del
f
o
r
pro
hib
it
o
ry
s
ig
ns
re
co
g
nitio
n
To
r
ec
o
g
n
ize
d
if
f
er
e
n
t
s
ig
n
s
of
p
r
o
h
ib
ited
class
,
we
u
s
ed
a
s
eq
u
en
tial
m
o
d
el
f
o
r
o
b
ject
r
e
co
g
n
itio
n
.
It
i
s
a
m
ac
h
in
e
lea
r
n
in
g
m
o
d
e
l
th
at
p
r
o
ce
s
s
es
d
ata
s
eq
u
e
n
tially
,
ty
p
ically
u
s
ed
in
r
ec
o
g
n
izin
g
o
b
jects
with
in
im
ag
es.
T
h
e
g
o
al
of
o
b
ject
r
e
co
g
n
itio
n
is
to
id
en
tify
an
d
cl
ass
if
y
o
b
jects
or
p
atter
n
s
with
in
th
e
v
is
u
al
d
ata
.
T
h
e
ar
ch
itectu
r
e
u
s
ed
is
d
escr
ib
ed
in
Fig
u
r
e
2
s
h
o
win
g
lay
er
ty
p
e,
o
u
tp
u
t
s
h
ap
e,
an
d
n
u
m
b
e
r
of
p
a
r
am
eter
s
.
Fig
u
r
e
2
.
C
NN
ar
ch
itectu
r
e
f
o
r
tr
af
f
ic
s
ig
n
class
if
icatio
n
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:
2088
-
8
7
0
8
Tr
a
ffic sig
n
s
d
etec
tio
n
a
n
d
p
r
o
h
ib
ito
r
s
ig
n
s
r
ec
o
g
n
itio
n
i
n
Mo
r
o
cc
o
r
o
a
d
s
ce
n
e
(
I
ma
n
e
T
a
o
u
q
i
)
6317
3
.
4
.
T
ra
ini
ng
s
et
u
p a
nd
da
t
a
s
et
s
All
test
s
in
th
is
p
ap
er
wer
e
ex
ec
u
ted
u
n
d
er
an
i
d
en
tical
ex
p
er
im
en
tal
en
v
ir
o
n
m
en
t:
C
o
lab
Go
o
g
le
to
tr
ain
an
d
test
th
e
YOL
Ov
7
,
ti
n
y
YOL
Ov
7
d
etec
to
r
u
s
in
g
G
PU
(
Nv
id
ia
SMI
,
C
u
d
a
v
er
s
io
n
12
to
r
ch
2
.
0
.
1
+
cu
1
1
8
C
UDA:0
,
T
esla
T
4
)
.
we
u
s
ed
Ger
m
a
n
tr
a
f
f
ic
s
ig
n
d
e
tectio
n
b
en
c
h
m
ar
k
(
GT
SDB
)
d
ataset
[
2
3
]
;
g
iv
en
th
at
m
o
s
t
of
th
eir
r
o
ad
s
ig
n
s
a
lig
n
with
th
o
s
e
u
s
ed
in
M
o
r
o
c
ca
n
TS
[
2
4
]
.
T
h
en
,
we
ad
a
p
t
t
h
e
d
ata
c
o
n
tain
in
g
842
im
ag
es,
by
r
em
o
v
in
g
im
ag
es
th
at
do
not
co
r
r
esp
o
n
d
to
d
ata
f
r
o
m
Mo
r
o
cc
o
an
d
ad
d
in
g
o
th
er
d
if
f
er
en
t
im
ag
es
ca
p
tu
r
ed
in
th
e
M
o
r
o
cc
an
r
o
a
d
s
ce
n
e
in
d
if
f
er
e
n
t
wea
th
er
co
n
d
itio
n
s
.
We
d
iv
id
ed
th
e
s
ig
n
s
in
to
5
class
es
:
s
to
p
,
m
an
d
ato
r
y
,
y
i
eld
,
war
n
in
g
,
an
d
p
r
o
h
ib
ito
r
y
.
We
u
s
e
R
o
b
o
f
lo
w
to
lab
el
TS
d
atasets
.
Su
b
s
eq
u
en
tly
,
d
ata
a
u
g
m
e
n
tatio
n
tech
n
iq
u
es
wer
e
em
p
lo
y
e
d
on
th
e
la
b
eled
d
atasets
,
to
f
in
ally
h
av
e
2
,
1
6
3
im
ag
es
f
o
r
tr
ain
i
n
g
.
Fo
r
th
e
r
ec
o
g
n
itio
n
m
o
d
u
le,
we
u
s
ed
Ger
m
an
tr
a
f
f
ic
s
ig
n
r
ec
o
g
n
itio
n
b
en
ch
m
a
r
k
(
GT
SR
B
)
d
atasets
.
th
at
co
n
tain
m
o
r
e
t
h
an
40
cla
s
s
es,
we
ad
ap
ted
it
f
o
r
o
n
ly
p
r
o
h
ib
ito
r
y
s
ig
n
s
by
d
eletin
g
u
n
wan
ted
class
es,
to
h
av
e
17
class
es
at
th
e
en
d
wit
h
th
e
f
o
llo
win
g
n
am
es:
‘
Sp
ee
d
lim
it
20
km
p
er
hour
’
,
‘
Sp
e
ed
lim
it
30
km
p
er
hour
’
,
‘
Sp
ee
d
lim
it
50
km
p
e
r
hour
’
,
‘
Sp
ee
d
lim
it
60
km
p
er
h
o
u
r
’
,
‘
Sp
ee
d
lim
it
70
km
p
er
h
o
u
r
’
,
‘
Sp
ee
d
lim
it
80
km
p
er
h
o
u
r
’
,
‘
E
n
d
of
s
p
ee
d
lim
it
80
km
p
er
h
’
,
‘
Sp
ee
d
lim
it
1
0
0
km
p
er
h
o
u
r
’
,
‘
S
p
ee
d
lim
it
120
km
p
er
h
o
u
r
’
,
‘
No
p
ass
in
g
’
,
‘
No
p
ass
in
g
v
eh
icles
o
v
e
r
3
.
5
to
n
s
’
,
‘
Sto
p
’
,
‘
No
v
eh
icles
’
,
‘
Ve
h
icles
o
v
er
3
.
5
t
o
n
s
p
r
o
h
ib
ited
’
,
‘
No
en
tr
y
’
,
‘
E
n
d
s
p
ee
d
an
d
p
ass
in
g
lim
its
’
,
‘
E
n
d
of
no
p
ass
in
g
’
,
‘
E
n
d
no
p
ass
in
g
v
e
h
icles
o
v
er
3
.
5
to
n
s
’
,
to
r
ec
o
g
n
ize
a
n
d
ex
tr
ac
t
in
f
o
r
m
atio
n
f
r
o
m
p
r
o
h
ib
itiv
e
class
.
T
h
e
tr
ain
in
g
p
ar
am
eter
s
s
etu
p
of
th
e
ex
p
e
r
im
en
t
’
s
tr
ain
in
g
p
r
o
c
ed
u
r
e
of
YOL
Ov
7
d
etec
tio
n
a
n
d
s
eq
u
e
n
tial
m
o
d
el
f
o
r
r
ec
o
g
n
itio
n
ar
e
s
h
o
wn
in
T
ab
le
s
1
an
d
2
r
esp
ec
tiv
ely
wi
th
th
eir
v
alu
es.
T
ab
le
1
.
T
r
ai
n
in
g
p
ar
am
eter
s
f
o
r
YOL
Ov
7
P
a
r
a
me
t
e
r
V
a
l
u
e
Ep
o
c
h
s
1
6
0
W
e
i
g
h
t
_
d
e
c
a
y
0
.
0
0
0
5
M
o
me
n
t
u
m
0
.
9
3
7
Le
a
r
n
i
n
g
r
a
t
e
0
.
0
1
B
a
t
c
h
si
z
e
16
I
mag
e
si
z
e
6
4
0
×
6
4
0
×
3
W
o
r
k
e
r
s
8
T
ab
le
2
.
T
r
ai
n
in
g
p
ar
am
eter
s
f
o
r
r
ec
o
g
n
itio
n
P
a
r
a
me
t
e
r
V
a
l
u
e
Ep
o
c
h
s
30
B
a
t
c
h
_
s
i
z
e
32
O
p
t
i
mi
z
e
r
s
A
d
a
m
I
mag
e
si
z
e
3
0
×
3
0
×
3
Le
a
r
n
i
n
g
r
a
t
e
0
.
0
0
1
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
4
.
1
.
E
v
a
lua
t
io
n ind
ica
t
o
rs:
In
th
is
p
ap
e
r
,
th
r
ee
m
etr
ics
r
ec
all,
ac
cu
r
ac
y
,
an
d
m
ea
n
a
v
er
ag
e
p
r
ec
is
io
n
(
m
AP)
ar
e
e
m
p
lo
y
ed
to
ass
es
s
th
e
m
o
d
el
’
s
ac
cu
r
ac
y
,
wh
ile
m
o
d
el
s
ize
an
d
FPS
ar
e
u
s
ed
to
ev
alu
ate
th
e
m
o
d
el
’
s
s
p
ee
d
[
2
5
]
.
−
Acc
u
r
ac
y
(
P):
Acc
u
r
ac
y
is
th
e
d
eg
r
ee
of
p
r
ec
is
io
n
of
th
e
m
o
d
el
in
i
d
en
tify
in
g
o
n
ly
r
el
ev
an
t
o
b
jects.
It
co
r
r
esp
o
n
d
s
to
th
e
r
atio
of
tr
u
e
p
o
s
itiv
es
(
T
P)
to
all
d
etec
tio
n
s
m
ad
e
by
th
e
m
o
d
el
[
2
5
]
.
=
+
−
R
ec
all:
T
h
e
r
ec
all
m
ea
s
u
r
es
th
e
p
er
ce
n
tag
e
of
co
r
r
ec
t
d
etec
tio
n
s
ag
ain
s
t
all
in
s
tan
ce
s
of
r
ea
l
o
b
jects.
It
is
a
m
ea
s
u
r
e
of
th
e
q
u
ality
of
b
o
t
h
p
o
s
itiv
e
a
n
d
n
eg
ativ
e
p
r
ed
i
ctio
n
s
.
T
h
e
h
ig
h
er
th
e
r
ec
all,
th
e
f
ewe
r
f
alse
n
eg
ativ
es
th
er
e
ar
e
[
2
6
]
.
=
+
−
I
n
f
er
en
ce
or
FPS:
I
n
f
er
en
ce
m
ea
s
u
r
es
th
e
tim
e
tak
en
to
p
er
f
o
r
m
d
etec
tio
n
on
a
s
in
g
l
e
im
ag
e.
It
is
a
m
ea
s
u
r
e
of
th
e
s
p
ee
d
of
th
e
d
e
tectio
n
alg
o
r
ith
m
.
A
m
in
o
r
in
f
er
en
ce
tim
e
is
u
s
u
ally
p
r
ef
e
r
ab
le
f
o
r
r
ea
l
-
tim
e
ap
p
licatio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
3
1
3
-
6
3
2
1
6318
−
Mean
av
er
a
g
e
p
r
ec
is
io
n
-
50
(
m
AP5
0
)
:
co
n
s
id
er
s
b
o
th
ac
c
u
r
ac
y
a
n
d
r
ec
all
to
r
ef
lect
m
o
d
el
p
er
f
o
r
m
an
ce
.
T
h
is
m
etr
ic
is
o
b
tain
ed
by
ca
lcu
latin
g
th
e
m
ea
n
ac
cu
r
ac
y
(
AP)
at
th
e
in
ter
s
ec
tio
n
o
v
er
u
n
io
n
(
I
OU
)
th
r
esh
o
ld
50
f
o
r
each
ca
te
g
o
r
y
,
an
d
th
e
n
av
er
a
g
in
g
all
AP
v
a
lu
es
ac
r
o
s
s
all
ca
teg
o
r
ies
[
2
6
]
.
AP
is
d
er
iv
ed
by
co
m
p
u
tin
g
t
h
e
ar
ea
u
n
d
er
each
p
r
ec
is
io
n
-
r
ec
all
(P
-
R)
cu
r
v
e,
with
p
r
ec
is
io
n
p
lo
tted
on
th
e
x
-
ax
is
an
d
r
ec
all
on
t
h
e
y
-
ax
is
.
T
h
e
f
o
r
m
u
las
f
o
r
ca
lc
u
latin
g
AP
an
d
m
AP
ar
e
as:
=
∫
(
)
1
0
,
=
∑
=
1
Her
e,
(
)
r
ep
r
esen
ts
th
e
p
r
ec
is
io
n
-
r
ec
all
cu
r
v
e
a
n
d
d
en
o
tes
th
e
n
u
m
b
er
of
ca
te
g
o
r
ies.
4
.
2
.
Resul
t
s
d
is
cu
s
s
io
n
T
h
is
s
tu
d
y
in
v
esti
g
ated
a
r
elev
an
t
d
etec
tio
n
an
d
r
ec
o
g
n
i
tio
n
alg
o
r
ith
m
o
v
er
d
if
f
er
e
n
t
class
e
s
,
en
v
ir
o
n
m
en
t
s
ev
er
ity
,
an
d
o
cc
u
lted
tr
af
f
ic
s
ig
n
s
,
th
at
can
p
r
o
f
icien
tly
b
alan
ce
th
e
s
p
ee
d
an
d
ac
cu
r
ac
y
of
th
e
m
o
d
el.
W
h
ile
ea
r
lier
s
tu
d
ies
h
av
e
ex
p
lo
r
ed
TS
d
etec
tio
n
on
lim
ited
class
es
an
d
f
o
cu
s
ed
on
ac
c
u
r
ac
y
r
ath
er
th
an
r
ea
l
-
tim
e
d
etec
tio
n
,
also
s
o
m
e
r
esear
c
h
f
o
cu
s
es
o
n
ly
on
d
etec
tio
n
or
r
ec
o
g
n
itio
n
s
ep
ar
ately
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
in
th
is
s
tu
d
y
ten
d
ed
to
u
s
e
an
ef
f
icien
t
alg
o
r
ith
m
th
at
co
m
b
in
es
b
o
th
ac
c
u
r
ac
y
an
d
r
ea
l
-
tim
e
p
r
o
ce
s
s
in
g
.
An
d
b
o
th
d
etec
tio
n
an
d
r
ec
o
g
n
itio
n
of
t
r
af
f
ic
s
ig
n
s
.
Ou
r
s
tu
d
y
s
u
g
g
ests
th
at
th
e
u
s
e
of
k
-
m
ea
n
s
++
in
cr
ea
s
es
th
e
ac
cu
r
ac
y
of
th
e
alg
o
r
ith
m
to
g
en
er
ate
r
elev
a
n
t
an
ch
o
r
b
o
x
es
a
n
d
t
h
en
u
s
e
a
tin
y
v
er
s
io
n
of
YOL
Ov
7
to
tak
e
its
r
ea
l
-
ti
m
e
p
r
o
ce
s
s
in
g
a
d
v
an
ta
g
e,
af
ter
th
at
if
th
e
d
etec
ted
class
is
p
r
o
h
ib
ito
r
y
,
t
h
e
r
ec
o
g
n
itio
n
m
o
d
u
le
will
s
p
ec
if
y
wh
at
th
e
p
r
o
h
ib
ited
s
ig
n
m
ea
n
s
.
Alth
o
u
g
h
th
e
tr
ain
in
g
of
t
h
e
r
ec
o
g
n
itio
n
m
o
d
u
le
d
escr
ib
e
d
in
Fig
u
r
e
2
u
s
in
g
th
e
p
ar
am
eter
of
T
ab
le
2.
g
iv
es
an
ac
cu
r
ac
y
of
9
9
%.
Fo
r
th
e
d
etec
tio
n
m
o
d
u
le,
d
if
f
er
en
t
ex
p
er
im
en
t
s
wer
e
co
n
d
u
cted
u
s
in
g
two
b
ac
k
b
o
n
e
n
etwo
r
k
s
,
to
d
e
m
o
n
s
tr
ate
an
d
co
m
p
ar
e
th
e
b
ac
k
b
o
n
e
n
etwo
r
k
’
s
ef
f
ec
t
s
on
d
etec
tio
n
ac
cu
r
ac
y
an
d
s
p
ee
d
.
th
e
YOL
Ov
7
b
ac
k
b
o
n
e
n
etwo
r
k
wh
ic
h
h
as
459
lay
e
r
s
,
an
d
t
h
e
tin
y
YO
L
O
v7
b
ac
k
b
o
n
e
n
etwo
r
k
wh
i
ch
co
n
tain
s
o
n
ly
2
6
3
lay
e
r
s
,
also
th
e
ef
f
ec
t
of
an
ch
o
r
s
box
ad
ap
tatio
n
u
s
in
g
k
-
m
ea
n
s
++
is
test
ed
.
T
ab
le
3
d
is
p
lay
s
th
e
o
u
tc
o
m
es
of
th
e
e
x
p
er
im
en
ts
.
T
a
b
le
4
s
h
o
ws
a
co
m
p
ar
is
o
n
b
etwe
en
th
e
p
r
o
p
o
s
ed
m
eth
o
d
co
n
d
u
cted
in
th
is
p
ap
e
r
an
d
o
th
er
s
tate
-
of
-
th
e
-
ar
t
TS
d
etec
tio
n
.
E
x
p
er
im
en
ts
s
h
o
we
d
th
at
th
e
u
s
e
of
tin
y
YOL
Ov
7
with
k
-
m
ea
n
s
++
,
tr
ain
ed
on
au
g
m
en
t
ed
d
atasets
th
at
co
n
s
id
er
d
if
f
er
e
n
t
s
itu
atio
n
s
(
b
lu
r
r
in
g
,
co
m
p
lex
en
v
ir
o
n
m
en
ts
)
,
not
o
n
ly
im
p
r
o
v
ed
th
e
ac
cu
r
ac
y
of
th
e
o
r
ig
in
al
YOL
Ov
7
alg
o
r
ith
m
b
u
t
also
s
ig
n
if
ican
tly
in
cr
ea
s
ed
th
e
s
p
ee
d
of
th
e
m
o
d
el,
r
e
d
u
ce
d
m
em
o
r
y
u
s
ag
e,
an
d
m
et
th
e
d
e
m
an
d
r
eq
u
ir
e
m
en
t
f
o
r
r
ea
l
-
tim
e
g
estu
r
e
r
e
co
g
n
itio
n
.
Als
o
,
by
co
m
p
ar
in
g
our
m
eth
o
d
to
th
e
p
r
ev
io
u
s
r
esear
ch
m
en
tio
n
ed
in
T
ab
le
4,
It
can
be
s
ee
n
t
h
a
t
th
e
u
s
e
of
YOL
Ov
7
with
k
-
m
ea
n
s
++
to
ad
ap
t
an
ch
o
r
b
o
x
c
o
o
r
d
in
ates
in
cr
e
ases
th
e
ac
cu
r
ac
y
d
etec
tio
n
to
9
7
%
co
m
p
ar
ed
to
all
m
en
tio
n
ed
m
et
h
o
d
s
in
T
ab
le
4,
alt
h
o
u
g
h
em
p
lo
y
in
g
a
tin
y
n
etwo
r
k
as
th
e
f
o
u
n
d
ati
o
n
al
b
ac
k
b
o
n
e
g
r
ea
tl
y
r
ed
u
ce
d
th
e
m
o
d
el
s
ize
to
1
2
.
3
MB
an
d
in
f
er
e
n
ce
to
1
1
.
2
(
m
s
)
co
m
p
ar
e
d
to
YOL
Ov
4
an
d
f
aster
R
-
C
NN
[
2
7
]
.
Ho
wev
er
,
T
ab
le
5
s
h
o
ws
more
d
etails
ab
o
u
t
th
e
r
esu
lts
of
th
e
p
r
o
p
o
s
ed
T
in
y
YOL
Ov
7
alg
o
r
ith
m
in
th
e
f
iv
e
ca
teg
o
r
ies
of
T
S.
T
h
e
r
esu
lts
s
h
o
w
th
at
em
p
lo
y
in
g
tin
y
YOL
Ov
7
in
our
co
n
tex
t
is
more
ac
c
u
r
ate
r
eg
ar
d
in
g
its
ex
ec
u
tio
n
tim
e.
T
h
u
s
,
e
m
b
ed
d
ed
on
a
v
e
h
icle
ca
lcu
lato
r
can
p
er
f
o
r
m
e
d
in
r
ea
l
tim
e
b
e
ca
u
s
e
of
t
h
e
lo
wer
ca
lcu
latio
n
ch
ar
g
e
of
th
e
tin
y
v
er
s
io
n
.
Ho
wev
er
f
u
tu
r
e
s
tu
d
ies
will
be
h
eld
to
co
n
f
ir
m
th
e
o
u
tco
m
es
of
th
is
p
ap
er
.
By
im
p
lem
e
n
tin
g
th
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
in
an
e
m
b
ed
d
e
d
ca
r
d
wh
ic
h
is
lim
it
ed
in
m
em
o
r
y
a
n
d
r
eso
u
r
ce
s
not
lik
e
t
h
e
co
m
p
u
te
r
.
T
ab
le
3.
E
v
alu
atio
n
of
YOL
O
v
7
-
T
in
y
an
d
YOL
Ov
7
alg
o
r
ith
m
M
o
d
e
l
u
se
d
A
c
c
u
r
a
c
y
R
e
c
a
l
l
mA
P
@
.
5
M
o
d
e
l
s
i
z
e
I
n
f
e
r
e
n
c
e
(
ms)
D
e
f
a
u
l
t
Y
O
LO
v
7
0
.
8
5
2
0
.
8
3
3
0
.
8
6
6
74
.
8
MB
1
9
.
3
Y
O
LO
v
7
+
k
-
m
e
a
n
s++
0
.
8
8
1
0
.
7
9
1
0
.
8
5
7
74
.
8
MB
2
2
.
5
Ti
n
y
Y
O
LO
v
7
0
.
8
6
7
0
.
7
6
9
0
.
8
2
4
1
2
.
3
MB
1
1
.
5
Ti
n
y
Y
O
LO
v
7
+
k
-
mea
n
s
a
n
d
a
u
g
m
e
n
t
e
d
d
a
t
a
0
.
9
6
9
0
.
8
0
2
0
.
9
0
3
1
2
.
3
MB
1
1
.
2
T
ab
le
4.
C
o
m
p
a
r
is
o
n
to
o
th
er
s
tate
-
of
-
ar
t
n
etwo
r
k
s
M
e
t
h
o
d
A
c
c
u
r
a
c
y
R
e
c
a
l
l
mA
P
@
.
5
M
o
d
e
l
s
i
z
e
I
n
f
e
r
e
n
c
e
(
ms)
FFSS
D
[
1
0
]
9
2
.
7
9
%
8
3
.
5
2
%
-
-
-
D
S
R
A
-
D
ETR
[
1
4
]
7
6
;
1
3
%
-
9
8
.
1
2
%
-
-
Y
O
LO
v
5
x
[
2
7
]
9
5
.
1
%
8
6
.
5
%
7
5
%
-
-
F
a
st
e
r
R
–
C
N
N
[
2
7
]
4
3
.
2
6
%
-
-
-
1
6
6
.
6
7
Y
O
LO
v
4
[
2
8
]
5
9
.
8
8
%
-
-
-
2
8
.
5
7
O
u
r
s
97
.
0
0
%
8
0
%
9
0
%
1
2
.
3
M
B
1
1
.
2
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:
2088
-
8
7
0
8
Tr
a
ffic sig
n
s
d
etec
tio
n
a
n
d
p
r
o
h
ib
ito
r
s
ig
n
s
r
ec
o
g
n
itio
n
i
n
Mo
r
o
cc
o
r
o
a
d
s
ce
n
e
(
I
ma
n
e
T
a
o
u
q
i
)
6319
T
ab
le
5
.
E
v
alu
atio
n
o
f
YOL
O
v
7
-
tin
y
o
v
er
th
e
5
class
C
l
a
s
ses
La
b
e
l
s
A
c
c
u
r
a
c
y
R
e
c
a
l
l
mA
P
@
.
5
A
l
l
1
4
6
9
6
%
8
0
%
9
0
%
P
r
o
h
i
b
i
t
o
r
y
27
9
5
%
7
8
%
8
6
%
M
a
n
d
a
t
o
r
y
68
1
0
0
%
9
7
%
9
8
%
S
t
o
p
3
9
9
%
6
7
%
8
6
%
W
a
r
n
i
n
g
17
9
9
%
8
2
%
9
2
%
Y
i
e
l
d
31
9
2
%
7
7
%
8
8
%
4
.
3
.
T
est
env
iro
nm
ent
of
t
ra
f
f
ic
s
ig
ns
im
a
g
es
T
h
e
r
esu
lts
f
o
u
n
d
wh
ile
test
in
g
o
u
r
p
r
o
p
o
s
ed
s
y
s
tem
,
ar
e
d
i
s
p
lay
ed
in
Fig
u
r
e
3
,
wh
ich
is
co
r
r
ec
tly
lo
ca
lized
m
an
d
ato
r
y
,
p
r
o
h
ib
it
o
r
y
,
war
n
in
g
,
y
ield
,
an
d
s
to
p
s
ig
n
s
.
Af
ter
war
d
,
s
h
o
ws
th
e
la
b
el
o
f
p
r
o
h
ib
it
s
ig
n
s
if
th
ey
ex
is
t.
T
h
e
s
y
s
tem
d
e
m
o
n
s
tr
ated
im
p
r
ess
iv
e
ac
cu
r
a
cy
,
s
u
cc
ess
f
u
lly
class
if
y
in
g
p
r
o
h
ib
ito
r
y
T
S
with
h
ig
h
p
r
ec
is
io
n
.
T
h
e
d
is
p
lay
ed
im
ag
es
ar
e
tak
en
at
d
if
f
er
en
t
p
ar
ts
o
f
th
e
d
a
y
an
d
in
d
if
f
er
en
t
Mo
r
o
cc
o
cities.
W
h
ich
d
em
o
n
s
tr
ates th
e
ac
cu
r
ate
r
esu
lt with
o
u
t d
ep
e
n
d
en
c
y
o
n
s
p
ec
if
ied
ca
s
es.
Su
b
s
eq
u
en
tly
,
wh
ile
ev
alu
ati
n
g
th
e
YOL
Ov
7
d
etec
to
r
o
n
T
S
th
at
d
o
es
n
o
t
b
elo
n
g
to
tr
ain
in
g
d
atasets
,
o
r
th
er
e
is
an
o
cc
lu
s
i
o
n
,
it
attem
p
ts
to
id
en
tify
th
ei
r
class
es
b
y
lo
o
k
in
g
f
o
r
s
im
ili
tu
d
e
with
e
x
is
tin
g
T
S.
b
ec
au
s
e
ea
ch
class
f
r
o
m
th
e
f
iv
e
class
es h
as it
s
ch
ar
ac
ter
is
tics
.
Nev
er
th
eless
,
o
u
r
d
etec
to
r
ca
n
d
etec
t sig
n
s
th
at
ar
e
n
o
t
i
n
clu
d
ed
in
tr
ain
in
g
d
atasets
with
an
ac
cu
r
ac
y
o
f
8
0
%.
I
n
o
th
er
ter
m
s
,
o
u
r
tr
ain
ed
alg
o
r
ith
m
d
etec
tio
n
d
em
o
n
s
tr
ates
s
tr
o
n
g
g
en
er
aliza
tio
n
ab
ilit
y
b
y
h
an
d
lin
g
all
f
o
u
n
d
s
ig
n
s
in
r
o
ad
s
ce
n
es
as
te
s
ted
o
n
d
if
f
er
en
t im
a
g
es in
Fig
u
r
e
4
.
Fig
u
r
e
3
.
Dif
f
e
r
en
t tr
af
f
ic
s
ig
n
d
etec
tio
n
an
d
r
ec
o
g
n
itio
n
r
esu
lts
Fig
u
r
e
4
.
T
est o
n
im
ag
es d
o
es n
o
t b
elo
n
g
th
e
tr
ai
n
ed
d
atasets
T
o
s
u
m
m
ar
ize,
th
e
wh
o
le
m
o
d
el
d
etec
tio
n
an
d
r
ec
o
g
n
itio
n
s
y
s
tem
ex
h
ib
ited
r
o
b
u
s
t
p
e
r
f
o
r
m
a
n
ce
ac
r
o
s
s
d
if
f
er
en
t
s
ce
n
ar
io
s
,
s
h
o
wca
s
in
g
its
ef
f
ec
tiv
en
ess
in
d
etec
tin
g
,
an
d
ca
teg
o
r
izi
n
g
a
m
u
ltit
u
d
e
o
f
T
S
an
d
class
if
y
in
g
p
r
o
h
i
b
ito
r
y
class
es,
esp
ec
ially
s
p
ee
d
lim
it
s
ig
n
s
.
T
h
e
r
esu
lts
u
n
d
er
s
co
r
e
t
h
e
r
eliab
ilit
y
o
f
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
e
r
20
24
:
6
3
1
3
-
6
3
2
1
6320
r
ec
o
g
n
itio
n
s
y
s
tem
,
m
a
k
in
g
it
a
p
r
o
m
is
in
g
to
o
l
f
o
r
e
n
h
an
ci
n
g
r
o
a
d
s
af
ety
th
r
o
u
g
h
im
p
r
o
v
ed
s
ig
n
d
etec
tio
n
an
d
r
ec
o
g
n
itio
n
.
5.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
in
v
o
lv
ed
th
e
d
etec
tio
n
o
f
T
S
u
s
in
g
th
e
YOL
Ov
7
an
d
YOL
Ov
7
-
tin
y
d
etec
tio
n
alg
o
r
ith
m
,
to
d
em
o
n
s
tr
ate
its
ab
ilit
y
to
d
e
liv
er
m
o
r
e
ac
c
u
r
ate
r
esu
lts
in
r
ea
l
-
tim
e,
h
o
wev
er
as
s
h
o
wn
in
th
e
ab
o
v
e
r
esu
lts
,
it
i
s
ca
p
ab
le
o
f
d
etec
tin
g
s
ig
n
s
in
d
if
f
e
r
en
t
a
n
g
les,
o
cc
lu
s
io
n
,
an
d
c
o
m
p
lex
en
v
ir
o
n
m
en
ts
.
Mo
r
eo
v
er
,
em
p
lo
y
in
g
k
-
m
ea
n
s
++
im
p
r
o
v
es
th
e
o
v
er
all
p
e
r
f
o
r
m
an
ce
o
f
th
e
d
etec
tio
n
b
y
ad
a
p
tin
g
th
e
an
ch
o
r
b
o
x
es,
a
s
th
ey
s
u
g
g
est
p
o
ten
tial
lo
ca
tio
n
s
f
o
r
o
b
jects
in
th
e
im
a
g
e.
T
h
is
f
ac
ilit
ates
th
e
task
o
f
p
r
ed
ictin
g
b
o
u
n
d
i
n
g
b
o
x
es
an
d
o
b
ject
class
es
wh
il
e
en
ab
lin
g
th
e
d
etec
tio
n
o
f
d
i
f
f
er
en
t
o
b
ject
s
izes
an
d
s
h
ap
es
in
th
e
im
ag
e.
T
h
e
ex
p
er
im
en
ts
o
f
th
e
YOL
Ov
7
a
lg
o
r
ith
m
wer
e
ca
r
r
ied
o
u
t
u
s
in
g
o
u
r
o
wn
ad
a
p
ted
Mo
r
o
cc
an
d
atasets
.
Al
s
o
,
th
e
in
teg
r
atio
n
o
f
a
r
ec
o
g
n
itio
n
m
o
d
u
le
f
o
r
c
r
itical
p
r
o
h
i
b
ito
r
y
s
ig
n
s
in
r
o
ad
s
ce
n
es.
Hen
ce
,
f
u
tu
r
e
r
esear
ch
ef
f
o
r
ts
will a
im
to
d
ep
l
o
y
th
e
t
r
ain
ed
alg
o
r
ith
m
in
to
em
b
ed
d
ed
h
ar
d
war
e.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
R
a
n
j
b
a
r
z
a
d
e
h
e
t
a
l
.
,
“
M
R
F
E
-
C
N
N
:
mu
l
t
i
-
r
o
u
t
e
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
mo
d
e
l
f
o
r
b
r
e
a
st
t
u
mo
r
s
e
g
me
n
t
a
t
i
o
n
i
n
m
a
mm
o
g
r
a
ms
u
si
n
g
a
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
,
”
An
n
a
l
s
o
f
O
p
e
r
a
t
i
o
n
s
Re
s
e
a
r
c
h
,
v
o
l
.
3
2
8
,
n
o
.
1
,
p
p
.
1
0
2
1
–
1
0
4
2
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
7
9
-
0
2
2
-
0
4
7
5
5
-
8.
[
2
]
P
.
N
.
C
h
o
w
d
h
u
r
y
,
A
.
K
.
B
h
u
n
i
a
,
A
.
S
a
i
n
,
S
.
K
o
l
e
y
,
T
.
X
i
a
n
g
,
a
n
d
Y
.
-
Z.
S
o
n
g
,
“
W
h
a
t
c
a
n
h
u
m
a
n
s
k
e
t
c
h
e
s
d
o
f
o
r
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
?
,
”
i
n
2
0
2
3
I
EE
E/
C
VF
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
(
C
VPR)
,
J
u
n
.
2
0
2
3
,
p
p
.
1
5
0
8
3
–
1
5
0
9
4
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
5
2
7
2
9
.
2
0
2
3
.
0
1
4
4
8
.
[
3
]
B
.
B
h
a
t
t
a
r
a
i
,
R
.
S
u
b
e
d
i
,
R
.
R
.
G
a
i
r
e
,
E.
V
a
z
q
u
e
z
,
a
n
d
D
.
S
t
o
y
a
n
o
v
,
“
H
i
st
o
g
r
a
m
o
f
o
r
i
e
n
t
e
d
g
r
a
d
i
e
n
t
s
mee
t
d
e
e
p
l
e
a
r
n
i
n
g
:
a
n
o
v
e
l
mu
l
t
i
-
t
a
s
k
d
e
e
p
n
e
t
w
o
r
k
f
o
r
2
D
s
u
r
g
i
c
a
l
i
m
a
g
e
sem
a
n
t
i
c
se
g
me
n
t
a
t
i
o
n
,
”
M
e
d
i
c
a
l
I
m
a
g
e
A
n
a
l
y
s
i
s
,
v
o
l
.
8
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
me
d
i
a
.
2
0
2
3
.
1
0
2
7
4
7
.
[
4
]
A
.
K
u
r
a
n
i
,
P
.
D
o
s
h
i
,
A
.
V
a
k
h
a
r
i
a
,
a
n
d
M
.
S
h
a
h
,
“
A
c
o
mp
r
e
h
e
n
s
i
v
e
c
o
m
p
a
r
a
t
i
v
e
s
t
u
d
y
o
f
a
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
(
A
N
N
)
a
n
d
su
p
p
o
r
t
v
e
c
t
o
r
mac
h
i
n
e
s
(
S
V
M
)
o
n
st
o
c
k
f
o
r
e
c
a
st
i
n
g
,
”
A
n
n
a
l
s
o
f
D
a
t
a
S
c
i
e
n
c
e
,
v
o
l
.
1
0
,
n
o
.
1
,
p
p
.
1
8
3
–
2
0
8
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
4
0
7
4
5
-
0
2
1
-
0
0
3
4
4
-
x.
[
5
]
B
.
Ja
g
a
d
e
e
sh
a
n
d
D
.
V
.
V
i
d
h
y
a
S
r
e
e
,
“
D
e
t
e
c
t
i
o
n
a
n
d
r
e
c
o
g
n
i
t
i
o
n
o
f
t
r
a
f
f
i
c
si
g
n
b
o
a
r
d
s
u
s
i
n
g
r
a
n
d
o
m
f
o
r
e
s
t
c
l
a
ss
i
f
i
e
r
,
”
Re
v
i
e
w
o
f
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
Re
s
e
a
r
c
h
,
v
o
l
.
9
,
n
o
.
3
,
p
p
.
1
3
5
–
1
4
9
,
2
0
2
2
,
d
o
i
:
1
0
.
1
8
4
8
8
/
7
6
.
v
9
i
3
.
3
1
0
9
.
[
6
]
S
.
W
a
n
g
a
n
d
X
.
H
e
,
“
T
h
e
d
e
v
e
l
o
p
me
n
t
a
n
d
a
p
p
l
i
c
a
t
i
o
n
o
f
c
o
m
p
u
t
e
r
v
i
si
o
n
t
e
c
h
n
o
l
o
g
y
i
n
t
h
e
e
r
a
o
f
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
,
”
J
o
u
r
n
a
l
o
f
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Pr
a
c
t
i
c
e
,
v
o
l
.
5
,
n
o
.
3
,
2
0
2
2
,
d
o
i
:
1
0
.
2
3
9
7
7
/
j
a
i
p
.
2
0
2
2
.
0
5
0
3
0
1
.
[
7
]
J.
K
a
n
g
,
S
.
Ta
r
i
q
,
H
.
O
h
,
a
n
d
S
.
S
.
W
o
o
,
“
A
s
u
r
v
e
y
o
f
d
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
me
t
h
o
d
s
a
n
d
d
a
t
a
set
s
f
o
r
o
v
e
r
h
e
a
d
i
ma
g
e
r
y
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
2
0
1
1
8
–
2
0
1
3
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
1
4
9
0
5
2
.
[
8
]
Y
.
W
u
,
F
.
M
e
n
g
,
Y
.
Q
i
n
,
Y
.
Q
i
a
n
,
F
.
X
u
,
a
n
d
L
.
Ji
a
,
“
U
A
V
i
ma
g
e
r
y
b
a
se
d
p
o
t
e
n
t
i
a
l
safe
t
y
h
a
z
a
r
d
e
v
a
l
u
a
t
i
o
n
f
o
r
h
i
g
h
-
sp
e
e
d
r
a
i
l
r
o
a
d
u
si
n
g
R
e
a
l
-
t
i
m
e
i
n
s
t
a
n
c
e
se
g
me
n
t
a
t
i
o
n
,
”
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
ri
n
g
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
5
5
,
J
a
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
e
i
.
2
0
2
2
.
1
0
1
8
1
9
.
[
9
]
V
.
S
r
i
v
a
s
t
a
v
a
,
S
.
M
i
sh
r
a
,
a
n
d
N
.
G
u
p
t
a
,
“
A
u
t
o
m
a
t
i
c
d
e
t
e
c
t
i
o
n
a
n
d
c
a
t
e
g
o
r
i
z
a
t
i
o
n
o
f
r
o
a
d
t
r
a
f
f
i
c
si
g
n
s
u
s
i
n
g
a
k
n
o
w
l
e
d
g
e
-
a
ss
i
st
e
d
met
h
o
d
,
”
Pr
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
1
8
,
p
p
.
1
2
8
0
–
1
2
8
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s
.
2
0
2
3
.
0
1
.
1
0
6
.
[
1
0
]
P
.
Ti
a
n
,
“
F
F
S
S
D
:
F
e
a
t
u
r
e
f
u
si
o
n
S
S
D
a
l
g
o
r
i
t
h
m
w
i
t
h
S
E
o
n
t
r
a
f
f
i
c
si
g
n
d
e
t
e
c
t
i
o
n
,
”
Pro
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
0
8
,
p
p
.
3
3
8
–
3
4
6
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s.2
0
2
2
.
1
0
.
0
4
8
.
[
1
1
]
Q
.
Li
u
,
J.
B
i
,
J
.
Z
h
a
n
g
,
X
.
B
u
,
a
n
d
N
.
H
a
n
a
j
i
m
a
,
“
B
-
F
P
N
S
S
D
:
a
n
S
S
D
a
l
g
o
r
i
t
h
m
b
a
se
d
o
n
a
b
i
d
i
r
e
c
t
i
o
n
a
l
f
e
a
t
u
r
e
f
u
si
o
n
p
y
r
a
mi
d
,
”
T
h
e
V
i
s
u
a
l
C
o
m
p
u
t
e
r
,
v
o
l
.
3
9
,
n
o
.
1
2
,
p
p
.
6
2
6
5
–
6
2
7
7
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
3
7
1
-
0
2
2
-
0
2
7
2
7
-
4.
[
1
2
]
S
.
S
e
l
ç
u
k
,
S
.
B
e
k
e
r
,
a
n
d
Ö
.
F
.
B
o
y
r
a
z
,
“
D
e
v
e
l
o
p
m
e
n
t
o
f
a
t
r
a
f
f
i
c
s
p
e
e
d
l
i
mi
t
s
i
g
n
d
e
t
e
c
t
i
o
n
sy
s
t
e
m
b
a
s
e
d
o
n
Y
o
l
o
v
4
n
e
t
w
o
r
k
,
”
Ad
v
a
n
c
e
s i
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Re
sea
r
c
h
,
v
o
l
.
3
,
n
o
.
2
,
p
p
.
6
6
–
7
5
,
O
c
t
.
2
0
2
3
,
d
o
i
:
1
0
.
5
4
5
6
9
/
a
a
i
r
.
1
1
8
4
5
6
9
.
[
1
3
]
R
.
K
.
M
e
g
a
l
i
n
g
a
m
,
K
.
Th
a
n
i
g
u
n
d
a
l
a
,
S
.
R
.
M
u
sa
n
i
,
H
.
N
i
d
a
m
a
n
u
r
u
,
a
n
d
L
.
G
a
d
d
e
,
“
I
n
d
i
a
n
t
r
a
f
f
i
c
si
g
n
d
e
t
e
c
t
i
o
n
a
n
d
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
T
r
a
n
s
p
o
rt
a
t
i
o
n
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
2
,
n
o
.
3
,
p
p
.
6
8
3
–
6
9
9
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
j
t
st
.
2
0
2
2
.
0
6
.
0
0
2
.
[
1
4
]
J.
X
i
a
,
M
.
Li
,
W
.
Li
u
,
a
n
d
X
.
C
h
e
n
,
“
D
S
R
A
-
D
ET
R
:
a
n
i
mp
r
o
v
e
d
D
E
TR
f
o
r
mu
l
t
i
sc
a
l
e
t
r
a
f
f
i
c
s
i
g
n
d
e
t
e
c
t
i
o
n
,
”
S
u
st
a
i
n
a
b
i
l
i
t
y
,
v
o
l
.
1
5
,
n
o
.
1
4
,
J
u
l
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s
u
1
5
1
4
1
0
8
6
2
.
[
1
5
]
C.
-
Y
.
W
a
n
g
,
A
.
B
o
c
h
k
o
v
s
k
i
y
,
a
n
d
H
.
-
Y
.
M
.
L
i
a
o
,
“
Y
O
LO
v
7
:
Tr
a
i
n
a
b
l
e
b
a
g
-
of
-
f
r
e
e
b
i
e
s
se
t
s
n
e
w
st
a
t
e
-
of
-
t
h
e
-
a
r
t
f
o
r
r
e
a
l
-
t
i
me
o
b
j
e
c
t
d
e
t
e
c
t
o
r
s,
”
a
rxi
v
.
o
r
g
/
a
b
s
/
2
2
0
7
.
0
2
6
9
6
,
J
u
l
.
2
0
2
2
.
[
1
6
]
H
.
Li
a
n
d
J
.
W
a
n
g
,
“
C
o
l
l
a
b
o
r
a
t
i
v
e
a
n
n
e
a
l
i
n
g
p
o
w
e
r
k
-
m
e
a
n
s+
+
c
l
u
s
t
e
r
i
n
g
,
”
K
n
o
w
l
e
d
g
e
-
Ba
s
e
d
S
y
st
e
m
s
,
v
o
l
.
2
5
5
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
k
n
o
sy
s
.
2
0
2
2
.
1
0
9
5
9
3
.
[
1
7
]
F
.
A
.
K
a
t
e
b
,
M
.
M
.
M
o
n
o
w
a
r
,
M
.
A
.
H
a
mi
d
,
A
.
Q
.
O
h
i
,
a
n
d
M
.
F
.
M
r
i
d
h
a
,
“
F
r
u
i
t
D
e
t
:
A
t
t
e
n
t
i
v
e
f
e
a
t
u
r
e
a
g
g
r
e
g
a
t
i
o
n
f
o
r
r
e
a
l
-
t
i
m
e
f
r
u
i
t
d
e
t
e
c
t
i
o
n
i
n
o
r
c
h
a
r
d
s
,
”
Ag
ro
n
o
m
y
,
v
o
l
.
1
1
,
n
o
.
1
2
,
N
o
v
.
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
a
g
r
o
n
o
m
y
1
1
1
2
2
4
4
0
.
[
1
8
]
B
.
R
.
C
h
a
n
g
,
H
.
-
F
.
Ts
a
i
,
a
n
d
C
.
-
W
.
H
si
e
h
,
“
A
c
c
e
l
e
r
a
t
i
n
g
t
h
e
r
e
s
p
o
n
s
e
o
f
sel
f
-
d
r
i
v
i
n
g
c
o
n
t
r
o
l
b
y
u
si
n
g
r
a
p
i
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
st
e
e
r
i
n
g
a
n
g
l
e
p
r
e
d
i
c
t
i
o
n
,
”
E
l
e
c
t
r
o
n
i
c
s
,
v
o
l
.
1
2
,
n
o
.
1
0
,
M
a
y
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s1
2
1
0
2
1
6
1
.
[
1
9
]
G
.
D
.
R
a
j
a
n
d
B
.
P
r
a
b
a
d
e
v
i
,
“
S
t
e
e
l
st
r
i
p
q
u
a
l
i
t
y
a
ssu
r
a
n
c
e
w
i
t
h
Y
O
LO
V
7
-
C
S
F
:
a
c
o
o
r
d
i
n
a
t
e
a
t
t
e
n
t
i
o
n
a
n
d
S
I
o
U
f
u
si
o
n
a
p
p
r
o
a
c
h
,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
1
2
9
4
9
3
–
1
2
9
5
0
6
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
3
3
8
9
4
.
[
2
0
]
R
.
Z
h
a
n
g
,
Y
.
Ti
a
n
,
Z.
X
u
,
a
n
d
D
.
L
i
u
,
“
D
e
s
i
g
n
o
f
a
n
c
h
o
r
b
o
x
e
s
a
n
d
d
a
t
a
a
u
g
me
n
t
a
t
i
o
n
f
o
r
t
r
a
n
sf
o
r
me
r
-
b
a
se
d
v
e
h
i
c
l
e
l
o
c
a
l
i
z
a
t
i
o
n
,
”
J
o
u
r
n
a
l
o
f
Vi
s
u
a
l
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
I
m
a
g
e
Re
p
res
e
n
t
a
t
i
o
n
,
v
o
l
.
9
0
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
v
c
i
r
.
2
0
2
2
.
1
0
3
7
1
1
.
[
2
1
]
R
.
M
a
h
i
m
a
,
M
.
M
a
h
e
sw
a
r
i
,
S
.
R
o
s
h
a
n
a
,
E.
P
r
i
y
a
n
k
a
,
N
.
M
o
h
a
n
a
n
,
a
n
d
N
.
N
a
n
d
h
i
n
i
,
“
A
c
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
f
t
h
e
m
o
s
t
c
o
mm
o
n
l
y
u
s
e
d
a
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
s
i
n
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
i
n
2
0
2
3
4
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
El
e
c
t
r
o
n
i
c
s
a
n
d
S
u
s
t
a
i
n
a
b
l
e
C
o
m
m
u
n
i
c
a
t
i
o
n
S
y
st
e
m
s (
I
C
ES
C
)
,
J
u
l
.
2
0
2
3
,
p
p
.
1
3
3
4
–
1
3
3
9
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
ESC
5
7
6
8
6
.
2
0
2
3
.
1
0
1
9
3
3
9
0
.
[
2
2
]
M
.
Z
u
b
a
i
r
,
M
.
A
.
I
q
b
a
l
,
A
.
S
h
i
l
,
M
.
J
.
M
.
C
h
o
w
d
h
u
r
y
,
M
.
A
.
M
o
n
i
,
a
n
d
I
.
H
.
S
a
r
k
e
r
,
“
A
n
i
m
p
r
o
v
e
d
K
-
m
e
a
n
s c
l
u
st
e
r
i
n
g
a
l
g
o
r
i
t
h
m
t
o
w
a
r
d
s
a
n
e
f
f
i
c
i
e
n
t
d
a
t
a
-
d
r
i
v
e
n
mo
d
e
l
i
n
g
,
”
An
n
a
l
s
o
f
D
a
t
a
S
c
i
e
n
c
e
,
J
u
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s
4
0
7
4
5
-
022
-
0
0
4
2
8
-
2.
[
2
3
]
V
.
S
i
c
h
k
a
r
,
“
Tr
a
f
f
i
c
s
i
g
n
s
d
e
t
e
c
t
i
o
n
b
y
Y
O
LO
v
3
,
O
p
e
n
C
V
,
K
e
r
a
s,
”
K
a
g
g
l
e
,
2
0
2
1
.
A
c
c
e
sse
d
:
A
u
g
.
1
9
,
2
0
2
3
.
[
O
n
l
i
n
e
]
,
A
v
a
i
l
a
b
l
e
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
Tr
a
ffic sig
n
s
d
etec
tio
n
a
n
d
p
r
o
h
ib
ito
r
s
ig
n
s
r
ec
o
g
n
itio
n
i
n
Mo
r
o
cc
o
r
o
a
d
s
ce
n
e
(
I
ma
n
e
T
a
o
u
q
i
)
6321
h
t
t
p
s
:
/
/
w
w
w
.
k
a
g
g
l
e
.
c
o
m
/
c
o
d
e
/
v
a
l
e
n
t
y
n
si
c
h
k
a
r
/
t
r
a
f
f
i
c
-
si
g
n
s
-
d
e
t
e
c
t
i
o
n
-
by
-
y
o
l
o
-
v3
-
o
p
e
n
c
v
-
k
e
r
a
s/
d
a
t
a
[
2
4
]
I
.
Ta
o
u
q
i
,
A
.
K
l
i
l
o
u
,
K
.
C
h
a
j
i
,
a
n
d
A
.
A
r
sal
a
n
e
,
“
Y
O
LO
v
2
i
m
p
l
e
me
n
t
a
t
i
o
n
a
n
d
o
p
t
i
mi
z
a
t
i
o
n
f
o
r
M
o
r
o
c
c
a
n
t
r
a
f
f
i
c
si
g
n
d
e
t
e
c
t
i
o
n
,
”
2
0
2
3
,
p
p
.
8
3
7
–
8
4
3
.
[
2
5
]
K
.
Ji
a
n
g
e
t
a
l
.
,
“
A
n
a
t
t
e
n
t
i
o
n
m
e
c
h
a
n
i
sm
-
i
m
p
r
o
v
e
d
Y
O
LO
v
7
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
a
l
g
o
r
i
t
h
m
f
o
r
h
e
m
p
d
u
c
k
c
o
u
n
t
e
st
i
ma
t
i
o
n
,
”
Ag
ri
c
u
l
t
u
r
e
,
v
o
l
.
1
2
,
n
o
.
1
0
,
O
c
t
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
a
g
r
i
c
u
l
t
u
r
e
1
2
1
0
1
6
5
9
.
[
2
6
]
R
.
P
a
d
i
l
l
a
,
S
.
L.
N
e
t
t
o
,
a
n
d
E.
A
.
B
.
d
a
S
i
l
v
a
,
“
A
s
u
r
v
e
y
o
n
p
e
r
f
o
r
m
a
n
c
e
m
e
t
r
i
c
s
f
o
r
o
b
j
e
c
t
-
d
e
t
e
c
t
i
o
n
a
l
g
o
r
i
t
h
ms
,
”
i
n
2
0
2
0
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
y
st
e
m
s,
S
i
g
n
a
l
s
a
n
d
I
m
a
g
e
Pr
o
c
e
s
si
n
g
(
I
WS
S
I
P)
,
J
u
l
.
2
0
2
0
,
p
p
.
2
3
7
–
2
4
2
,
d
o
i
:
1
0
.
1
1
0
9
/
I
W
S
S
I
P
4
8
2
8
9
.
2
0
2
0
.
9
1
4
5
1
3
0
.
[
2
7
]
N
.
Y
o
u
sso
u
f
,
“
Tr
a
f
f
i
c
s
i
g
n
c
l
a
ssi
f
i
c
a
t
i
o
n
u
s
i
n
g
C
N
N
a
n
d
d
e
t
e
c
t
i
o
n
u
s
i
n
g
f
a
st
e
r
-
R
C
N
N
a
n
d
Y
O
LO
v
4
,
”
H
e
l
i
y
o
n
,
v
o
l
.
8
,
n
o
.
1
2
,
D
e
c
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
2
.
e
1
1
7
9
2
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Im
a
n
e
T
a
o
u
q
i
re
c
e
iv
e
d
t
h
e
m
a
ste
r
’
s
d
e
g
re
e
in
c
o
m
m
u
n
ica
ti
o
n
sy
ste
m
a
n
d
e
m
b
e
d
d
e
d
e
lec
tro
n
ic
fr
o
m
t
h
e
Ab
d
e
lma
lek
Essa
a
d
i
U
n
iv
e
rsit
y
,
M
o
ro
c
c
o
,
i
n
2
0
2
0
.
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
E
n
g
i
n
e
e
rin
g
.
He
r
a
re
a
o
f
re
se
a
rc
h
in
c
lu
d
e
s
a
rti
f
i
c
ial
in
telli
g
e
n
c
e
,
c
o
m
p
u
ter
v
isio
n
,
ima
g
e
p
ro
c
e
ss
in
g
a
n
d
e
m
b
e
d
d
e
d
sy
ste
m
s.
S
h
e
is
c
u
rre
n
tl
y
a
P
h
.
D.
stu
d
e
n
t
in
t
h
e
field
s
o
f
e
m
b
e
d
d
e
d
sy
ste
m
s
a
n
d
a
rti
fi
c
ial
in
telli
g
e
n
c
e
i
n
F
a
c
u
lt
y
o
f
S
c
ien
c
e
s
a
n
d
Tec
h
n
o
lo
g
y
,
Un
iv
e
rsity
o
f
S
u
lt
a
n
M
o
u
la
y
S
li
m
a
n
e
,
Be
n
i
M
e
ll
a
l
M
o
ro
c
c
o
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
ima
n
e
.
tao
u
q
i@u
sm
s.m
a
.
Abd
e
ss
a
m
a
d
K
li
lo
u
re
c
e
iv
e
d
a
n
e
n
g
in
e
e
r
’
s
d
e
g
re
e
in
2
0
1
0
a
n
d
a
P
h
.
D
.
d
e
g
re
e
i
n
2
0
1
6
fr
o
m
t
h
e
Un
i
v
e
rsity
o
f
Ca
d
y
Ay
y
a
d
,
M
a
rra
k
e
c
h
,
M
o
r
o
c
c
o
.
S
in
c
e
2
0
1
7
,
h
e
is
a
p
r
o
fe
ss
o
r
a
t
th
e
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
En
g
i
n
e
e
rin
g
i
n
th
e
F
a
c
u
lt
y
o
f
S
c
ien
c
e
s
a
n
d
Tec
h
n
o
l
o
g
y
,
Un
iv
e
rsity
o
f
S
u
lt
a
n
M
o
u
lay
S
li
m
a
n
e
,
Be
n
i
M
e
ll
a
l
M
o
r
o
c
c
o
.
His
a
re
a
o
f
re
se
a
rc
h
in
c
lu
d
e
s
o
n
p
a
ra
ll
e
l
a
n
d
re
a
l
ti
m
e
o
p
ti
m
iza
ti
o
n
o
f
si
g
n
a
l
p
r
o
c
e
ss
in
g
a
lg
o
rit
h
m
s
o
n
m
u
lt
i
-
c
o
re
a
n
d
m
u
lt
i
-
p
ro
c
e
ss
o
rs p
a
ra
ll
e
l
m
a
c
h
in
e
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
.
k
l
il
o
u
@u
sm
s.m
a
.
K
e
b
ir
Cha
ji
is
a
p
ro
fe
ss
o
r
o
f
El
e
c
tri
c
a
l
E
n
g
i
n
e
e
rin
g
a
t
F
a
c
u
lt
y
o
f
S
c
ien
c
e
s
a
n
d
Tec
h
n
o
l
o
g
y
,
U
n
iv
e
rsit
y
o
f
S
u
lt
a
n
M
o
u
lay
S
li
m
a
n
e
,
Be
n
i
M
e
ll
a
l
M
o
ro
c
c
o
,
sin
c
e
1
9
9
6
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
k
e
b
ir.
c
h
a
ji
@g
m
a
il
.
c
o
m
.
As
sia
Ar
sa
la
n
e
re
c
e
iv
e
d
a
n
e
n
g
i
n
e
e
r
’
s
d
e
g
re
e
in
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
fro
m
th
e
Na
ti
o
n
a
l
S
c
h
o
o
l
o
f
Ap
p
li
e
d
S
c
ien
c
e
s
o
f
Kh
o
u
ri
b
g
a
i
n
2
0
1
4
a
n
d
a
P
h
.
D
.
d
e
g
re
e
i
n
2
0
1
9
fr
o
m
th
e
Un
i
v
e
rsity
o
f
Ha
ss
a
n
I,
S
e
tt
a
t,
M
o
ro
c
c
o
.
S
i
n
c
e
2
0
2
0
,
sh
e
is
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
in
t
h
e
De
p
a
rtme
n
t
o
f
M
e
c
h
a
tr
o
n
ics
a
t
th
e
Hig
h
e
r
S
c
h
o
o
l
o
f
Tec
h
n
o
l
o
g
ies
,
U
n
iv
e
rsit
y
o
f
S
u
lt
a
n
M
o
u
la
y
S
li
m
a
n
e
,
Be
n
i
M
e
ll
a
l
M
o
ro
c
c
o
.
He
r
a
re
a
o
f
re
se
a
rc
h
i
n
c
l
u
d
e
s
a
rti
ficia
l
i
n
telli
g
e
n
c
e
,
m
a
c
h
in
e
v
isio
n
,
ima
g
e
p
r
o
c
e
ss
in
g
,
d
a
ta
a
n
a
l
y
sis,
a
n
d
e
m
b
e
d
d
e
d
s
y
s
tem
s.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
rsa
lan
.
a
ss
ia@
g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.