I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
14
,
No
.
3
,
J
u
n
e
2
0
2
5
,
p
p
.
1696
~
1
7
0
7
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
3
.
p
p
1
6
9
6
-
1
7
0
7
1696
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
A review
of re
cen
t
deep
learning
a
pplica
tions
in
wo
o
d surface
defec
t
iden
tif
ica
ti
o
n
M
a
rt
ina
Ali
1
,
Um
m
i R
a
ba
’
a
h H
a
s
him
1
,
K
a
s
t
uri K
a
nchy
m
a
la
y
1
,
Aj
i P
ra
s
et
y
a
Wiba
wa
2
,
L
iza
wa
t
i Sa
la
hu
dd
i
n
1
,
Ra
hil
lda
Na
dh
ira
h No
rizz
a
t
y
Ra
hid
din
1
1
C
e
n
t
r
e
f
o
r
A
d
v
a
n
c
e
d
C
o
m
p
u
t
i
n
g
T
e
c
h
n
o
l
o
g
y
,
F
a
c
u
l
t
y
o
f
I
n
f
o
r
m
a
t
i
o
n
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
,
U
n
i
v
e
r
s
i
t
i
T
e
k
n
i
k
a
l
M
a
l
a
y
s
i
a
M
e
l
a
k
a
,
M
e
l
a
k
a
,
M
a
l
a
y
s
i
a
2
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
a
n
d
I
n
f
o
r
m
a
t
i
c
s,
F
a
c
u
l
t
y
o
f
E
n
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
si
t
a
s
N
e
g
e
r
i
M
a
l
a
n
g
,
M
a
l
a
n
g
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
2
8
,
2
0
2
4
R
ev
is
ed
No
v
1
9
,
2
0
2
4
Acc
ep
ted
J
an
2
7
,
2
0
2
5
Wo
o
d
is
wi
d
e
ly
u
se
d
i
n
c
o
n
stru
c
ti
o
n
,
a
rt,
a
n
d
h
o
m
e
a
p
p
li
c
a
ti
o
n
s
d
u
e
t
o
it
s
a
e
sth
e
ti
c
a
p
p
e
a
l
a
n
d
fa
v
o
ra
b
le
m
e
c
h
a
n
ica
l
p
r
o
p
e
rti
e
s.
Ho
we
v
e
r,
e
n
v
iro
n
m
e
n
tal
fa
c
to
rs
sig
n
ifi
c
a
n
tl
y
a
ffe
c
t
th
e
g
r
o
wth
a
n
d
p
re
se
rv
a
ti
o
n
o
f
wo
o
d
,
o
ften
lea
d
in
g
to
d
e
fe
c
t
s
th
a
t
c
a
n
re
d
u
c
e
it
s
p
e
rfo
rm
a
n
c
e
a
n
d
o
rn
a
m
e
n
tal
v
a
l
u
e
.
Re
se
a
rc
h
e
rs
h
a
v
e
in
t
ro
d
u
c
e
d
m
a
c
h
in
e
v
isi
o
n
a
n
d
d
e
e
p
lea
rn
in
g
m
e
th
o
d
s
to
a
d
d
re
ss
t
h
e
c
h
a
ll
e
n
g
e
s
o
f
h
ig
h
lab
o
r
c
o
sts
a
n
d
in
e
fficie
n
c
ies
in
i
d
e
n
ti
f
y
in
g
wo
o
d
d
e
fe
c
ts.
De
e
p
lea
rn
in
g
h
a
s
sh
o
wn
g
re
a
t
su
c
c
e
ss
in
ima
g
e
r
e
c
o
g
n
it
i
o
n
tas
k
s,
y
ield
i
n
g
im
p
re
ss
iv
e
re
su
lt
s.
Th
is
p
a
p
e
r
re
v
iew
s
p
re
v
i
o
u
s
w
o
rk
o
n
d
e
e
p
-
lea
rn
in
g
stra
teg
ies
fo
r
id
e
n
ti
fy
i
n
g
w
o
o
d
su
rfa
c
e
d
e
fe
c
ts.
It
a
lso
d
isc
u
ss
e
s
d
a
ta
a
u
g
m
e
n
tati
o
n
tec
h
n
i
q
u
e
s
t
o
a
d
d
re
ss
li
m
it
e
d
d
e
fe
c
t
d
a
ta
a
n
d
e
x
p
lo
re
s
tran
sfe
r
lea
rn
in
g
t
o
e
n
h
a
n
c
e
c
las
sifica
ti
o
n
a
c
c
u
ra
c
y
o
n
sm
a
ll
d
a
tas
e
ts.
F
i
n
a
ll
y
,
th
e
p
a
p
e
r
e
x
a
m
in
e
s
t
h
e
p
o
te
n
ti
a
l
li
m
it
a
ti
o
n
s
o
f
d
e
e
p
lea
rn
i
n
g
f
o
r
d
e
fe
c
t
id
e
n
ti
fica
ti
o
n
a
n
d
s
u
g
g
e
sts
fu
tu
re
re
se
a
rc
h
d
irec
ti
o
n
s.
K
ey
w
o
r
d
s
:
Au
to
m
ated
in
s
p
ec
tio
n
Dee
p
lear
n
in
g
Def
ec
t id
en
tific
atio
n
T
r
an
s
f
er
lear
n
i
n
g
W
o
o
d
s
u
r
f
ac
e
d
e
f
ec
ts
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
:
Um
m
i Rab
a’
ah
Hash
im
C
en
tr
e
f
o
r
Ad
v
a
n
ce
d
C
o
m
p
u
ti
n
g
T
ec
h
n
o
lo
g
y
,
Facu
lty
o
f
I
n
f
o
r
m
atio
n
a
n
d
C
o
m
m
u
n
icatio
n
T
ec
h
n
o
lo
g
y
Un
iv
er
s
iti T
ek
n
ik
al
Ma
lay
s
ia
Me
lak
a
St.
Han
g
T
u
ah
J
ay
a,
7
6
1
0
0
D
u
r
ian
T
u
n
g
g
al,
Me
lak
a,
Ma
lay
s
ia
E
m
ail:
u
m
m
i@
u
tem
.
ed
u
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
Sin
ce
th
e
Par
is
Ag
r
e
em
en
t
,
en
d
o
r
s
ed
b
y
th
e
Un
ited
Natio
n
s
in
New
Yo
r
k
in
2
0
1
6
,
co
u
n
tr
ies
wo
r
ld
wid
e
h
a
v
e
s
tr
iv
ed
t
o
p
r
eser
v
e
r
eso
u
r
ce
s
,
u
s
e
en
er
g
y
ef
f
icien
tly
,
an
d
r
e
d
u
ce
ca
r
b
o
n
em
is
s
io
n
s
.
I
n
t
h
is
co
n
tex
t,
co
n
s
er
v
in
g
f
o
r
est
r
eso
u
r
ce
s
an
d
o
p
tim
izin
g
wo
o
d
u
tili
za
tio
n
h
av
e
b
ec
o
m
e
in
cr
e
asin
g
ly
im
p
o
r
tan
t.
His
to
r
ically
,
wo
o
d
h
as
b
ee
n
o
n
e
o
f
th
e
m
o
s
t
ab
u
n
d
an
t
a
n
d
v
al
u
ab
le
m
ater
ials
.
I
n
an
cien
t
tim
es,
it
was
ess
en
tial
f
o
r
s
h
elter
,
to
o
ls
,
f
u
el,
an
d
wea
p
o
n
r
y
.
As
civ
ilizatio
n
s
ad
v
an
ce
d
,
alter
n
ativ
e
m
ater
ials
b
eg
an
to
r
ep
lace
wo
o
d
f
o
r
s
p
ec
if
ic
u
s
e
s
.
Ho
wev
er
,
th
er
e
h
as
b
ee
n
a
r
esu
r
g
en
ce
in
th
e
p
r
o
d
u
ctio
n
o
f
n
ew
an
d
d
iv
er
s
e
wo
o
d
-
b
ased
p
r
o
d
u
cts.
W
ith
r
ap
id
g
lo
b
al
ec
o
n
o
m
ic
g
r
o
wth
,
th
e
d
em
a
n
d
f
o
r
wo
o
d
a
n
d
wo
o
d
p
r
o
d
u
cts
h
as
s
u
r
g
ed
as
p
eo
p
le
s
ee
k
an
im
p
r
o
v
ed
q
u
ality
o
f
life
.
Un
f
o
r
tu
n
ately
,
th
e
cu
r
r
en
t
wo
o
d
s
to
r
ag
e
an
d
p
r
o
ce
s
s
in
g
ca
p
ac
ity
is
in
s
u
f
f
icien
t
to
m
ee
t
th
is
d
em
an
d
.
L
im
ited
wo
o
d
s
u
p
p
ly
an
d
lo
w
u
tili
za
tio
n
r
ates
h
av
e
h
in
d
er
e
d
th
e
d
ev
elo
p
m
e
n
t
o
f
th
e
wo
o
d
in
d
u
s
tr
y
.
T
h
er
ef
o
r
e,
a
co
m
p
r
eh
en
s
iv
e
ev
alu
atio
n
o
f
lo
g
an
d
b
o
ar
d
p
r
o
ce
s
s
in
g
q
u
ality
is
n
ec
ess
ar
y
to
im
p
r
o
v
e
u
tili
za
tio
n
r
ates a
n
d
en
h
an
ce
th
e
q
u
ality
o
f
wo
o
d
p
r
o
d
u
cts.
W
o
o
d
is
wi
d
el
y
u
s
e
d
i
n
m
a
n
u
f
ac
t
u
r
i
n
g
d
u
e
t
o
i
ts
s
tr
en
g
t
h
,
d
u
r
ab
ilit
y
,
an
d
v
e
r
s
at
ilit
y
,
m
ak
in
g
it
s
u
it
ab
le
f
o
r
v
a
r
i
o
u
s
a
p
p
li
ca
t
io
n
s
.
H
o
w
ev
er
,
it is
r
a
r
e
to
f
in
d
lo
g
s
w
it
h
f
la
wless
s
u
r
f
ac
es i
n
w
o
o
d
m
an
u
f
a
ct
u
r
in
g
.
As
a
n
at
u
r
al
b
io
lo
g
i
ca
l
m
at
er
i
al,
w
o
o
d
is
v
u
l
n
e
r
ab
le
to
m
ic
r
o
o
r
g
a
n
is
m
s
th
at
ca
n
d
am
a
g
e
it
s
s
tr
u
ct
u
r
e,
le
a
d
i
n
g
to
d
ef
ec
ts
.
T
h
es
e
d
ef
ec
ts
,
s
u
ch
as
i
r
r
eg
u
l
ar
t
is
s
u
e
f
o
r
m
a
ti
o
n
s
an
d
s
tr
u
ct
u
r
al
d
am
ag
e,
c
a
n
af
f
ec
t
t
h
e
q
u
alit
y
a
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
r
ev
iew
o
f rece
n
t d
ee
p
lea
r
n
in
g
a
p
p
lica
tio
n
s
in
w
o
o
d
s
u
r
fa
ce
d
efec
t id
en
tifi
ca
tio
n
(
Ma
r
ti
n
a
A
li
)
1697
lif
esp
a
n
o
f
w
o
o
d
p
r
o
d
u
cts
[
1
]
,
[
2
]
.
W
o
o
d
d
e
f
e
cts
c
an
b
e
ca
t
eg
o
r
iz
ed
in
to
t
h
r
e
e
ty
p
es:
g
r
o
wth
d
e
f
e
cts
(
ca
u
s
e
d
b
y
p
h
y
s
i
o
l
o
g
i
ca
l
f
ac
t
o
r
s
)
,
p
est
d
am
ag
e
d
e
f
e
cts
(
d
u
e
to
p
at
h
o
lo
g
ic
al
f
ac
t
o
r
s
)
,
an
d
p
r
o
c
ess
i
n
g
d
e
f
e
cts
(
r
es
u
lt
in
g
f
r
o
m
h
u
m
an
e
r
r
o
r
)
[
3
]
.
M
o
r
e
o
v
e
r
,
s
u
r
f
a
ce
d
ef
ec
ts
s
i
g
n
if
ic
an
t
ly
im
p
a
ct
th
e
q
u
al
it
y
o
f
w
o
o
d
f
in
is
h
es
,
in
f
l
u
e
n
ci
n
g
s
tr
e
n
g
t
h
a
n
d
ae
s
t
h
e
tic
a
p
p
e
al
[
4
]
.
De
f
e
cts
ca
n
b
e
cl
ass
i
f
ie
d
as
p
e
r
m
is
s
i
b
le
o
r
n
o
n
-
p
e
r
m
is
s
ib
l
e,
d
e
p
e
n
d
i
n
g
o
n
th
e
ir
s
e
v
e
r
it
y
.
Pe
r
m
is
s
i
b
le
d
e
f
e
cts
a
r
e
m
i
n
o
r
a
n
d
all
o
w
a
b
le,
wh
ile
n
o
n
-
p
e
r
m
is
s
i
b
l
e
d
e
f
e
cts
r
e
n
d
e
r
w
o
o
d
u
n
s
u
i
ta
b
l
e
f
o
r
p
r
o
d
u
cts
t
h
a
t
r
e
q
u
ir
e
h
i
g
h
s
tr
u
c
tu
r
al
i
n
te
g
r
it
y
o
r
v
is
u
al
p
e
r
f
ec
t
io
n
[
5
]
.
W
o
o
d
o
r
its
c
o
m
p
o
n
e
n
ts
ca
n
b
e
s
o
r
te
d
b
as
ed
o
n
t
h
e
t
y
p
e,
s
i
ze
,
a
n
d
n
u
m
b
e
r
o
f
d
e
f
ec
ts
,
wh
i
c
h
h
el
p
s
d
e
te
r
m
i
n
e
w
h
et
h
er
it c
a
n
b
e
r
e
p
a
ir
ed
,
r
e
cy
cle
d
,
o
r
d
is
c
ar
d
ed
.
I
d
en
ti
f
y
in
g
d
e
f
e
cti
v
e
wo
o
d
s
u
r
f
a
ce
s
w
as
ess
e
n
t
ial
f
o
r
m
ai
n
t
ai
n
i
n
g
h
i
g
h
p
r
o
d
u
cti
o
n
q
u
ali
ty
a
n
d
s
a
f
e
ty
[
6
]
–
[
9
]
.
I
n
wo
o
d
m
a
n
u
f
a
ct
u
r
in
g
,
d
e
f
e
cts
r
e
d
u
ce
d
wo
o
d
y
i
el
d
b
y
a
n
av
er
ag
e
o
f
1
0
%
[
1
0
]
.
T
h
er
ef
o
r
e
,
t
h
e
ea
r
l
y
i
d
e
n
t
if
ic
at
io
n
o
f
d
e
f
ec
t
iv
e
it
em
s
o
n
t
h
e
p
r
o
d
u
c
ti
o
n
li
n
e
w
as
c
r
i
tic
al
t
o
m
ai
n
t
ai
n
i
n
g
o
v
e
r
all
q
u
ali
ty
[
1
1
]
.
A
t
h
o
r
o
u
g
h
i
n
s
p
e
cti
o
n
was
n
ec
ess
a
r
y
t
o
e
n
s
u
r
e
th
a
t
w
o
o
d
m
et
t
h
e
s
p
ec
if
ic
ati
o
n
s
f
o
r
its
i
n
t
e
n
d
ed
u
s
e
b
y
i
m
p
le
m
e
n
t
in
g
r
o
b
u
s
t
q
u
ali
ty
c
o
n
tr
o
l
p
r
o
c
ess
es
[
1
2
]
,
[
1
3
]
.
Q
u
a
lit
y
c
o
n
t
r
o
l
i
n
w
o
o
d
m
a
n
u
f
ac
tu
r
i
n
g
en
s
u
r
e
d
t
h
a
t
p
r
o
d
u
cts
m
et
p
e
r
f
o
r
m
a
n
ce
,
s
a
f
e
ty
,
an
d
ae
s
th
eti
c
s
tan
d
ar
d
s
.
E
ar
ly
i
d
e
n
ti
f
i
ca
ti
o
n
o
f
d
e
f
ec
ts
e
n
ab
le
d
m
a
n
u
f
ac
t
u
r
e
r
s
to
t
ak
e
c
o
r
r
e
cti
v
e
a
cti
o
n
,
h
el
p
i
n
g
to
m
ai
n
t
ai
n
p
r
o
d
u
ct
q
u
ali
ty
a
n
d
c
o
n
s
is
te
n
c
y
[
8
]
,
[
1
2
]
.
B
ef
o
r
e
th
e
in
tr
o
d
u
ctio
n
o
f
a
u
t
o
m
ated
v
is
u
al
in
s
p
ec
tio
n
(
AV
I
)
,
th
e
wo
o
d
in
d
u
s
tr
y
r
elie
d
o
n
m
an
u
al
in
s
p
ec
tio
n
m
eth
o
d
s
,
wh
er
e
h
u
m
an
o
p
er
ato
r
s
p
h
y
s
ically
in
s
p
ec
ted
wo
o
d
s
u
r
f
ac
es
to
i
d
e
n
tify
d
ef
ec
ts
.
T
h
is
tr
ad
itio
n
al
ap
p
r
o
ac
h
was
wid
e
ly
u
s
ed
i
n
p
r
im
ar
y
an
d
s
ec
o
n
d
ar
y
wo
o
d
i
n
d
u
s
tr
ies
an
d
d
id
n
o
t
r
eq
u
ir
e
c
o
m
p
lex
tech
n
ical
s
etu
p
s
.
Ho
wev
e
r
,
it
o
f
f
e
r
ed
lim
ited
p
o
ten
tial
f
o
r
f
u
tu
r
e
d
e
v
elo
p
m
e
n
t
d
u
e
t
o
f
r
eq
u
en
t
ch
an
g
es
in
s
tan
d
ar
d
o
p
er
atin
g
p
r
o
ce
d
u
r
e
s
as
n
ew
d
ef
ec
ts
wer
e
id
en
tif
ied
.
T
h
e
m
an
u
al
in
s
p
ec
tio
n
m
eth
o
d
s
wer
e
o
f
ten
in
ac
cu
r
ate,
in
e
f
f
icien
t,
a
n
d
r
e
s
tr
icted
p
r
o
d
u
ctio
n
v
o
lu
m
es,
with
ac
cu
r
ac
y
r
ates
o
f
o
n
ly
ab
o
u
t
7
0
%,
r
aisi
n
g
co
n
ce
r
n
s
ab
o
u
t
th
eir
r
eliab
ilit
y
[
1
4
]
–
[
1
6
]
.
Ad
d
itio
n
ally
,
m
an
u
al
in
s
p
ec
tio
n
s
wer
e
co
s
tly
,
as
tr
ain
in
g
p
er
s
o
n
n
el
r
eq
u
ir
ed
s
ig
n
if
ican
t
tim
e
an
d
r
eso
u
r
ce
s
,
an
d
v
is
u
al
f
atig
u
e
o
f
ten
led
to
m
is
id
en
tifie
d
d
e
f
ec
ts
[
1
7
]
.
H
u
m
an
er
r
o
r
also
v
ar
ies
d
ep
en
d
in
g
o
n
th
e
wo
r
k
er
s
'
ex
p
er
ien
ce
,
s
k
ills
,
an
d
aler
tn
es
s
[
1
8
]
.
Fu
r
th
er
m
o
r
e,
h
ig
h
p
r
o
d
u
ctio
n
v
o
l
u
m
es
an
d
r
ep
etitiv
e
task
s
o
v
er
p
r
o
lo
n
g
ed
p
er
io
d
s
n
eg
ativ
ely
im
p
ac
te
d
h
u
m
an
o
p
er
ato
r
s
,
lead
in
g
to
ex
h
a
u
s
tio
n
,
s
tr
ess
,
an
d
r
ed
u
ce
d
in
s
p
ec
tio
n
q
u
ality
[
1
9
]
.
As
a
r
esu
lt,
m
an
u
al
in
s
p
ec
tio
n
s
wer
e
n
o
t
o
n
ly
s
lo
wer
b
u
t a
ls
o
less
ac
cu
r
ate
co
m
p
ar
ed
to
au
to
m
ated
m
e
th
o
d
s
[
2
0
]
–
[
2
2
]
.
T
o
o
v
e
r
co
m
e
th
e
lim
itatio
n
s
o
f
m
a
n
u
al
i
n
s
p
ec
tio
n
,
t
h
e
wo
o
d
in
d
u
s
tr
y
i
n
cr
ea
s
in
g
ly
em
b
r
ac
ed
tech
n
o
lo
g
y
in
teg
r
ated
with
in
tellig
en
t
alg
o
r
ith
m
s
[
2
3
]
.
T
h
ese
alg
o
r
ith
m
s
s
ig
n
if
ica
n
tly
en
h
an
ce
d
d
ef
ec
t
id
en
tific
atio
n
p
r
o
ce
s
s
es,
wh
ic
h
b
ec
am
e
a
p
r
im
a
r
y
f
o
cu
s
with
in
th
e
in
d
u
s
tr
y
[
2
4
]
,
[
2
5
]
.
R
esear
ch
er
s
ac
tiv
ely
ex
p
lo
r
ed
ad
v
an
ce
d
tech
n
iq
u
es
,
s
u
ch
as
d
ee
p
lear
n
in
g
to
im
p
r
o
v
e
ac
cu
r
ac
y
a
n
d
ef
f
icien
c
y
.
As
a
r
esu
lt,
th
er
e
was
g
r
o
win
g
in
ter
est
in
m
ac
h
i
n
e
v
is
io
n
s
y
s
tem
s
,
wh
ich
o
f
f
e
r
ed
f
aster
an
d
m
o
r
e
p
r
ec
is
e
ap
p
r
o
ac
h
es
to
d
ef
ec
t
id
en
tific
atio
n
[
2
6
]
.
AVI
s
y
s
t
em
s
,
eq
u
ip
p
ed
with
ar
tific
ial
in
tellig
en
ce
,
g
ain
ed
tr
ac
tio
n
as
a
s
o
lu
tio
n
to
en
h
an
ce
q
u
ality
co
n
tr
o
l
in
th
e
wo
o
d
-
b
ased
in
d
u
s
tr
y
[
5
]
,
[
1
5
]
,
[
2
7
]
.
W
ith
ad
v
a
n
ce
m
en
ts
in
ar
tific
ial
in
tellig
en
ce
an
d
c
o
m
p
u
ter
v
i
s
io
n
tech
n
o
lo
g
ies,
d
ee
p
lear
n
in
g
em
e
r
g
ed
as
a
h
ig
h
ly
ef
f
ec
tiv
e
m
eth
o
d
f
o
r
id
en
tify
in
g
wo
o
d
d
e
f
ec
ts
[
2
8
]
–
[
3
0
]
.
Fu
r
th
e
r
m
o
r
e,
d
u
e
to
th
eir
s
im
p
licity
an
d
af
f
o
r
d
ab
ilit
y
,
AVI
s
y
s
tem
s
b
ec
am
e
p
o
p
u
lar
f
o
r
en
s
u
r
in
g
h
ig
h
er
ac
cu
r
ac
y
an
d
p
r
o
d
u
c
tio
n
v
o
lu
m
es
b
y
elim
in
atin
g
h
u
m
an
lim
itatio
n
s
,
im
p
r
o
v
in
g
r
eliab
ilit
y
,
an
d
m
ain
tain
in
g
q
u
ality
s
tan
d
a
r
d
s
[
1
4
]
.
Desp
ite
ad
v
an
ce
m
en
ts
,
th
e
wo
o
d
in
d
u
s
tr
y
s
till
r
eq
u
ir
es
s
o
lu
tio
n
s
to
im
p
r
o
v
e
p
r
o
ce
s
s
in
g
ef
f
icien
cy
an
d
in
cr
ea
s
e
y
ield
with
o
u
t
co
m
p
r
o
m
is
in
g
p
r
o
d
u
ct
q
u
ality
.
AVI
h
as b
ee
n
h
ig
h
lig
h
ted
f
o
r
i
ts
r
o
le
in
en
s
u
r
in
g
co
n
s
is
ten
t p
r
o
d
u
ct
r
elia
b
ilit
y
an
d
a
d
d
r
ess
in
g
y
ield
lo
s
s
es d
u
e
to
lim
itatio
n
s
in
m
an
u
al
in
s
p
e
ctio
n
s
.
R
esear
ch
in
d
icate
s
th
a
t
AVI
o
f
f
er
s
2
5
%
h
ig
h
er
i
d
en
tific
atio
n
ac
cu
r
ac
y
th
an
tr
ad
itio
n
al
m
eth
o
d
s
,
lead
in
g
to
a
5
.
3
%
in
cr
ea
s
e
in
y
ield
an
d
co
n
s
id
er
ab
le
co
s
t
s
av
in
g
s
f
o
r
th
e
av
e
r
ag
e
r
o
u
g
h
m
ill
[
3
1
]
.
Fu
r
th
er
m
o
r
e,
au
to
m
ated
g
r
ad
in
g
h
as
s
h
o
wn
g
r
ea
ter
p
r
ec
is
io
n
an
d
co
n
s
is
ten
cy
t
h
an
co
n
v
en
tio
n
al
in
s
p
ec
tio
n
s
,
wh
i
ch
o
f
te
n
s
tr
u
g
g
le
to
o
p
tim
ize
wo
o
d
r
eso
u
r
ce
s
[
3
2
]
.
Stu
d
ie
s
also
d
em
o
n
s
tr
ate
th
at
AVI
o
u
tp
er
f
o
r
m
s
h
u
m
an
in
s
p
ec
to
r
s
in
i
d
en
tify
in
g
d
ef
ec
ts
an
d
p
lay
s
a
c
r
u
cial
r
o
le
i
n
m
ain
tain
in
g
q
u
ality
co
n
tr
o
l,
u
ltima
tely
b
en
ef
itin
g
t
h
e
s
ec
o
n
d
ar
y
wo
o
d
in
d
u
s
tr
y
b
y
im
p
r
o
v
in
g
y
ield
s
an
d
p
r
o
d
u
c
tio
n
q
u
ality
[
2
6
]
.
Mo
r
eo
v
er
,
o
n
e
s
ig
n
if
ican
t
ad
v
an
ce
m
en
t
in
d
ee
p
lear
n
in
g
f
o
r
wo
o
d
d
ef
ec
t
id
en
tific
atio
n
is
u
s
in
g
d
ata
au
g
m
en
tatio
n
an
d
tr
an
s
f
er
lea
r
n
in
g
tech
n
iq
u
es
[
3
3
]
.
Data
a
u
g
m
en
tatio
n
ad
d
r
ess
es
th
e
ch
allen
g
e
o
f
lim
ited
d
atasets
b
y
en
ab
lin
g
m
o
d
els
to
g
en
er
alize
b
etter
an
d
im
p
r
o
v
e
class
if
icatio
n
ac
cu
r
ac
y
.
On
th
e
o
th
er
h
an
d
,
tr
an
s
f
er
lear
n
i
n
g
le
v
er
ag
es
p
r
e
-
tr
ain
ed
m
o
d
els,
s
u
ch
as
r
esi
d
u
al
n
etwo
r
k
(
R
esNet)
an
d
I
n
ce
p
tio
n
,
f
in
e
-
tu
n
in
g
th
em
f
o
r
s
p
ec
i
f
ic
task
s
lik
e
wo
o
d
d
ef
ec
t
class
if
icatio
n
,
r
ed
u
cin
g
th
e
n
ee
d
f
o
r
r
etr
ain
i
n
g
f
r
o
m
s
cr
atch
.
As
d
ee
p
lear
n
in
g
tech
n
i
q
u
es
ev
o
lv
e
,
f
u
r
th
er
ex
p
l
o
r
atio
n
o
f
s
o
p
h
is
ticated
au
g
m
en
tatio
n
m
eth
o
d
s
an
d
ad
v
an
ce
d
tr
an
s
f
e
r
lear
n
in
g
f
r
am
ewo
r
k
s
will
b
e
cr
u
cial
f
o
r
en
h
a
n
cin
g
d
ef
e
ct
id
en
tific
atio
n
s
y
s
tem
s
.
T
h
is
co
m
b
in
atio
n
o
f
tech
n
iq
u
es
o
v
er
co
m
es
t
h
e
li
m
itatio
n
s
o
f
m
a
n
u
al
in
s
p
ec
ti
o
n
an
d
s
ig
n
if
ican
tl
y
im
p
r
o
v
e
s
wo
o
d
p
r
o
d
u
ctio
n
ef
f
icien
cy
,
q
u
ality
ass
u
r
an
ce
,
an
d
r
eso
u
r
ce
u
tili
za
tio
n
.
2.
WO
O
D
DE
F
E
C
T
I
DE
N
T
I
F
I
CAT
I
O
N
AP
P
RO
ACH
E
S
USI
NG
D
E
E
P
L
E
AR
NING
Dete
ctin
g
an
d
id
e
n
tify
in
g
d
ef
ec
ts
ar
e
cr
u
cial
in
m
an
u
f
ac
t
u
r
in
g
to
m
ai
n
tain
co
n
tr
o
lled
a
n
d
ef
f
icien
t
p
r
o
d
u
ctio
n
p
r
o
ce
s
s
es
[
3
4
]
.
T
r
ad
itio
n
ally
,
t
h
ese
task
s
r
el
ied
o
n
s
k
illed
h
u
m
an
o
p
er
a
to
r
s
.
Ho
wev
er
,
th
e
in
tr
o
d
u
ctio
n
o
f
AVI
h
as
en
h
a
n
ce
d
th
e
a
u
to
n
o
m
y
o
f
m
a
n
u
f
a
ctu
r
in
g
o
p
er
atio
n
s
.
AVI
-
b
ased
q
u
ality
co
n
tr
o
l
h
as
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
1
6
9
6
-
1
7
0
7
1698
g
ain
ed
p
o
p
u
lar
ity
in
t
h
e
s
ec
o
n
d
ar
y
wo
o
d
in
d
u
s
tr
y
d
u
e
to
i
ts
ab
ilit
y
to
im
p
r
o
v
e
in
s
p
ec
ti
o
n
ac
cu
r
ac
y
,
b
o
o
s
t
p
r
o
d
u
ctio
n
r
ates,
an
d
l
o
wer
la
b
o
r
c
o
s
ts
.
An
AVI
s
y
s
tem
ty
p
ically
in
v
o
lv
es
s
ev
er
al
s
tag
es:
im
ag
e
ac
q
u
is
itio
n
,
im
ag
e
en
h
an
ce
m
e
n
t,
s
eg
m
en
tatio
n
,
f
ea
tu
r
e
ex
tr
ac
tio
n
,
an
d
f
ea
tu
r
e
class
if
icatio
n
[
3
5
]
.
E
f
f
icien
t
m
ater
ial
h
an
d
lin
g
en
s
u
r
es
s
m
o
o
t
h
lo
g
is
tics
,
m
ain
tain
in
g
s
tead
y
m
a
ter
ial
f
lo
w,
r
ed
u
cin
g
v
ib
r
atio
n
s
,
an
d
r
eg
u
latin
g
s
p
ee
d
d
u
r
i
n
g
tim
b
e
r
im
ag
i
n
g
.
C
r
itical
s
u
b
s
y
s
tem
s
lik
e
s
e
n
s
o
r
s
an
d
lig
h
tin
g
a
r
e
ess
en
tial
f
o
r
ca
p
t
u
r
in
g
,
d
ig
itizin
g
,
an
d
s
to
r
in
g
im
ag
e
d
ata.
T
h
e
in
s
p
ec
tio
n
p
r
o
ce
s
s
b
eg
in
s
with
d
ef
ec
t
d
etec
tio
n
,
p
in
p
o
in
tin
g
th
e
lo
ca
tio
n
o
f
f
laws
in
th
e
wo
o
d
.
T
h
is
is
f
o
llo
wed
b
y
d
ef
ec
t
id
en
tific
atio
n
,
wh
er
e
d
ef
ec
ts
'
ty
p
e,
s
ize,
an
d
f
r
eq
u
e
n
cy
ar
e
an
aly
ze
d
.
T
h
ese
co
m
p
o
n
en
ts
also
o
f
f
er
g
u
id
elin
es
f
o
r
o
p
tim
ized
tim
b
e
r
c
u
ttin
g
b
ased
o
n
t
h
e
id
en
tifie
d
d
e
f
ec
ts
.
Fin
ally
,
th
e
co
llected
d
ata
is
u
s
ed
in
th
e
g
r
ad
in
g
p
h
ase,
wh
er
e
th
e
t
im
b
er
is
class
if
ied
ac
co
r
d
in
g
to
s
p
ec
if
ic
p
r
o
d
u
ctio
n
s
tan
d
ar
d
s
.
I
n
t
h
e
w
o
o
d
i
n
d
u
s
t
r
y
,
o
n
e
o
f
t
h
e
m
o
s
t
e
f
f
e
c
t
i
v
e
m
et
h
o
d
s
f
o
r
i
d
e
n
t
i
f
y
i
n
g
d
e
f
e
c
t
s
i
n
v
o
l
v
es
p
r
o
c
e
s
s
i
n
g
a
n
d
a
n
a
l
y
z
i
n
g
i
m
a
g
e
s
o
f
w
o
o
d
s
u
r
f
a
c
e
s
.
N
u
m
e
r
o
u
s
s
t
u
d
i
es
h
a
v
e
e
x
p
l
o
r
e
d
u
s
i
n
g
A
V
I
s
y
s
t
e
m
s
,
e
m
p
l
o
y
i
n
g
t
r
a
d
i
t
i
o
n
a
l
i
m
a
g
e
p
r
o
c
e
s
s
i
n
g
,
s
p
e
c
i
al
i
z
e
d
t
e
c
h
n
i
q
u
es
,
a
n
d
a
r
t
i
f
i
c
ia
l
i
n
t
e
ll
i
g
e
n
c
e
m
e
t
h
o
d
s
[
3
6
]
.
D
e
e
p
l
e
a
r
n
i
n
g
r
e
p
r
e
s
e
n
t
s
a
s
i
g
n
i
f
i
c
a
n
t
a
d
v
a
n
c
e
m
e
n
t
w
it
h
i
n
m
a
c
h
i
n
e
l
e
a
r
n
in
g
,
c
h
a
r
a
c
t
e
r
i
z
e
d
b
y
t
h
e
c
o
n
s
t
r
u
c
t
i
o
n
o
f
m
u
l
t
i
-
l
a
y
e
r
e
d
n
e
u
r
a
l
n
e
t
w
o
r
k
s
t
h
a
t
em
u
l
a
t
e
t
h
e
c
o
m
p
l
e
x
p
r
o
c
e
s
s
es
o
f
t
h
e
h
u
m
a
n
b
r
a
i
n
.
I
t
e
n
a
b
l
e
s
t
h
e
r
e
c
o
g
n
i
t
i
o
n
o
f
d
i
v
e
r
s
e
i
n
p
u
ts
s
u
c
h
as
i
m
a
g
es
,
s
o
u
n
d
s
,
a
n
d
t
e
x
ts
[
3
7
]
.
W
it
h
in
t
h
e
s
e
n
e
u
r
a
l
n
et
w
o
r
k
s
,
al
g
o
r
i
t
h
m
s
i
n
ea
c
h
l
a
y
e
r
c
o
n
t
i
n
u
o
u
s
l
y
p
e
r
f
o
r
m
c
al
c
u
la
t
i
o
n
s
a
n
d
p
r
e
d
i
c
ti
o
n
s
,
p
r
o
g
r
e
s
s
i
v
e
l
y
e
n
h
a
n
ci
n
g
a
c
c
u
r
a
cy
o
v
e
r
t
i
m
e
.
T
h
is
m
e
t
h
o
d
o
l
o
g
y
a
d
h
e
r
e
s
t
o
d
a
ta
-
d
r
i
v
e
n
p
r
i
n
c
i
p
l
es
,
r
e
p
r
e
s
e
n
t
i
n
g
o
b
s
e
r
v
e
d
i
m
a
g
es
i
n
v
a
r
i
o
u
s
f
o
r
m
s
,
w
h
e
t
h
e
r
a
s
p
i
x
e
l
d
e
n
s
i
t
y
v
a
l
u
e
v
e
ct
o
r
s
o
r
a
b
s
t
r
a
c
t
f
e
a
t
u
r
es
l
i
k
e
e
d
g
e
s
e
q
u
e
n
ce
s
[
3
8
]
.
W
h
il
e
d
e
e
p
n
e
t
w
o
r
k
s
o
f
t
e
n
i
n
c
l
u
d
e
m
o
r
e
h
i
d
d
e
n
l
a
y
e
r
s
t
h
a
n
t
r
a
d
i
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
,
t
h
e
e
f
f
e
c
ti
v
e
n
es
s
o
f
l
ea
r
n
i
n
g
o
u
t
c
o
m
e
s
d
o
es
n
o
t
d
e
p
e
n
d
s
o
l
e
l
y
o
n
l
a
y
e
r
d
e
p
t
h
.
S
u
c
c
es
s
f
u
l
d
e
e
p
l
ea
r
n
i
n
g
d
e
s
i
g
n
h
i
n
g
e
s
o
n
d
e
t
e
r
m
i
n
i
n
g
t
h
e
o
p
t
i
m
a
l
n
u
m
b
e
r
o
f
h
id
d
e
n
l
a
y
e
r
s
t
ai
l
o
r
e
d
t
o
s
p
e
ci
f
i
c
t
a
s
k
s
[
3
]
.
T
h
i
s
e
v
o
lu
t
i
o
n
h
a
s
p
r
o
p
e
l
l
e
d
d
e
e
p
l
e
a
r
n
i
n
g
i
n
t
o
a
l
e
a
d
i
n
g
a
p
p
r
o
a
c
h
i
n
a
c
a
d
e
m
i
c
r
e
s
e
a
r
c
h
a
n
d
i
n
d
u
s
t
r
y
a
p
p
l
i
c
a
t
i
o
n
s
.
D
e
ep
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m
s
a
u
t
o
m
a
te
d
t
h
e
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
p
r
o
c
e
s
s
,
e
li
m
i
n
a
ti
n
g
t
h
e
n
e
e
d
f
o
r
m
a
n
u
a
l
i
n
p
u
t
f
r
o
m
e
x
p
e
r
t
s
.
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
(
C
NN
s
)
e
x
e
m
p
l
if
i
e
d
t
h
i
s
a
p
p
r
o
a
c
h
,
e
x
t
r
a
c
t
i
n
g
f
e
a
t
u
r
es
d
i
r
e
c
t
l
y
f
r
o
m
i
m
a
g
e
s
i
n
a
f
u
l
l
y
a
u
t
o
m
a
t
e
d
m
a
n
n
e
r
[
2
9
]
.
T
h
e
g
o
a
l
o
f
a
p
p
l
y
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
w
a
s
t
o
i
n
d
e
p
e
n
d
e
n
t
l
y
p
r
o
c
es
s
n
e
w
d
a
t
a
,
m
a
k
e
a
c
c
u
r
a
te
d
e
c
is
i
o
n
s
,
a
n
d
p
r
o
v
i
d
e
r
e
li
a
b
l
e
r
e
c
o
m
m
e
n
d
a
t
i
o
n
s
b
as
e
d
o
n
t
h
o
u
s
a
n
d
s
o
f
c
a
l
c
u
l
at
i
o
n
s
wh
i
l
e
s
i
g
n
i
f
i
c
a
n
tl
y
r
e
d
u
c
i
n
g
t
h
e
p
o
t
e
n
t
i
a
l
f
o
r
h
u
m
a
n
e
r
r
o
r
.
Var
io
u
s
ad
v
a
n
ce
d
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
s
h
av
e
p
r
o
v
en
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
f
o
r
wo
o
d
s
u
r
f
ac
e
d
ef
ec
t d
etec
tio
n
,
s
im
u
ltan
eo
u
s
ly
lear
n
in
g
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
class
if
icatio
n
d
u
r
in
g
tr
ain
i
n
g
.
T
h
is
in
n
o
v
ativ
e
ap
p
r
o
ac
h
h
ig
h
lig
h
ts
th
e
a
d
ap
t
ab
ilit
y
an
d
r
o
b
u
s
tn
ess
o
f
d
ee
p
lear
n
in
g
m
eth
o
d
o
lo
g
ies
in
a
d
d
r
ess
in
g
r
ea
l
-
wo
r
ld
ch
allen
g
es.
T
ab
le
1
s
u
m
m
ar
i
ze
s
s
ig
n
if
ican
t
s
tu
d
ies
f
r
o
m
2
0
1
9
to
2
0
2
4
,
s
h
o
wca
s
in
g
th
e
ev
o
lu
tio
n
o
f
d
ee
p
lear
n
in
g
a
r
ch
itectu
r
es
f
o
r
wo
o
d
d
e
f
ec
t
id
en
tific
atio
n
.
T
h
ese
s
tu
d
ies
d
em
o
n
s
tr
ate
th
e
g
r
o
win
g
ap
p
licatio
n
o
f
d
ee
p
lear
n
in
g
tech
n
i
q
u
es,
wh
i
ch
h
av
e
s
ig
n
i
f
ican
tly
im
p
r
o
v
e
d
d
ef
ec
t d
etec
tio
n
ac
cu
r
ac
y
in
th
e
wo
o
d
in
d
u
s
tr
y
.
T
ab
le
1
.
Ov
e
r
v
iew
o
f
d
ee
p
lea
r
n
in
g
m
o
d
els u
s
ed
in
w
o
o
d
s
u
r
f
ac
e
d
ef
ec
t id
e
n
tific
atio
n
R
e
f
e
r
e
n
c
e
Y
e
a
r
A
p
p
r
o
a
c
h
D
e
f
e
c
t
t
y
p
e
[
3
9
]
2
0
2
4
I
n
c
e
p
t
i
o
n
-
R
e
sN
e
t
-
V
2
C
N
N
m
o
d
e
l
C
r
a
c
k
s,
k
n
o
t
s,
a
n
d
u
n
d
a
m
a
g
e
d
[
4
0
]
2
0
2
4
I
mp
r
o
v
e
d
R
e
sN
e
t
-
5
0
m
o
d
e
l
B
r
o
w
n
st
a
i
n
,
b
l
u
e
s
t
a
i
n
,
k
n
o
t
,
b
o
r
e
r
h
o
l
e
s,
r
o
t
,
b
a
r
k
p
o
c
k
e
t
,
w
a
n
e
,
a
n
d
s
p
l
i
t
[
4
1
]
2
0
2
3
I
n
c
e
p
t
i
o
n
-
V
3
m
o
d
e
l
C
r
a
c
k
s,
k
n
o
t
s,
a
n
d
u
n
d
a
m
a
g
e
d
[
4
2
]
2
0
2
3
I
mp
r
o
v
e
d
R
e
sN
e
t
-
5
0
m
o
d
e
l
K
n
o
t
s,
c
r
a
c
k
s,
a
n
d
c
o
l
o
r
-
r
e
l
a
t
e
d
d
e
f
e
c
t
s
[
4
3
]
2
0
2
3
I
mp
r
o
v
e
d
R
e
g
N
e
t
m
o
d
e
l
W
o
r
mh
o
l
e
,
d
e
a
d
k
n
o
t
,
a
n
d
l
i
v
e
k
n
o
t
[
3
3
]
2
0
2
2
R
e
sN
e
t
5
0
mo
d
e
l
B
r
o
w
n
st
a
i
n
,
b
l
u
e
s
t
a
i
n
,
k
n
o
t
,
b
o
r
e
r
h
o
l
e
s,
r
o
t
,
b
a
r
k
p
o
c
k
e
t
,
w
a
n
e
,
a
n
d
s
p
l
i
t
[
1
1
]
2
0
2
2
F
a
st
e
r
R
-
C
N
N
m
o
d
e
l
K
n
o
t
,
w
a
n
e
,
e
d
g
e
,
s
t
a
i
n
,
a
n
d
b
r
a
n
c
h
.
[
4
4
]
2
0
2
2
I
n
c
e
p
t
i
o
n
-
R
e
sN
e
t
-
V
2
C
N
N
m
o
d
e
l
Ed
g
e
-
g
l
u
e
d
[
4
5
]
2
0
2
2
D
e
r
i
v
e
d
R
e
sN
e
t
-
v
2
m
o
d
e
l
W
o
r
mh
o
l
e
,
l
i
v
e
j
o
i
n
t
,
a
n
d
d
e
a
d
j
o
i
n
t
.
[
4
6
]
2
0
2
1
TL
-
R
e
sN
e
t
3
4
m
o
d
e
l
D
e
c
a
y
k
n
o
t
s,
d
r
y
k
n
o
t
s
,
e
d
g
e
k
n
o
t
s,
e
n
c
a
se
d
k
n
o
t
s
,
h
o
r
n
k
n
o
t
s,
l
e
a
f
k
n
o
t
s
,
a
n
d
s
o
u
n
d
k
n
o
t
s.
[
4
7
]
2
0
2
1
B
a
se
d
b
i
l
i
n
e
a
r
f
i
n
e
-
g
r
a
i
n
e
d
(
B
LN
N
)
K
n
o
t
[
4
8
]
2
0
2
1
C
N
N
mo
d
e
l
D
e
f
e
c
t
i
v
e
w
o
o
d
[
4
9
]
2
0
2
1
D
e
e
p
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
(
D
C
N
N
)
mo
d
e
l
N
o
t
s
p
e
c
i
f
i
e
d
[
2
9
]
2
0
2
0
M
u
l
t
i
-
c
h
a
n
n
e
l
mas
k
R
-
C
N
N
D
e
a
d
k
n
o
t
s
,
l
i
v
e
k
n
o
t
s,
a
n
d
c
r
a
c
k
s
.
[
3
0
]
2
0
2
0
D
C
N
N
mo
d
e
l
C
r
a
c
k
,
k
n
o
t
,
a
n
d
mi
l
d
e
w
[
5
0
]
2
0
2
0
M
a
s
k
R
-
C
N
N
m
o
d
e
l
D
e
a
d
k
n
o
t
s
,
l
i
v
e
k
n
o
t
s,
a
n
d
i
n
sec
t
h
o
l
e
s
[
5
1
]
2
0
1
9
M
i
x
-
F
C
N
mo
d
e
l
D
e
a
d
k
n
o
t
,
l
i
v
e
k
n
o
t
,
b
l
u
e
st
a
i
n
,
c
r
a
c
k
,
b
r
o
w
n
st
a
i
n
,
a
n
d
p
i
t
c
h
s
t
r
e
a
k
[
5
2
]
2
0
1
9
A
l
e
x
N
e
t
mo
d
e
l
Ex
t
e
r
n
a
l
d
e
f
e
c
t
s
[
5
3
]
2
0
1
9
R
e
sN
e
t
1
8
mo
d
e
l
N
o
t
s
p
e
c
i
f
i
e
d
[
1
4
]
2
0
1
9
F
a
st
e
r
R
-
C
N
N
m
o
d
e
l
S
p
l
i
t
,
c
o
r
e
,
b
r
a
n
c
h
,
a
n
d
st
a
i
n
[
2
2
]
2
0
1
9
F
a
st
e
r
R
-
C
N
N
m
o
d
e
l
K
n
o
t
s a
n
d
h
o
l
e
s
[
5
4
]
2
0
1
9
C
N
N
mo
d
e
l
K
n
o
t
s
[
5
5
]
2
0
1
9
VGG
-
16
m
o
d
e
l
N
o
t
s
p
e
c
i
f
i
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
r
ev
iew
o
f rece
n
t d
ee
p
lea
r
n
in
g
a
p
p
lica
tio
n
s
in
w
o
o
d
s
u
r
fa
ce
d
efec
t id
en
tifi
ca
tio
n
(
Ma
r
ti
n
a
A
li
)
1699
R
e
c
e
n
t
s
t
u
d
i
e
s
o
n
w
o
o
d
s
u
r
f
a
c
e
d
e
f
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
h
a
v
e
l
e
d
t
o
t
h
e
d
e
v
e
l
o
p
m
e
n
t
o
f
s
e
v
e
r
a
l
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
s
.
I
n
2
0
2
4
,
E
h
t
i
s
h
a
m
e
t
a
l
.
[
3
9
]
d
e
v
e
l
o
p
e
d
a
n
a
u
t
o
m
a
t
e
d
m
e
t
h
o
d
f
o
r
a
s
s
e
s
s
i
n
g
d
e
f
e
c
t
s
i
n
w
o
o
d
e
n
s
t
r
u
c
t
u
r
e
s
u
s
i
n
g
C
N
N
s
a
n
d
i
m
a
g
e
p
r
o
c
e
s
s
i
n
g
t
e
c
h
n
i
q
u
e
s
t
o
e
n
h
a
n
c
e
i
n
s
p
e
c
t
i
o
n
e
f
f
i
c
i
e
n
c
y
.
T
r
a
d
i
t
i
o
n
a
l
m
a
n
u
a
l
i
n
s
p
e
c
t
i
o
n
s
,
t
h
o
u
g
h
e
f
f
e
c
t
i
v
e
,
a
r
e
t
i
m
e
-
c
o
n
s
u
m
i
n
g
a
n
d
c
o
s
t
l
y
.
T
h
e
s
t
u
d
y
e
m
p
l
o
y
e
d
t
h
e
I
n
c
e
p
t
i
o
n
-
R
e
s
N
e
t
-
V
2
C
N
N
m
o
d
e
l
,
t
r
a
i
n
e
d
w
i
t
h
a
d
a
t
a
s
e
t
o
f
9
0
0
0
i
m
a
g
e
s
d
i
v
i
d
e
d
i
n
t
o
c
r
a
c
k
s
,
k
n
o
t
s
,
a
n
d
u
n
d
a
m
a
g
e
d
s
e
c
t
i
o
n
s
.
T
h
e
m
o
d
e
l
a
c
h
i
e
v
e
d
9
2
%
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
o
n
n
e
w
t
e
s
t
i
m
a
g
e
s
,
w
i
t
h
m
i
n
i
m
a
l
e
r
r
o
r
s
i
n
d
e
f
e
c
t
q
u
a
n
t
i
f
i
c
a
t
i
o
n
.
T
h
i
s
a
u
t
o
m
a
t
e
d
a
p
p
r
o
a
c
h
p
r
o
v
i
d
e
d
s
i
g
n
i
f
i
c
a
n
t
p
r
a
c
t
i
c
a
l
b
e
n
e
f
i
t
s
f
o
r
i
n
d
u
s
t
r
y
p
r
o
f
e
s
s
i
o
n
a
l
s
b
y
r
e
d
u
c
i
n
g
l
a
b
o
r
c
o
s
t
s
,
s
p
e
e
d
i
n
g
u
p
i
n
s
p
e
c
t
i
o
n
s
,
a
n
d
i
m
p
r
o
v
i
n
g
a
c
c
u
r
a
c
y
,
m
a
k
i
n
g
i
t
a
v
a
l
u
a
b
l
e
t
o
o
l
f
o
r
m
a
i
n
t
a
i
n
i
n
g
t
h
e
s
a
f
e
t
y
a
n
d
d
u
r
a
b
i
l
i
t
y
o
f
w
o
o
d
e
n
s
t
r
u
c
t
u
r
e
s
.
N
e
x
t
,
i
n
t
h
e
s
a
m
e
y
e
a
r
,
C
h
u
n
e
t
a
l
.
[
4
0
]
i
n
v
e
s
t
i
g
a
t
e
d
w
a
y
s
t
o
e
n
h
a
n
c
e
t
h
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
o
f
t
i
m
b
e
r
d
e
f
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
p
a
r
t
i
c
u
l
a
r
l
y
R
e
s
N
e
t
s
.
T
h
e
f
o
c
u
s
i
s
r
e
f
i
n
i
n
g
t
h
e
s
e
n
e
t
w
o
r
k
s
b
y
i
n
c
r
e
a
s
i
n
g
d
e
p
t
h
a
n
d
i
n
c
o
r
p
o
r
a
t
i
n
g
m
u
l
t
i
-
l
e
v
e
l
f
e
a
t
u
r
e
s
t
o
b
u
i
l
d
a
m
o
r
e
e
f
f
e
c
t
i
v
e
f
r
a
m
e
w
o
r
k
f
o
r
i
d
e
n
t
i
f
y
i
n
g
d
i
f
f
e
r
e
n
t
d
e
f
e
c
t
t
y
p
e
s
.
A
s
e
r
i
e
s
o
f
a
b
l
a
t
i
o
n
e
x
p
e
r
i
m
e
n
t
s
w
e
r
e
p
e
r
f
o
r
m
e
d
,
t
e
s
t
i
n
g
t
h
e
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
r
e
e
n
e
w
a
r
c
h
i
t
e
c
t
u
r
e
s
(
R
1
,
R
2
,
a
n
d
R
3
)
a
g
a
i
n
s
t
t
h
e
s
t
a
n
d
a
r
d
R
e
s
N
e
t
5
0
m
o
d
e
l
.
W
h
i
l
e
t
h
e
R
1
a
r
c
h
i
t
e
c
t
u
r
e
a
c
h
i
e
v
e
d
s
l
i
g
h
t
i
m
p
r
o
v
e
m
e
n
t
s
o
v
e
r
R
e
s
N
e
t
5
0
,
p
a
r
t
i
c
u
l
a
r
l
y
w
i
t
h
a
d
d
i
n
g
a
n
e
x
t
r
a
l
a
y
e
r
c
a
l
l
e
d
"
C
o
n
v
G
,
"
R
e
s
N
e
t
5
0
s
t
i
l
l
s
h
o
w
e
d
b
e
t
t
e
r
o
v
e
r
a
l
l
p
e
r
f
o
r
m
a
n
c
e
.
S
i
m
i
l
a
r
l
y
,
t
h
e
R
2
f
r
a
m
e
w
o
r
k
p
r
o
v
i
d
e
d
s
o
m
e
i
m
p
r
o
v
e
m
e
n
t
s
b
u
t
s
t
i
l
l
f
e
l
l
b
e
h
i
n
d
R
1
.
T
h
e
m
o
s
t
s
i
g
n
i
f
i
c
a
n
t
r
e
s
u
l
t
s
w
e
r
e
a
c
h
i
e
v
e
d
w
i
t
h
t
h
e
R
3
a
r
c
h
i
t
e
c
t
u
r
e
,
w
h
i
c
h
i
n
t
e
g
r
a
t
e
d
f
u
l
l
y
p
r
e
-
a
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
s
,
l
e
a
d
i
n
g
t
o
a
1
4
.
1
8
%
i
n
c
r
e
a
s
e
i
n
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
c
o
m
p
a
r
e
d
t
o
R
e
s
N
e
t
5
0
.
T
h
e
R
3
a
r
c
h
i
t
e
c
t
u
r
e
a
l
s
o
p
e
r
f
o
r
m
e
d
w
e
l
l
a
c
r
o
s
s
v
a
r
i
o
u
s
t
i
m
b
e
r
s
p
e
c
i
e
s
,
t
h
o
u
g
h
i
t
s
h
o
w
e
d
s
l
i
g
h
t
l
y
r
e
d
u
c
e
d
a
c
c
u
r
a
c
y
f
o
r
r
u
b
b
e
r
w
o
o
d
.
N
o
n
e
t
h
e
l
e
s
s
,
R
3
o
u
t
p
e
r
f
o
r
m
e
d
b
o
t
h
R
e
s
N
e
t
5
0
a
n
d
t
h
e
o
t
h
e
r
p
r
o
p
o
s
e
d
m
o
d
e
l
s
,
h
i
g
h
l
i
g
h
t
i
n
g
t
h
e
e
f
f
e
c
t
i
v
e
n
e
s
s
o
f
d
e
e
p
e
r
n
e
t
w
o
r
k
s
a
n
d
p
r
e
-
a
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
s
.
S
t
a
t
i
s
t
i
c
a
l
a
n
a
l
y
s
e
s
,
i
n
c
l
u
d
i
n
g
i
n
d
e
p
e
n
d
e
n
t
t
-
t
e
s
t
s
a
n
d
o
n
e
-
w
a
y
A
N
O
V
A
,
c
o
n
f
i
r
m
e
d
t
h
a
t
t
h
e
i
m
p
r
o
v
e
m
e
n
t
s
w
e
r
e
s
i
g
n
i
f
i
c
a
n
t
a
c
r
o
s
s
m
u
l
t
i
p
l
e
s
p
e
c
i
e
s
,
d
e
m
o
n
s
t
r
a
t
i
n
g
t
h
e
p
o
t
e
n
t
i
a
l
o
f
t
h
e
R
3
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
t
i
m
b
e
r
d
e
f
e
c
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
.
Nex
t,
in
2
0
2
3
,
Z
o
u
et
a
l.
[
4
2
]
p
r
o
p
o
s
ed
a
n
im
p
r
o
v
ed
m
o
d
elin
g
m
eth
o
d
b
ased
o
n
R
esNet
-
5
0
to
en
h
an
ce
th
e
ac
cu
r
ac
y
an
d
ef
f
i
cien
cy
o
f
wo
o
d
d
ef
ec
t
id
e
n
tif
icatio
n
,
wh
ich
was
cr
u
cial
f
o
r
en
s
u
r
in
g
f
u
r
n
itu
r
e
m
an
u
f
ac
tu
r
in
g
q
u
ality
.
R
ec
o
g
n
izin
g
th
e
co
m
p
lex
ities
in
v
o
l
v
ed
in
d
ata
an
d
th
e
n
ee
d
f
o
r
h
ig
h
e
f
f
icien
cy
,
th
e
s
tu
d
y
in
tr
o
d
u
ce
d
a
n
ew
o
p
tim
izatio
n
s
ch
em
e
th
at
co
m
b
in
ed
a
co
n
v
o
lu
tio
n
al
b
lo
c
k
atten
tio
n
m
o
d
u
le
with
a
cr
o
s
s
-
s
tag
e
p
ar
tial
n
etwo
r
k
(
C
SP
Ne
t)
tailo
r
ed
f
o
r
th
e
R
esNet
-
5
0
ar
c
h
itectu
r
e.
T
h
e
r
ese
ar
ch
ex
p
lo
r
ed
h
o
w
v
ar
y
in
g
th
e
cr
o
s
s
-
s
tag
e
p
ar
a
m
eter
s
in
C
SP
Net
af
f
ec
ted
class
if
icatio
n
p
er
f
o
r
m
an
ce
,
r
e
v
ea
lin
g
th
at
th
e
d
ef
au
lt
p
ar
am
eter
s
wer
e
n
o
t
alwa
y
s
o
p
tim
al.
Ad
d
itio
n
ally
,
th
e
au
th
o
r
p
r
o
p
o
s
ed
th
e
r
an
g
e
r
o
p
tim
izer
,
wh
ich
s
h
o
wed
s
u
p
er
io
r
p
er
f
o
r
m
a
n
ce
o
v
e
r
th
e
tr
ad
itio
n
al
Ad
am
o
p
tim
izer
in
ter
m
s
o
f
tr
ain
in
g
ef
f
icien
cy
an
d
p
r
e
d
ictio
n
ac
cu
r
ac
y
.
E
x
p
er
im
en
ts
co
n
d
u
cted
u
s
in
g
a
d
ataset
o
f
r
ea
l
-
wo
r
ld
wo
o
d
s
u
r
f
ac
e
d
ef
ec
ts
d
e
m
o
n
s
tr
ated
th
at
th
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
ie
v
ed
an
i
m
p
r
ess
iv
e
o
v
er
all
d
etec
tio
n
a
cc
u
r
ac
y
o
f
8
6
.
2
5
%
in
id
en
tify
in
g
d
ef
ec
ts
s
u
ch
as
k
n
o
ts
,
cr
ac
k
s
,
an
d
co
lo
r
-
r
elate
d
is
s
u
es.
T
h
ese
r
esu
lts
u
n
d
er
l
in
ed
th
e
e
f
f
ec
tiv
en
ess
an
d
f
ea
s
ib
ilit
y
o
f
th
e
n
ew
m
o
d
elin
g
a
p
p
r
o
ac
h
in
im
p
r
o
v
i
n
g
wo
o
d
d
ef
ec
t i
d
en
tific
atio
n
co
m
p
ar
ed
to
ex
is
tin
g
s
tate
-
of
-
th
e
-
ar
t m
eth
o
d
s
.
N
e
x
t
,
i
n
2
0
2
2
,
L
i
n
g
a
n
d
X
i
e
[
4
5
]
d
e
v
e
l
o
p
e
d
a
d
e
r
i
v
e
d
m
o
d
e
l
b
a
s
e
d
o
n
R
e
s
N
e
t
-
v
2
t
o
a
d
d
r
e
s
s
t
h
e
l
i
m
i
t
a
t
i
o
n
s
o
f
t
r
a
d
i
t
i
o
n
a
l
m
a
n
u
a
l
d
e
t
e
c
t
i
o
n
m
e
t
h
o
d
s
f
o
r
w
o
o
d
d
e
f
e
c
t
s
,
w
h
i
c
h
a
r
e
o
f
t
e
n
t
i
m
e
-
c
o
n
s
u
m
i
n
g
,
i
n
e
f
f
i
c
i
e
n
t
,
a
n
d
i
n
a
c
c
u
r
a
t
e
.
T
h
i
s
n
e
w
m
o
d
e
l
a
i
m
s
t
o
a
c
c
u
r
a
t
e
l
y
i
d
e
n
t
i
f
y
d
e
f
e
c
t
t
y
p
e
s
s
u
c
h
a
s
w
o
r
m
h
o
l
e
s
,
l
i
v
e
j
o
i
n
t
s
,
a
n
d
d
e
a
d
j
o
i
n
t
s
o
n
w
o
o
d
e
n
s
u
r
f
a
c
e
s
,
s
i
g
n
i
f
i
c
a
n
t
l
y
i
m
p
r
o
v
i
n
g
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
w
h
i
l
e
r
e
d
u
c
i
n
g
l
a
b
o
r
r
e
q
u
i
r
e
m
e
n
t
s
.
T
h
e
s
t
u
d
y
h
i
g
h
l
i
g
h
t
s
t
h
a
t
t
h
e
R
e
s
N
e
t
-
v2
-
d
e
r
i
v
e
d
m
o
d
e
l
d
e
m
o
n
s
t
r
a
t
e
s
b
e
t
t
e
r
r
e
c
o
g
n
i
t
i
o
n
c
a
p
a
b
i
l
i
t
i
e
s
a
n
d
e
n
h
a
n
c
e
d
g
e
n
e
r
a
l
i
z
a
t
i
o
n
a
b
i
l
i
t
i
e
s
t
h
a
n
t
r
a
d
i
t
i
o
n
a
l
C
N
N
.
E
x
p
e
r
i
m
e
n
t
a
l
r
e
s
u
l
t
s
i
n
d
i
c
a
t
e
t
h
a
t
t
h
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
o
f
t
h
e
R
e
s
N
e
t
-
v2
-
d
e
r
i
v
e
d
n
e
t
w
o
r
k
e
x
c
e
e
d
s
8
0
%
f
o
r
v
a
r
i
o
u
s
c
o
n
f
i
g
u
r
a
t
i
o
n
s
,
w
i
t
h
a
p
e
a
k
a
c
c
u
r
a
c
y
o
f
9
7
.
2
7
%
u
n
d
e
r
c
e
r
t
a
i
n
c
o
n
d
i
t
i
o
n
s
.
T
h
e
s
e
f
i
n
d
i
n
g
s
s
u
g
g
e
s
t
t
h
a
t
t
h
e
p
r
o
p
o
s
e
d
m
o
d
e
l
o
f
f
e
r
s
a
m
o
r
e
e
f
f
i
c
i
e
n
t
a
n
d
r
e
l
i
a
b
l
e
s
o
l
u
t
i
o
n
f
o
r
w
o
o
d
d
e
f
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
,
m
a
k
i
n
g
i
t
a
v
a
l
u
a
b
l
e
t
o
o
l
i
n
t
h
e
i
n
d
u
s
t
r
y
f
o
r
i
m
p
r
o
v
i
n
g
q
u
a
l
i
t
y
c
o
n
t
r
o
l
a
n
d
o
p
e
r
a
t
i
o
n
a
l
e
f
f
i
c
i
e
n
c
y
.
I
n
t
h
e
s
a
m
e
y
e
a
r
,
M
o
h
s
i
n
e
t
a
l
.
[
1
1
]
p
r
o
p
o
s
e
d
a
n
o
v
e
l
m
e
t
h
o
d
f
o
r
a
u
t
o
m
a
t
i
c
r
e
a
l
-
t
i
m
e
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
o
n
w
o
o
d
s
u
r
f
a
c
e
s
.
T
h
i
s
a
p
p
r
o
a
c
h
u
t
i
l
i
z
e
d
a
D
C
N
N
f
r
a
m
e
w
o
r
k
c
a
l
l
e
d
F
a
s
t
e
r
R
-
C
N
N
f
o
r
d
e
t
e
c
t
i
o
n
,
c
o
m
b
i
n
e
d
w
i
t
h
M
o
b
i
l
e
N
e
t
V
3
a
s
t
h
e
b
a
c
k
b
o
n
e
n
e
t
w
o
r
k
f
o
r
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
.
T
h
e
k
e
y
i
n
n
o
v
a
t
i
o
n
o
f
t
h
i
s
m
e
t
h
o
d
l
i
e
s
i
n
i
t
s
a
b
i
l
i
t
y
t
o
e
f
f
i
c
i
e
n
t
l
y
d
e
t
e
c
t
k
n
o
t
s
a
n
d
o
t
h
e
r
d
e
f
e
c
t
s
w
h
i
l
e
p
e
r
f
o
r
m
i
n
g
c
l
a
s
s
i
f
i
c
a
t
i
o
n
i
n
r
e
a
l
-
t
i
m
e
f
r
o
m
i
n
p
u
t
v
i
d
e
o
f
r
a
m
e
s
.
T
h
e
s
t
u
d
y
f
o
c
u
s
e
d
o
n
a
c
h
i
e
v
i
n
g
s
p
e
e
d
a
n
d
a
c
c
u
r
a
c
y
,
w
h
i
c
h
w
e
r
e
c
r
u
c
i
a
l
f
o
r
i
n
d
u
s
t
r
i
a
l
q
u
a
l
i
t
y
c
o
n
t
r
o
l
a
n
d
i
n
s
p
e
c
t
i
o
n
t
a
s
k
s
r
e
q
u
i
r
i
n
g
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
i
n
r
e
a
l
-
t
i
m
e
,
p
a
r
t
i
c
u
l
a
r
l
y
o
n
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
l
i
m
i
t
e
d
p
r
o
c
e
s
s
i
n
g
u
n
i
t
s
o
r
c
o
m
m
o
d
i
t
y
p
r
o
c
e
s
s
o
r
s
.
T
h
e
m
o
d
e
l
w
a
s
d
e
s
i
g
n
e
d
t
o
b
e
l
i
g
h
t
w
e
i
g
h
t
,
m
a
k
i
n
g
i
t
s
u
i
t
a
b
l
e
f
o
r
d
e
p
l
o
y
m
e
n
t
o
n
m
o
b
i
l
e
a
n
d
e
d
g
e
d
e
v
i
c
e
s
.
M
o
b
i
l
e
N
e
t
V
3
w
a
s
p
r
e
-
t
r
a
i
n
e
d
o
n
a
l
a
r
g
e
i
m
a
g
e
d
a
t
a
s
e
t
t
o
e
n
h
a
n
c
e
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
,
w
h
i
l
e
F
a
s
t
e
r
R
-
C
N
N
w
a
s
e
m
p
l
o
y
e
d
f
o
r
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
.
T
h
e
s
y
s
t
e
m
p
r
o
c
e
s
s
e
d
i
n
p
u
t
v
i
d
e
o
f
r
a
m
e
s
a
t
a
n
a
v
e
r
a
g
e
s
p
e
e
d
o
f
3
7
p
e
r
s
e
c
o
n
d
u
s
i
n
g
a
l
o
w
-
c
o
s
t
,
l
o
w
-
m
e
m
o
r
y
g
r
a
p
h
i
c
s
p
r
o
c
e
s
s
i
n
g
u
n
i
t
(
G
P
U
)
.
T
h
e
m
e
t
h
o
d
a
c
h
i
e
v
e
d
a
n
o
v
e
r
a
l
l
a
c
c
u
r
a
c
y
o
f
9
9
%
i
n
d
e
t
e
c
t
i
n
g
a
n
d
c
l
a
s
s
i
f
y
i
n
g
d
e
f
e
c
t
s
,
d
e
m
o
n
s
t
r
a
t
i
n
g
i
t
s
e
f
f
e
c
t
i
v
e
n
e
s
s
f
o
r
p
r
a
c
t
i
c
a
l
a
p
p
l
i
c
a
t
i
o
n
s
i
n
w
o
o
d
s
u
r
f
a
c
e
i
n
s
p
e
c
t
i
o
n
.
I
n
2
0
2
1
,
G
a
o
et
a
l
.
[
4
6
]
d
e
v
e
lo
p
e
d
t
h
e
T
L
-
R
es
Net
-
3
4
m
o
d
el,
a
t
r
a
n
s
f
e
r
r
esi
d
u
al
n
eu
r
a
l
n
etw
o
r
k
,
t
o
en
h
a
n
ce
t
h
e
i
d
e
n
ti
f
i
ca
t
io
n
o
f
wo
o
d
k
n
o
t
d
e
f
e
cts
wi
th
s
p
ee
d
a
n
d
ac
c
u
r
a
cy
.
T
h
is
m
o
d
e
l
i
m
p
r
o
v
e
d
a
cc
u
r
ac
y
b
y
o
v
e
r
0
.
7
8
%
th
r
o
u
g
h
ca
r
e
f
u
l
e
x
tr
a
cti
o
n
o
f
s
t
r
u
c
tu
r
a
l
d
e
f
ec
t
f
ea
tu
r
es
an
d
o
p
ti
m
iz
ati
o
n
o
f
t
r
ai
n
in
g
p
a
r
a
m
et
er
s
a
n
d
d
at
ase
ts
.
B
y
i
n
c
o
r
p
o
r
at
in
g
t
r
a
n
s
f
er
le
ar
n
i
n
g
,
th
e
m
o
d
e
l
b
e
n
ef
i
te
d
f
r
o
m
a
p
r
e
-
tr
ai
n
i
n
g
p
h
a
s
e
th
at
all
o
w
ed
f
o
r
m
o
r
e
ef
f
e
cti
v
e
l
ea
r
n
i
n
g
f
r
o
m
t
h
e
li
m
i
te
d
a
v
ai
la
b
le
d
a
ta
.
E
x
p
er
i
m
e
n
t
al
r
es
u
lts
d
e
m
o
n
s
t
r
a
te
d
t
h
at
T
L
-
R
esN
et
-
34
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
1
6
9
6
-
1
7
0
7
1700
ac
h
ie
v
e
d
a
n
i
m
p
r
ess
i
v
e
r
ec
o
g
n
iti
o
n
r
ate
o
f
9
9
.
2
2
%
o
n
t
h
e
tr
ai
n
i
n
g
d
at
aset
,
wit
h
a
lo
w
t
r
a
in
in
g
l
o
s
s
o
f
2
.
8
3
%
o
n
th
e
v
ali
d
a
ti
o
n
d
atas
et
w
h
i
le
i
d
en
t
if
y
i
n
g
s
e
v
en
d
is
ti
n
ct
wo
o
d
k
n
o
t
d
e
f
ec
ts
.
Ov
er
all
ac
cu
r
ac
y
r
ea
c
h
e
d
9
8
.
6
9
%
,
wit
h
m
i
n
i
m
al
f
lu
ct
u
a
ti
o
n
s
i
n
t
h
e
l
o
s
s
an
d
a
cc
u
r
ac
y
c
u
r
v
es
d
u
r
in
g
tes
ti
n
g
,
i
n
d
ica
ti
n
g
t
h
e
m
o
d
el'
s
s
t
a
b
ili
ty
a
n
d
r
el
ia
b
il
it
y
.
Si
g
n
if
ica
n
t
ly
,
th
is
m
e
th
o
d
m
i
n
i
m
iz
ed
t
h
e
n
ee
d
f
o
r
e
x
t
en
s
iv
e
i
m
a
g
e
p
r
ep
r
o
c
ess
in
g
a
n
d
f
ea
t
u
r
e
ex
t
r
ac
ti
o
n
,
m
a
k
i
n
g
i
t
e
f
f
ic
ie
n
t
an
d
ac
cu
r
ate
i
n
b
o
t
h
t
h
e
t
r
a
in
i
n
g
a
n
d
test
in
g
p
h
as
es.
T
h
e
f
i
n
d
i
n
g
s
s
u
g
g
este
d
th
at
TL
-
R
esN
et
-
3
4
h
el
d
s
i
g
n
i
f
ic
an
t
p
o
t
e
n
tia
l
f
o
r
a
p
p
lic
ati
o
n
s
i
n
wo
o
d
n
o
n
-
d
est
r
u
c
ti
v
e
test
in
g
a
n
d
w
o
o
d
d
ef
ec
t
id
e
n
t
if
ica
ti
o
n
,
al
lo
wi
n
g
f
o
r
q
u
ick
a
n
d
ac
cu
r
a
te
ass
ess
m
en
ts
o
f
c
o
l
lec
te
d
wo
o
d
k
n
o
t
d
ef
ec
t
s
.
T
h
is
i
n
n
o
v
ati
o
n
co
u
l
d
s
i
g
n
if
i
ca
n
t
ly
i
m
p
r
o
v
e
w
o
o
d
u
til
iza
ti
o
n
an
d
r
eso
u
r
c
e
co
n
s
e
r
v
at
io
n
i
n
t
h
e
i
n
d
u
s
t
r
y
.
3.
DIS
CU
SS
I
O
N
B
ased
o
n
th
e
f
i
n
d
i
n
g
s
f
r
o
m
r
e
ce
n
t
s
t
u
d
ies
o
n
w
o
o
d
d
e
f
e
ct
i
d
e
n
ti
f
i
ca
t
io
n
,
s
e
v
e
r
al
cr
i
tic
al
d
is
cu
s
s
io
n
s
ca
n
b
e
d
r
awn
.
T
h
e
r
ese
ar
ch
h
i
g
h
li
g
h
ts
t
h
e
g
r
o
w
in
g
i
m
p
o
r
t
a
n
ce
o
f
u
til
izi
n
g
d
ee
p
l
ea
r
n
in
g
t
ec
h
n
i
q
u
es,
p
a
r
ti
c
u
la
r
l
y
C
NNs,
t
o
e
n
h
a
n
ce
t
h
e
ac
c
u
r
ac
y
a
n
d
e
f
f
ic
ie
n
c
y
o
f
d
e
f
ec
t
i
d
e
n
ti
f
i
ca
ti
o
n
i
n
wo
o
d
.
T
h
e
e
v
o
lu
ti
o
n
o
f
C
NN
a
r
c
h
ite
ct
u
r
es,
s
u
c
h
as
I
n
ce
p
ti
o
n
-
R
esNet
-
V
2
a
n
d
v
ar
i
o
u
s
R
esN
et
m
o
d
els
,
h
as
s
h
o
w
n
p
r
o
m
is
in
g
r
es
u
lts
i
n
im
p
r
o
v
i
n
g
class
if
ica
ti
o
n
ac
c
u
r
ac
y
.
F
o
r
i
n
s
ta
n
ce
,
th
e
a
u
t
o
m
ate
d
m
et
h
o
d
d
ev
el
o
p
e
d
i
n
2
0
2
4
ac
h
i
ev
e
d
a
class
i
f
i
ca
t
io
n
a
cc
u
r
ac
y
o
f
9
2
%,
d
e
m
o
n
s
tr
ati
n
g
h
o
w
d
e
ep
lea
r
n
i
n
g
c
a
n
s
i
g
n
i
f
i
ca
n
tl
y
o
u
t
p
e
r
f
o
r
m
t
r
ad
iti
o
n
al
m
a
n
u
al
i
n
s
p
ec
t
i
o
n
m
et
h
o
d
s
,
w
h
ic
h
a
r
e
o
f
t
en
ti
m
e
-
c
o
n
s
u
m
in
g
an
d
c
o
s
tl
y
[
3
9
]
,
[
4
2
]
,
[
4
6
]
.
Nex
t,
as
s
ee
n
i
n
t
h
e
Fas
te
r
R
-
C
NN
an
d
Mo
b
i
le
NetV
3
f
r
am
ew
o
r
k
,
t
h
e
e
m
p
h
asis
o
n
r
e
al
-
ti
m
e
d
ete
cti
o
n
an
d
cl
ass
if
ic
ati
o
n
ca
p
a
b
il
iti
es
a
d
d
r
ess
es
p
r
ac
tic
al
n
ee
d
s
in
in
d
u
s
t
r
ia
l
s
e
tti
n
g
s
wh
er
e
q
u
i
c
k
d
ec
is
i
o
n
-
m
a
k
i
n
g
is
ess
e
n
tia
l.
T
h
e
ab
i
lit
y
t
o
p
r
o
ce
s
s
v
i
d
e
o
f
r
a
m
e
s
at
3
7
f
r
a
m
es
p
er
s
ec
o
n
d
wit
h
a
n
ac
c
u
r
a
cy
o
f
9
9
%
ill
u
s
t
r
a
tes
t
h
e
p
o
t
en
tial
o
f
th
es
e
a
d
v
a
n
ce
d
m
o
d
e
ls
t
o
e
n
h
a
n
c
e
q
u
alit
y
c
o
n
tr
o
l
in
w
o
o
d
p
r
o
c
ess
i
n
g
[
1
1
]
,
[
4
6
]
.
M
o
r
e
o
v
er
,
i
n
t
e
g
r
ati
n
g
o
p
ti
m
iz
ati
o
n
s
t
r
a
te
g
ies
,
s
u
c
h
as
c
o
n
v
o
l
u
t
io
n
al
b
l
o
c
k
att
en
ti
o
n
m
o
d
u
les
a
n
d
t
r
a
n
s
f
er
le
ar
n
in
g
t
ec
h
n
i
q
u
es,
h
as
im
p
r
o
v
e
d
m
o
d
el
p
e
r
f
o
r
m
a
n
c
e,
a
d
d
r
ess
i
n
g
c
h
all
en
g
es
l
ik
e
li
m
it
ed
l
ab
el
ed
d
at
asets
.
T
h
es
e
ad
v
an
ce
m
e
n
ts
p
r
o
m
is
e
s
u
b
s
t
a
n
ti
al
b
e
n
ef
i
ts
,
in
c
lu
d
i
n
g
r
e
d
u
c
ed
la
b
o
r
c
o
s
t
s
an
d
e
n
h
a
n
ce
d
s
a
f
et
y
,
a
n
d
th
e
y
als
o
o
p
ti
m
i
ze
r
es
o
u
r
ce
u
ti
liz
ati
o
n
i
n
ti
m
b
er
m
a
n
u
f
a
ct
u
r
i
n
g
,
d
e
m
o
n
s
t
r
a
ti
n
g
a
s
i
g
n
i
f
ic
an
t
s
h
if
t
to
wa
r
d
a
u
t
o
m
ati
o
n
i
n
w
o
o
d
d
e
f
e
ct
id
en
ti
f
ic
ati
o
n
.
Fu
r
t
h
er
m
o
r
e
,
in
tr
o
d
u
ci
n
g
t
h
e
r
a
n
g
er
o
p
ti
m
i
ze
r
a
n
d
ad
ju
s
ti
n
g
t
h
e
C
SP
Net
p
ar
am
ete
r
s
co
n
t
r
i
b
u
te
d
t
o
b
et
te
r
ac
cu
r
a
c
y
,
u
n
d
er
s
c
o
r
i
n
g
th
e
n
ec
ess
it
y
o
f
c
o
n
ti
n
u
a
l
i
m
p
r
o
v
e
m
e
n
t
i
n
th
e
m
o
d
e
l
t
r
ai
n
i
n
g
p
r
o
ce
s
s
[
3
9
]
,
[
4
2
]
,
[
4
6
]
.
T
h
e
s
t
u
d
ies
als
o
h
i
g
h
li
g
h
te
d
c
h
a
lle
n
g
es
,
p
a
r
ti
cu
la
r
l
y
t
h
e
lim
ite
d
av
aila
b
i
lit
y
o
f
la
b
e
le
d
d
at
ase
ts
in
wo
o
d
d
e
f
ec
t
d
ete
c
tio
n
.
M
o
d
els
li
k
e
T
L
-
R
es
Net
-
3
4
,
w
h
i
ch
ef
f
e
cti
v
el
y
le
v
e
r
a
g
e
tr
an
s
f
er
l
ea
r
n
i
n
g
,
s
h
o
w
t
h
at
e
v
en
w
it
h
f
ewe
r
t
r
ai
n
i
n
g
ex
a
m
p
les
,
it
is
p
o
s
s
i
b
l
e
t
o
ac
h
i
e
v
e
h
i
g
h
r
e
co
g
n
it
io
n
r
at
es,
e
m
p
h
asiz
in
g
t
h
e
im
p
o
r
t
an
ce
o
f
i
n
n
o
v
ati
v
e
a
p
p
r
o
a
ch
es
to
d
at
a
s
ca
r
c
it
y
[
4
6
]
.
I
n
a
d
d
iti
o
n
,
i
m
p
le
m
e
n
t
in
g
t
h
es
e
ad
v
a
n
ce
d
d
e
te
cti
o
n
s
y
s
te
m
s
i
n
r
ea
l
-
w
o
r
l
d
s
ce
n
a
r
i
o
s
p
r
o
m
is
es
s
i
g
n
i
f
ic
an
t
b
en
ef
its
,
in
cl
u
d
i
n
g
r
e
d
u
ce
d
l
ab
o
r
c
o
s
t
s
,
im
p
r
o
v
ed
s
a
f
et
y
,
an
d
e
n
h
a
n
c
e
d
d
u
r
a
b
il
it
y
o
f
w
o
o
d
en
s
tr
u
c
tu
r
es.
B
y
a
u
t
o
m
ati
n
g
d
ef
ec
t
d
e
te
cti
o
n
,
i
n
d
u
s
t
r
ies
c
a
n
o
p
ti
m
i
ze
t
h
eir
p
r
o
ce
s
s
es
,
le
ad
in
g
t
o
b
ette
r
r
es
o
u
r
ce
c
o
n
s
e
r
v
ati
o
n
a
n
d
m
o
r
e
e
f
f
ici
en
t
u
s
e
o
f
t
im
b
e
r
[
3
9
]
,
[
4
2
]
,
[
4
6
]
.
T
h
e
an
al
y
s
is
o
f
w
o
o
d
d
ef
ec
t
id
en
tific
atio
n
m
eth
o
d
s
f
r
o
m
2
0
1
9
to
2
0
2
4
r
ev
ea
ls
a
p
r
e
d
o
m
in
an
t
r
elian
ce
o
n
C
NN
-
b
ased
m
o
d
el
s
,
h
ig
h
lig
h
tin
g
t
h
eir
e
f
f
ec
tiv
en
ess
in
id
en
tify
in
g
s
u
r
f
ac
e
d
ef
e
cts
in
wo
o
d
.
C
NNs
en
h
an
ce
d
e
f
ec
t
id
en
tific
atio
n
t
h
r
o
u
g
h
h
ier
a
r
ch
ical
f
ea
tu
r
e
le
ar
n
in
g
,
s
ig
n
if
ica
n
tly
r
ed
u
cin
g
th
e
tim
e
an
d
ef
f
o
r
t
ass
o
ciate
d
with
m
an
u
al
in
s
p
ec
tio
n
s
[
3
9
]
.
Var
io
u
s
ar
ch
itectu
r
es,
in
clu
d
in
g
R
esNet,
Fas
ter
R
-
C
NN,
an
d
I
n
ce
p
tio
n
-
R
esNet,
s
h
o
wca
s
e
th
e
v
er
s
atility
an
d
d
o
m
in
an
ce
o
f
C
NNs
in
th
is
f
ield
.
T
h
eir
p
o
p
u
lar
ity
s
tem
s
f
r
o
m
two
m
ain
f
ac
to
r
s
:
h
ig
h
p
er
f
o
r
m
an
ce
a
n
d
ea
s
e
o
f
tr
ain
i
n
g
[
5
6
]
.
T
h
e
m
o
d
u
lar
d
esig
n
o
f
C
NNs,
co
m
p
r
is
in
g
co
n
v
o
l
u
tio
n
al
b
ases
an
d
f
u
ll
y
co
n
n
ec
ted
lay
er
s
,
allo
ws
f
o
r
r
o
b
u
s
t
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
ac
c
u
r
ate
d
ef
ec
t
class
if
icatio
n
.
T
h
is
ca
p
ab
ilit
y
im
p
r
o
v
es
th
e
p
r
ec
is
io
n
o
f
d
e
f
ec
t
id
en
tific
atio
n
b
y
ca
p
tu
r
i
n
g
in
tr
icate
f
ea
t
u
r
es
with
in
h
ig
h
-
d
im
en
s
io
n
al
d
ata
[
5
7
]
–
[
5
9
]
.
Ho
we
v
er
,
c
h
allen
g
es
p
er
s
is
t,
n
o
tab
ly
th
e
s
u
b
s
tan
tial
co
m
p
u
tatio
n
al
p
o
wer
r
eq
u
i
r
ed
f
o
r
e
f
f
ec
tiv
e
C
NN
o
p
er
atio
n
,
o
f
ten
n
ec
ess
itatin
g
ex
p
en
s
iv
e
GPUs
an
d
lar
g
e
d
atasets
[
6
0
]
.
T
o
o
v
er
co
m
e
t
h
ese
b
ar
r
ier
s
,
r
ese
ar
ch
is
n
ee
d
ed
to
d
ev
elo
p
c
o
m
p
ac
t,
ef
f
icien
t
C
NN
m
o
d
els
th
at
ca
n
f
u
n
ctio
n
o
n
lo
w
-
p
o
wer
co
m
p
u
tin
g
r
eso
u
r
ce
s
,
s
u
ch
as
C
PU
s
o
r
FP
G
As,
m
ak
in
g
th
em
m
o
r
e
ac
ce
s
s
ib
le
f
o
r
in
d
u
s
tr
ial
ap
p
licatio
n
s
.
Fu
r
th
e
r
m
o
r
e
,
cr
e
atin
g
an
d
s
h
ar
in
g
lar
g
e,
h
ig
h
-
q
u
ality
d
atasets
o
f
s
u
r
f
ac
e
d
e
f
ec
ts
is
cr
u
cial
f
o
r
en
h
an
cin
g
m
o
d
el
tr
ai
n
in
g
an
d
p
er
f
o
r
m
a
n
ce
.
Ad
d
r
ess
in
g
th
ese
g
ap
s
will
f
ac
ilit
ate
th
e
b
r
o
ad
e
r
ad
o
p
tio
n
o
f
C
NNs
in
au
to
m
ated
s
u
r
f
ac
e
in
s
p
ec
tio
n
s
y
s
tem
s
,
b
alan
cin
g
co
m
p
u
tatio
n
al
d
em
a
n
d
s
with
co
s
t
-
ef
f
ec
tiv
en
ess
an
d
d
ata
av
ailab
ilit
y
.
O
v
er
all,
C
NNs
p
lay
a
v
ital
r
o
le
in
ad
v
an
cin
g
wo
o
d
d
ef
ec
t
id
e
n
tific
atio
n
,
o
f
f
er
in
g
s
u
p
er
io
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
an
d
class
if
icatio
n
ca
p
ab
ilit
ies es
s
e
n
tial f
o
r
im
p
r
o
v
in
g
au
to
m
atio
n
in
th
e
in
d
u
s
tr
y
.
F
u
r
t
h
e
r
m
o
r
e
,
t
h
e
a
n
a
l
y
s
i
s
o
f
m
o
d
e
l
s
p
r
e
s
e
n
t
e
d
i
n
T
a
b
l
e
1
i
n
d
i
c
a
t
e
s
s
i
g
n
i
f
i
c
a
n
t
a
d
v
a
n
c
e
m
e
n
t
s
i
n
t
h
e
s
o
p
h
i
s
t
i
c
a
t
i
o
n
o
f
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
s
,
w
i
t
h
n
o
t
a
b
l
e
m
o
d
e
l
s
l
i
k
e
I
n
c
e
p
t
i
o
n
-
R
e
s
N
e
t
-
V
2
a
n
d
R
e
s
N
e
t
-
5
0
d
e
m
o
n
s
t
r
a
t
i
n
g
e
n
h
a
n
c
e
d
p
e
r
f
o
r
m
a
n
c
e
a
c
r
o
s
s
v
a
r
i
o
u
s
d
e
f
e
c
t
c
a
t
e
g
o
r
i
e
s
.
T
h
e
s
e
a
d
v
a
n
c
e
m
e
n
t
s
h
i
g
h
l
i
g
h
t
t
h
e
r
o
b
u
s
t
n
e
s
s
a
n
d
a
d
a
p
t
a
b
i
l
i
t
y
o
f
C
N
N
s
i
n
r
e
a
l
-
t
i
m
e
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
,
a
s
s
e
e
n
w
i
t
h
m
o
d
e
l
s
s
u
c
h
a
s
F
a
s
t
e
r
R
-
C
N
N
,
w
h
i
c
h
e
f
f
e
c
t
i
v
e
l
y
i
d
e
n
t
i
f
y
d
e
f
e
c
t
s
l
i
k
e
k
n
o
t
s
,
h
o
l
e
s
,
a
n
d
s
t
a
i
n
s
i
n
i
n
d
u
s
t
r
i
a
l
a
p
p
l
i
c
a
t
i
o
n
s
.
I
n
2
0
2
4
,
t
h
e
f
o
c
u
s
h
a
s
s
h
i
f
t
e
d
t
o
w
a
r
d
s
r
e
f
i
n
i
n
g
a
d
v
a
n
c
e
d
m
o
d
e
l
s
,
p
a
r
t
i
c
u
l
a
r
l
y
I
n
c
e
p
t
i
o
n
-
R
e
s
N
e
t
-
V
2
a
n
d
i
m
p
r
o
v
e
d
v
e
r
s
i
o
n
s
o
f
R
e
s
N
e
t
-
5
0
,
u
n
d
e
r
s
c
o
r
i
n
g
o
n
g
o
i
n
g
i
n
n
o
v
a
t
i
o
n
i
n
d
e
f
e
c
t
i
d
e
n
t
i
f
i
c
a
t
i
o
n
m
e
t
h
o
d
o
l
o
g
i
e
s
.
T
h
e
p
r
e
v
i
o
u
s
y
e
a
r
s
,
2
0
2
2
a
n
d
2
0
2
3
,
a
l
s
o
r
e
f
l
e
c
t
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
r
ev
iew
o
f rece
n
t d
ee
p
lea
r
n
in
g
a
p
p
lica
tio
n
s
in
w
o
o
d
s
u
r
fa
ce
d
efec
t id
en
tifi
ca
tio
n
(
Ma
r
ti
n
a
A
li
)
1701
s
i
g
n
i
f
i
c
a
n
t
i
m
p
r
o
v
e
m
e
n
t
s
i
n
R
e
s
N
e
t
-
5
0
,
e
m
p
h
a
s
i
z
i
n
g
e
f
f
o
r
t
s
t
o
o
p
t
i
m
i
z
e
a
c
c
u
r
a
c
y
a
n
d
e
f
f
i
c
i
e
n
c
y
.
W
h
i
l
e
t
h
e
m
o
d
e
l
s
i
n
T
a
b
l
e
1
e
f
f
e
c
t
i
v
e
l
y
a
d
d
r
e
s
s
a
r
a
n
g
e
o
f
d
e
f
e
c
t
s
f
r
o
m
c
o
m
m
o
n
i
s
s
u
e
s
l
i
k
e
c
r
a
c
k
s
a
n
d
k
n
o
t
s
t
o
s
p
e
c
i
f
i
c
p
r
o
b
l
e
m
s
s
u
c
h
a
s
w
o
r
m
h
o
l
e
s
a
n
d
i
n
s
e
c
t
d
a
m
a
g
e
t
h
e
y
o
f
t
e
n
c
o
n
c
e
n
t
r
a
t
e
o
n
a
l
i
m
i
t
e
d
n
u
m
b
e
r
o
f
d
e
f
e
c
t
t
y
p
e
s
.
M
o
s
t
s
t
u
d
i
e
s
f
o
c
u
s
o
n
k
n
o
t
s
,
s
i
g
n
i
f
i
c
a
n
t
l
y
i
m
p
a
c
t
i
n
g
s
t
r
u
c
t
u
r
a
l
s
t
r
e
n
g
t
h
a
n
d
p
r
o
d
u
c
t
q
u
a
l
i
t
y
.
T
h
i
s
t
r
e
n
d
u
n
d
e
r
s
c
o
r
e
s
t
h
e
n
e
c
e
s
s
i
t
y
f
o
r
m
o
r
e
c
o
m
p
r
e
h
e
n
s
i
v
e
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
s
c
a
p
a
b
l
e
o
f
m
u
l
t
i
c
l
a
s
s
i
d
e
n
t
i
f
i
c
a
t
i
o
n
,
a
l
l
o
w
i
n
g
f
o
r
a
b
r
o
a
d
e
r
s
c
o
p
e
o
f
w
o
o
d
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
.
M
o
r
e
o
v
e
r
,
s
e
v
e
r
a
l
s
t
u
d
i
e
s
l
a
b
e
l
e
d
d
e
f
e
c
t
t
y
p
e
s
a
s
"
n
o
t
s
p
e
c
i
f
i
e
d
,
"
i
n
d
i
c
a
t
i
n
g
a
g
e
n
e
r
a
l
f
o
c
u
s
o
n
w
o
o
d
s
u
r
f
a
c
e
d
e
f
e
c
t
s
w
i
t
h
o
u
t
d
e
t
a
i
l
i
n
g
s
p
e
c
i
f
i
c
t
y
p
e
s
.
T
h
i
s
p
r
e
s
e
n
t
s
a
n
o
p
p
o
r
t
u
n
i
t
y
f
o
r
f
u
t
u
r
e
r
e
s
e
a
r
c
h
t
o
e
x
p
l
o
r
e
a
n
d
d
o
c
u
m
e
n
t
v
a
r
i
o
u
s
d
e
f
e
c
t
s
,
e
n
h
a
n
c
i
n
g
q
u
a
l
i
t
y
c
o
n
t
r
o
l
i
n
t
h
e
w
o
o
d
i
n
d
u
s
t
r
y
.
I
n
ad
d
itio
n
,
th
e
q
u
ality
o
f
th
e
tr
ain
in
g
d
ataset
is
cr
u
cial
f
o
r
ef
f
ec
tiv
ely
tr
ain
in
g
d
ee
p
lear
n
in
g
m
o
d
els.
Dee
p
n
eu
r
al
n
etwo
r
k
s
ty
p
icall
y
co
n
tain
m
an
y
tr
ain
a
b
le
p
ar
a
m
eter
s
,
m
ak
in
g
th
em
h
ig
h
l
y
p
r
o
n
e
to
o
v
er
f
itti
n
g
,
esp
ec
ially
wh
en
tr
ai
n
in
g
d
ata
is
lim
ited
.
Var
io
u
s
r
eg
u
lar
iz
atio
n
tech
n
i
q
u
es
h
a
v
e
b
ee
n
d
ev
elo
p
ed
to
a
d
d
r
ess
th
is
is
s
u
e
to
en
h
an
ce
th
e
m
o
d
e
l's
g
en
er
aliza
tio
n
ca
p
ab
ilit
y
.
On
e
ef
f
ec
tiv
e
m
et
h
o
d
is
d
ata
au
g
m
en
tatio
n
,
w
h
ich
d
ir
ec
tly
im
p
ac
ts
th
e
tr
ain
i
n
g
d
ataset.
Data
au
g
m
en
tatio
n
in
v
o
lv
es
ap
p
ly
i
n
g
v
ar
io
u
s
tr
a
n
s
f
o
r
m
atio
n
s
to
in
p
u
t
im
ag
es
to
e
x
p
an
d
th
e
tr
ain
i
n
g
d
ataset
ar
tific
ially
.
I
t
is
ess
en
tial
to
ca
r
ef
u
lly
s
elec
t
tr
an
s
f
o
r
m
atio
n
s
ap
p
r
o
p
r
iate
f
o
r
a
s
p
ec
if
ic
p
r
o
b
lem
,
as
th
e
s
e
tr
an
s
f
o
r
m
atio
n
s
s
h
o
u
ld
m
a
in
tain
th
e
s
em
an
tic
m
ea
n
in
g
o
f
th
e
lab
els.
Data
au
g
m
en
tatio
n
is
co
n
s
id
er
ed
a
s
tan
d
ar
d
ap
p
r
o
ac
h
in
d
ee
p
lear
n
in
g
,
esp
ec
ially
wh
en
f
ew
er
d
ata
s
am
p
les
ar
e
av
ailab
le.
I
t
is
wid
ely
u
s
ed
to
im
p
r
o
v
e
th
e
ad
ap
ta
b
ilit
y
o
f
n
eu
r
al
n
etwo
r
k
s
f
o
r
class
if
icatio
n
task
s
an
d
to
m
itig
ate
o
v
er
f
itti
n
g
is
s
u
es.
N
u
m
er
o
u
s
s
tu
d
ies
h
av
e
ex
p
lo
r
e
d
th
e
ef
f
ec
tiv
en
ess
o
f
d
ata
au
g
m
en
tatio
n
in
wo
o
d
d
ef
ec
t
id
en
tific
atio
n
,
d
em
o
n
s
t
r
atin
g
its
p
o
ten
tial
to
en
h
an
c
e
ac
cu
r
ac
y
an
d
ef
f
icien
cy
.
T
ab
le
2
s
u
m
m
ar
izes
v
ar
io
u
s
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es
em
p
lo
y
e
d
in
p
ast
r
esear
ch
,
h
ig
h
lig
h
tin
g
th
e
s
p
ec
if
ic
m
eth
o
d
s
u
s
ed
b
y
d
if
f
er
en
t
r
esear
ch
er
s
to
im
p
r
o
v
e
th
e
d
iv
e
r
s
ity
an
d
ef
f
ec
tiv
en
ess
o
f
th
eir
tr
ain
in
g
d
atasets
.
T
ab
le
2
.
T
ec
h
n
iq
u
es f
o
r
d
ata
a
u
g
m
en
tatio
n
in
wo
o
d
d
ef
ec
t i
d
en
tific
atio
n
R
e
se
a
r
c
h
e
r
s
D
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
[
3
3
]
M
o
r
p
h
o
l
o
g
i
c
a
l
t
r
a
n
sf
o
r
ma
t
i
o
n
s
:
x
-
r
e
f
l
e
c
t
i
o
n
a
n
d
r
o
t
a
t
i
o
n
[
1
4
]
G
e
o
me
t
r
i
c
t
r
a
n
sf
o
r
ma
t
i
o
n
s:
f
l
i
p
,
r
o
t
a
t
i
o
n
t
r
a
n
sf
o
r
mat
i
o
n
,
a
n
d
r
e
s
i
z
e
t
r
a
n
sf
o
r
mat
i
o
n
.
[
6
1
]
B
r
i
g
h
t
e
n
c
h
a
n
g
e
a
n
d
G
a
u
ssi
a
n
n
o
i
s
e
.
[
3
0
]
R
o
t
a
t
i
n
g
t
h
e
t
r
a
i
n
i
n
g
i
ma
g
e
s
a
t
9
0
°
,
1
8
0
°
,
a
n
d
2
7
0
°
,
r
a
n
d
o
m
l
y
c
r
o
p
p
i
n
g
t
h
e
t
r
a
i
n
i
n
g
i
ma
g
e
s
,
f
l
i
p
p
i
n
g
t
h
e
t
r
a
i
n
i
n
g
i
ma
g
e
s
h
o
r
i
z
o
n
t
a
l
l
y
,
v
e
r
t
i
c
a
l
l
y
,
a
n
d
d
i
a
g
o
n
a
l
l
y
,
a
d
j
u
st
i
n
g
t
h
e
b
r
i
g
h
t
n
e
s
s,
c
o
n
t
r
a
s
t
,
s
a
t
u
r
a
t
i
o
n
,
o
r
h
u
e
o
f
t
h
e
t
r
a
i
n
i
n
g
i
ma
g
e
s,
a
n
d
a
d
d
i
n
g
G
a
u
ssi
a
n
a
n
d
s
a
l
t
-
a
n
d
-
p
e
p
p
e
r
n
o
i
ses
t
o
t
h
e
i
m
a
g
e
s
R
es
e
a
r
c
h
b
y
C
h
u
n
e
t
a
l
.
[
3
3
]
h
i
g
h
l
i
g
h
t
e
d
t
h
e
e
f
f
i
c
a
c
y
o
f
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
i
n
e
n
h
a
n
c
i
n
g
t
h
e
p
e
r
f
o
r
m
a
n
c
e
o
f
C
N
Ns
f
o
r
cl
a
s
s
i
f
y
i
n
g
ti
m
b
e
r
d
e
f
e
c
ts
i
n
v
a
r
i
o
u
s
s
p
e
ci
e
s
.
T
h
e
y
e
m
p
l
o
y
e
d
m
o
r
p
h
o
l
o
g
i
c
a
l
t
r
a
n
s
f
o
r
m
a
t
i
o
n
s
,
s
u
c
h
as
x
-
r
e
f
le
c
t
i
o
n
a
n
d
r
o
t
at
i
o
n
,
t
o
e
x
p
a
n
d
th
e
t
r
a
i
n
i
n
g
d
at
a
s
e
t
,
w
h
i
c
h
a
d
d
r
e
s
s
e
d
t
h
e
c
h
a
l
l
e
n
g
es
p
o
s
e
d
b
y
l
i
m
i
t
e
d
d
at
a
a
v
ai
l
ab
i
l
i
t
y
.
T
h
e
r
e
s
u
l
ts
d
e
m
o
n
s
t
r
at
e
d
a
s
i
g
n
i
f
ic
a
n
t
i
m
p
r
o
v
e
m
e
n
t
i
n
cl
a
s
s
i
f
ic
a
t
i
o
n
p
e
r
f
o
r
m
a
n
c
e
,
w
i
t
h
t
h
e
b
es
t
m
o
d
e
l
,
R
e
s
N
e
t
5
0
,
ac
h
i
e
v
i
n
g
a
n
ac
c
u
r
a
c
y
o
f
9
4
.
5
9
%
.
S
t
at
i
s
t
i
c
al
a
n
a
l
y
s
es
,
i
n
cl
u
d
i
n
g
T
-
t
e
s
ts
,
c
o
n
f
i
r
m
e
d
t
h
a
t
t
h
e
e
n
h
a
n
c
e
m
e
n
t
s
i
n
c
l
ass
i
f
i
c
at
i
o
n
ac
c
u
r
a
c
y
d
u
e
t
o
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
w
e
r
e
s
i
g
n
i
f
i
ca
n
t
f
o
r
m
o
s
t
C
N
N
a
l
g
o
r
it
h
m
s
t
es
ted
,
e
x
c
e
p
t
f
o
r
A
l
e
x
N
e
t
.
F
u
r
t
h
e
r
m
o
r
e
,
t
h
e
a
u
t
h
o
r
s
o
b
s
e
r
v
e
d
t
h
at
s
i
m
p
l
y
i
n
c
r
e
a
s
i
n
g
t
h
e
n
u
m
b
e
r
o
f
t
r
a
i
n
i
n
g
e
p
o
c
h
s
a
n
d
t
h
e
l
e
a
r
n
i
n
g
r
at
e
d
i
d
n
o
t
n
e
c
e
s
s
a
r
il
y
y
i
el
d
b
e
t
t
e
r
c
l
as
s
i
f
i
c
a
ti
o
n
p
r
e
c
i
s
i
o
n
,
u
n
d
e
r
s
c
o
r
i
n
g
t
h
e
i
m
p
o
r
t
a
n
c
e
o
f
c
a
r
e
f
u
l
l
y
t
u
n
i
n
g
h
y
p
e
r
p
a
r
a
m
e
t
e
r
s
f
o
r
o
p
t
i
m
a
l
r
es
u
l
ts
.
O
v
e
r
a
l
l
,
t
h
e
s
t
u
d
y
c
o
n
c
l
u
d
e
d
t
h
a
t
d
a
t
a
a
u
g
m
e
n
t
a
ti
o
n
w
a
s
a
v
i
t
al
t
ec
h
n
i
q
u
e
f
o
r
o
v
e
r
c
o
m
i
n
g
t
h
e
l
i
m
it
a
t
i
o
n
s
o
f
s
m
a
ll
d
a
t
as
e
ts
a
n
d
i
m
p
r
o
v
i
n
g
t
h
e
c
l
a
s
s
i
f
ic
a
t
i
o
n
c
ap
a
b
i
l
i
t
ie
s
o
f
C
N
Ns
i
n
t
h
e
c
o
n
t
ex
t
o
f
t
i
m
b
e
r
d
e
f
e
c
t
i
d
e
n
ti
f
i
c
at
i
o
n
.
N
e
x
t
,
U
r
b
o
n
a
s
e
t
a
l
.
[
1
4
]
d
is
c
u
s
s
e
d
t
h
e
i
m
p
o
r
t
a
n
c
e
o
f
d
at
a
a
u
g
m
e
n
t
a
t
i
o
n
i
n
e
n
h
a
n
c
i
n
g
t
h
e
p
e
r
f
o
r
m
a
n
c
e
o
f
AVI
s
y
s
t
e
m
s
f
o
r
i
d
e
n
ti
f
y
i
n
g
d
e
f
e
c
t
s
o
n
w
o
o
d
v
e
n
e
e
r
s
u
r
f
a
ce
s
.
T
h
e
y
h
i
g
h
l
i
g
h
t
e
d
t
h
at
t
r
a
d
i
ti
o
n
a
l
v
i
s
u
a
l
q
u
al
i
ty
i
n
s
p
e
c
ti
o
n
s
i
n
t
h
e
l
u
m
b
e
r
a
n
d
w
o
o
d
p
r
o
c
e
s
s
i
n
g
i
n
d
u
s
t
r
y
w
e
r
e
o
f
t
e
n
p
e
r
f
o
r
m
e
d
b
y
h
u
m
a
n
o
p
e
r
a
t
o
r
s
,
l
ea
d
i
n
g
t
o
p
o
t
e
n
t
i
a
l
e
r
r
o
r
s
d
u
e
t
o
t
h
e
t
e
d
i
o
u
s
n
a
t
u
r
e
o
f
t
h
e
t
a
s
k
.
T
o
i
m
p
r
o
v
e
a
c
c
u
r
a
c
y
a
n
d
s
p
e
e
d
,
t
h
e
a
u
th
o
r
s
p
r
o
p
o
s
e
d
u
s
i
n
g
a
F
a
s
t
e
r
R
-
C
N
N
al
o
n
g
s
i
d
e
d
a
t
a
a
u
g
m
e
n
t
a
ti
o
n
a
n
d
t
r
a
n
s
f
er
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
.
T
h
e
au
t
h
o
r
s
s
p
e
c
i
f
i
c
a
ll
y
m
e
n
t
i
o
n
e
d
e
m
p
l
o
y
i
n
g
a
s
y
n
th
e
t
i
c
a
ll
y
a
u
g
m
e
n
t
e
d
d
a
t
as
e
t
to
t
r
a
i
n
t
h
e
i
r
m
o
d
el
s
,
w
h
i
c
h
in
c
l
u
d
e
d
p
r
e
-
t
r
a
i
n
e
d
n
e
t
w
o
r
k
s
l
i
k
e
A
l
e
x
N
et
,
VGG
1
6
,
B
N
I
n
c
e
p
t
i
o
n
,
a
n
d
R
esN
e
t
1
5
2
.
T
h
e
i
r
e
x
p
e
r
i
m
e
n
ts
s
h
o
w
e
d
t
h
at
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
s
i
g
n
i
f
i
c
a
n
t
l
y
c
o
n
t
r
i
b
u
t
e
d
t
o
t
h
e
o
v
e
r
a
ll
cl
a
s
s
i
f
i
ca
t
i
o
n
a
c
c
u
r
a
c
y
o
f
d
e
f
e
c
t
i
d
e
n
tif
i
c
a
t
i
o
n
,
ac
h
i
e
v
i
n
g
a
n
a
v
e
r
a
g
e
a
c
c
u
r
a
c
y
o
f
8
0
.
6
%
w
i
t
h
R
es
N
et
1
5
2
a
n
d
a
n
i
m
p
r
e
s
s
i
v
e
9
6
.
1
%
w
h
e
n
c
o
m
b
i
n
i
n
g
a
l
l
d
e
f
e
c
t
c
l
as
s
es.
T
h
i
s
i
n
d
i
ca
t
e
d
t
h
at
d
a
t
a
a
u
g
m
en
t
a
t
i
o
n
e
f
f
e
c
ti
v
e
l
y
e
n
h
a
n
c
e
d
t
h
e
r
o
b
u
s
t
n
e
s
s
o
f
t
h
e
m
o
d
e
ls
,
m
ak
i
n
g
t
h
e
m
s
u
i
t
a
b
l
e
f
o
r
r
e
a
l
-
t
i
m
e
i
n
d
u
s
t
r
i
al
a
p
p
l
i
ca
t
i
o
n
s
.
M
o
r
e
o
v
e
r
,
t
h
e
a
u
t
h
o
r
s
a
c
k
n
o
w
l
e
d
g
e
d
t
h
a
t
w
h
i
l
e
d
a
ta
a
u
g
m
e
n
t
a
t
i
o
n
a
n
d
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
i
m
p
r
o
v
e
d
d
ef
e
c
t
d
e
t
ec
t
i
o
n
,
t
h
e
m
e
t
h
o
d
'
s
r
e
l
i
a
n
c
e
o
n
m
a
n
u
a
l
l
y
l
a
b
e
l
e
d
im
a
g
e
s
f
o
r
t
r
a
i
n
i
n
g
p
r
e
s
e
n
t
e
d
a
li
m
i
t
at
i
o
n
,
a
s
t
h
e
s
e
l
a
b
e
ls
m
i
g
h
t
n
o
t
a
lw
a
y
s
b
e
a
c
c
u
r
a
t
e
.
T
h
e
y
s
u
g
g
es
t
e
d
t
h
at
f
u
t
u
r
e
w
o
r
k
c
o
u
l
d
e
x
p
l
o
r
e
m
o
r
e
c
o
m
p
l
e
x
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
a
n
d
ap
p
l
y
t
h
e
s
e
m
e
t
h
o
d
s
t
o
a
n
a
l
y
ze
s
u
r
f
a
c
e
d
e
f
e
c
ts
i
n
o
t
h
e
r
t
y
p
e
s
o
f
w
o
o
d
p
a
n
el
s
,
t
h
e
r
e
b
y
b
r
o
a
d
e
n
i
n
g
t
h
e
s
c
o
p
e
o
f
t
h
e
i
r
a
p
p
r
o
a
c
h
.
O
v
e
r
a
ll
,
t
h
e
au
t
h
o
r
s
e
m
p
h
a
s
i
z
e
d
t
h
a
t
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
w
a
s
c
r
u
c
i
a
l
i
n
d
e
v
el
o
p
i
n
g
e
f
f
e
c
t
i
v
e
au
t
o
m
a
t
e
d
i
n
s
p
ec
t
i
o
n
s
y
s
te
m
s
f
o
r
w
o
o
d
p
r
o
c
e
s
s
i
n
g
,
d
e
m
o
n
s
t
r
a
t
i
n
g
it
s
a
p
p
li
c
a
b
i
li
t
y
t
o
v
a
r
i
o
u
s
i
n
d
u
s
t
r
i
a
l
m
at
e
r
i
a
ls
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
1
6
9
6
-
1
7
0
7
1702
T
h
en
,
R
o
m
an
o
v
s
k
is
et
a
l.
[
6
1
]
d
is
cu
s
s
ed
th
e
im
p
o
r
tan
ce
o
f
d
ata
au
g
m
en
tatio
n
in
o
p
tim
izin
g
th
e
u
s
e
o
f
o
ak
wo
o
d
d
u
r
in
g
in
d
u
s
tr
ial
p
r
o
ce
s
s
in
g
.
T
h
e
au
th
o
r
tr
ain
e
d
two
d
if
f
er
en
t
Ma
s
k
R
-
C
NN
m
o
d
els
u
s
in
g
th
e
in
s
tan
ce
s
eg
m
en
tatio
n
m
eth
o
d
.
T
h
e
f
i
r
s
t
m
o
d
el
was
tr
ain
e
d
s
o
lely
o
n
t
h
e
o
r
i
g
in
al
d
ata
o
b
tain
ed
f
r
o
m
th
e
s
ca
n
n
er
.
I
n
c
o
n
tr
ast,
th
e
s
ec
o
n
d
m
o
d
el
in
cl
u
d
ed
t
h
e
o
r
ig
i
n
al
im
ag
es a
n
d
ad
d
itio
n
al
ar
tific
ia
lly
ad
ju
s
ted
im
ag
es
to
ex
p
a
n
d
th
e
d
ataset.
T
h
e
au
th
o
r
s
em
p
lo
y
e
d
two
d
ata
au
g
m
en
tatio
n
m
eth
o
d
s
“b
r
ig
h
ten
ch
a
n
g
e”
an
d
“Ga
u
s
s
ian
n
o
is
e”
wh
ich
wer
e
s
elec
ted
b
ec
au
s
e
th
ey
d
id
n
o
t
alter
th
e
p
h
y
s
ical
lo
ca
tio
n
o
f
o
b
jects
with
in
th
e
im
ag
es.
All
n
ew
im
ag
es
wer
e
g
en
e
r
ated
b
ased
o
n
lab
eled
im
ag
es,
elim
in
atin
g
th
e
n
ee
d
f
o
r
r
ela
b
elin
g
,
as
th
e
o
r
ig
in
al
im
ag
e
J
SON
f
ile
was
attac
h
ed
to
th
e
ar
tific
ially
g
e
n
er
ated
im
ag
es.
T
h
is
ap
p
r
o
ac
h
s
av
ed
s
ig
n
if
ican
t
tim
e
in
th
e
m
ar
k
in
g
p
r
o
ce
s
s
.
T
h
e
r
esu
lts
in
d
icate
d
th
at
th
e
o
r
ig
in
al
m
o
d
el
ac
h
iev
e
d
an
a
v
er
ag
e
ac
cu
r
ac
y
o
f
7
3
%.
Ho
wev
e
r
,
th
e
m
o
d
el
th
at
u
tili
ze
d
d
ata
au
g
m
en
tatio
n
s
h
o
wed
an
im
p
r
o
v
em
e
n
t
o
f
a
p
p
r
o
x
im
ately
1
6
%,
r
ea
ch
in
g
an
av
er
a
g
e
ac
c
u
r
ac
y
o
f
ar
o
u
n
d
8
9
%.
T
h
e
a
u
th
o
r
s
c
o
n
clu
d
e
d
th
at
th
e
au
g
m
e
n
ted
m
o
d
el
s
u
cc
ess
f
u
lly
id
en
tifie
d
an
d
lo
ca
lized
d
ef
ec
ts
s
p
ec
if
ic
to
o
ak
wo
o
d
,
d
em
o
n
s
tr
atin
g
th
at
d
ata
au
g
m
en
ta
tio
n
is
a
v
alu
ab
le
tech
n
iq
u
e
f
o
r
e
n
h
an
cin
g
m
o
d
e
l p
er
f
o
r
m
an
ce
in
t
h
e
co
n
te
x
t o
f
wo
o
d
p
r
o
ce
s
s
in
g
.
L
astl
y
,
He
et
a
l.
[
3
0
]
d
is
cu
s
s
ed
th
e
r
o
le
o
f
d
ata
a
u
g
m
e
n
tatio
n
in
th
eir
p
r
o
p
o
s
ed
m
eth
o
d
f
o
r
d
etec
tin
g
wo
o
d
f
ea
tu
r
es
an
d
class
if
y
in
g
d
ef
ec
ts
f
r
o
m
im
ag
es
co
llec
ted
u
s
in
g
a
laser
s
ca
n
n
er
.
I
n
th
eir
ap
p
r
o
ac
h
,
th
e
au
th
o
r
s
im
p
lem
e
n
ted
d
ata
au
g
m
en
tatio
n
as
a
s
tr
ateg
y
to
a
v
o
id
o
v
er
f
itti
n
g
d
u
r
i
n
g
th
e
tr
a
in
in
g
o
f
th
e
DC
NN
m
o
d
el.
T
h
ey
c
o
llected
a
d
ata
s
et
co
m
p
r
is
in
g
im
a
g
es
o
f
6
0
0
p
iece
s
o
f
r
ed
an
d
ca
m
p
h
o
r
p
in
e
wo
o
d
,
wh
ich
in
clu
d
ed
s
p
ec
if
ic
d
ef
ec
t
ca
teg
o
r
ies
s
u
ch
as
k
n
o
ts
,
cr
ac
k
s
,
an
d
m
ild
ew
s
tain
s
.
T
h
e
d
ataset
was
d
iv
id
ed
in
to
tr
ain
in
g
,
v
alid
atio
n
,
a
n
d
test
in
g
s
u
b
s
ets,
with
d
ata
au
g
m
e
n
ta
tio
n
ap
p
lied
t
o
en
h
a
n
ce
th
e
s
i
ze
an
d
d
i
v
er
s
ity
o
f
th
e
tr
ain
in
g
d
ata.
T
h
e
r
esu
lts
d
em
o
n
s
tr
ated
th
at
th
e
DC
NN
m
o
d
el
ac
h
iev
ed
an
o
v
e
r
all
ac
cu
r
ac
y
o
f
9
9
.
1
3
%,
with
o
n
ly
1
.
1
2
s
ec
o
n
d
s
r
eq
u
ir
ed
f
o
r
d
etec
tio
n
,
in
clu
d
in
g
im
ag
e
p
r
ep
r
o
ce
s
s
in
g
an
d
id
en
tific
atio
n
.
Data
au
g
m
en
tatio
n
s
ig
n
if
ican
tly
c
o
n
tr
ib
u
ted
to
th
e
m
o
d
el
’
s
p
er
f
o
r
m
an
ce
,
allo
win
g
it
to
r
ec
o
g
n
ize
wo
o
d
d
ef
ec
ts
m
o
r
e
ac
cu
r
ately
an
d
ef
f
ec
tiv
el
y
th
an
c
o
n
v
e
n
tio
n
al
m
eth
o
d
s
.
B
y
in
co
r
p
o
r
atin
g
tr
an
s
f
er
lear
n
in
g
,
t
h
e
m
o
d
el
b
en
ef
ited
f
r
o
m
a
p
r
e
-
tr
ain
in
g
p
h
ase
th
at
f
ac
ilit
ated
m
o
r
e
ef
f
ec
tiv
e
lear
n
i
n
g
f
r
o
m
th
e
lim
ited
av
ailab
le
d
ata.
I
n
th
e
r
ap
id
ly
ev
o
lv
i
n
g
f
ield
o
f
d
ee
p
lear
n
in
g
,
tr
an
s
f
e
r
lear
n
in
g
h
as
em
er
g
ed
as
a
p
iv
o
tal
tech
n
iq
u
e,
p
ar
ticu
lar
ly
f
o
r
task
s
in
v
o
l
v
in
g
lim
ited
lab
el
ed
d
ata.
Nu
m
er
o
u
s
p
r
e
-
tr
ain
ed
C
NN
m
o
d
els
ar
e
r
ea
d
ily
av
ailab
le
i
n
th
e
liter
at
u
r
e,
p
r
o
v
id
i
n
g
s
ig
n
if
ica
n
t
ad
v
an
tag
es
in
lear
n
in
g
co
m
p
r
eh
e
n
s
iv
e
v
is
u
al
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
.
T
h
ese
m
o
d
els
h
av
e
b
ee
n
ex
ten
s
iv
ely
tr
ain
e
d
o
n
lar
g
e
d
atasets
,
allo
win
g
th
em
to
ca
p
tu
r
e
c
o
m
p
lex
p
atter
n
s
a
n
d
in
tr
icate
f
ea
tu
r
es,
wh
ich
is
esp
ec
ially
b
e
n
ef
icial
wh
e
n
wo
r
k
in
g
with
s
m
all,
task
-
s
p
ec
if
ic
tr
ain
in
g
d
atasets
[
6
2
]
.
L
e
v
er
ag
in
g
tr
an
s
f
er
lear
n
in
g
e
n
a
b
les
r
esear
ch
er
s
to
ad
ap
t
p
r
e
-
tr
ain
ed
m
o
d
els
to
m
ee
t
th
eir
s
p
ec
if
ic
n
ee
d
s
,
o
f
f
er
i
n
g
a
p
r
ag
m
atic
an
d
ef
f
icien
t
a
p
p
r
o
ac
h
f
o
r
tr
ain
in
g
d
ee
p
lear
n
i
n
g
m
o
d
els
in
s
ce
n
a
r
io
s
with
lim
ited
lab
eled
d
ata.
T
h
is
m
eth
o
d
o
l
o
g
y
allo
ws
p
r
ac
titi
o
n
er
s
to
m
o
d
i
f
y
an
d
f
in
e
-
t
u
n
e
p
r
e
-
tr
ain
e
d
m
o
d
els
u
s
in
g
d
o
m
ain
-
s
p
ec
if
ic
in
f
o
r
m
atio
n
,
s
ig
n
if
ican
tly
ac
ce
l
er
atin
g
th
e
lear
n
i
n
g
p
r
o
ce
s
s
f
o
r
im
ag
e
class
if
icatio
n
an
d
id
en
tific
atio
n
task
s
.
I
n
s
tead
o
f
s
tar
tin
g
t
h
e
tr
ain
in
g
p
r
o
ce
s
s
f
r
o
m
s
cr
atch
,
tr
an
s
f
er
lear
n
in
g
u
tili
ze
s
ex
is
tin
g
m
o
d
els
tr
ain
ed
o
n
lar
g
er
d
atasets
an
d
ad
ju
s
ts
th
eir
weig
h
ts
to
alig
n
with
th
e
in
ten
d
ed
task
.
T
h
is
tech
n
iq
u
e
en
h
an
ce
s
b
o
t
h
ac
cu
r
ac
y
an
d
ef
f
ec
tiv
en
ess
ac
r
o
s
s
v
ar
io
u
s
ap
p
licatio
n
s
[
3
3
]
.
T
r
a
i
n
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
s
f
r
o
m
s
c
r
a
t
c
h
c
a
n
b
e
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
e
x
p
e
n
s
i
v
e
;
t
h
e
r
e
f
o
r
e
,
u
t
i
l
i
z
i
n
g
m
o
d
e
l
s
a
l
r
e
a
d
y
p
u
b
l
i
s
h
e
d
i
n
t
h
e
l
i
t
e
r
a
t
u
r
e
c
a
n
s
a
v
e
t
i
m
e
a
n
d
r
e
s
o
u
r
c
e
s
w
h
i
l
e
a
c
h
i
e
v
i
n
g
c
o
m
p
a
r
a
b
l
e
p
e
r
f
o
r
m
a
n
c
e
i
n
c
o
m
p
u
t
e
r
v
i
s
i
o
n
t
a
s
k
s
[
6
3
]
.
S
e
v
e
r
a
l
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
m
o
d
e
l
s
,
s
u
c
h
a
s
X
c
e
p
t
i
o
n
,
I
n
c
e
p
t
i
o
n
,
A
l
e
x
N
e
t
,
V
G
G
N
e
t
,
G
o
o
g
L
e
N
e
t
,
a
n
d
R
e
s
N
e
t
,
h
a
v
e
b
e
e
n
p
r
o
p
o
s
e
d
f
o
r
o
b
j
e
c
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
o
v
e
r
t
h
e
p
a
s
t
f
e
w
d
e
c
a
d
e
s
.
M
o
r
e
o
v
e
r
,
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
s
t
r
a
t
e
g
i
e
s
h
a
v
e
b
e
e
n
w
i
d
e
l
y
a
p
p
l
i
e
d
i
n
w
o
o
d
s
u
r
f
a
c
e
d
e
f
e
c
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
.
A
d
d
i
t
i
o
n
a
l
l
y
,
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
h
a
s
p
r
o
v
e
n
e
f
f
e
c
t
i
v
e
a
c
r
o
s
s
v
a
r
i
o
u
s
d
o
m
a
i
n
s
,
i
n
c
l
u
d
i
n
g
m
a
n
u
f
a
c
t
u
r
i
n
g
,
m
e
d
i
c
a
l
d
i
a
g
n
o
s
t
i
c
s
,
a
n
d
b
a
g
g
a
g
e
s
c
r
e
e
n
i
n
g
[
6
4
]
–
[
6
6
]
.
T
h
i
s
a
p
p
r
o
a
c
h
e
l
i
m
i
n
a
t
e
s
t
h
e
n
e
e
d
f
o
r
e
x
t
e
n
s
i
v
e
d
a
t
a
s
e
t
s
a
n
d
m
i
n
i
m
i
z
e
s
t
h
e
p
r
o
l
o
n
g
e
d
t
r
a
i
n
i
n
g
p
e
r
i
o
d
s
t
y
p
i
c
a
l
l
y
a
s
s
o
c
i
a
t
e
d
w
i
t
h
d
e
v
e
l
o
p
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m
s
f
r
o
m
s
c
r
a
t
c
h
[
6
7
]
,
[
6
8
]
.
Sev
er
al
s
tu
d
ies
ef
f
ec
tiv
ely
u
ti
lized
tr
an
s
f
er
lear
n
i
n
g
to
i
d
e
n
tify
wo
o
d
d
ef
ec
ts
.
Ur
b
o
n
as
et
a
l.
[
1
4
]
d
is
cu
s
s
ed
tr
an
s
f
er
lear
n
in
g
as
a
cr
itical
tech
n
iq
u
e
in
d
ev
el
o
p
in
g
an
AVI
s
y
s
tem
f
o
r
d
etec
ti
n
g
d
ef
ec
ts
o
n
wo
o
d
v
en
ee
r
s
u
r
f
ac
es.
B
y
em
p
lo
y
in
g
p
r
e
-
tr
ain
ed
n
eu
r
al
n
etwo
r
k
m
o
d
els
s
u
ch
as
Alex
Net,
VG
G1
6
,
B
NI
n
ce
p
tio
n
,
an
d
R
esNet1
5
2
,
th
e
r
esear
ch
e
r
s
aim
ed
to
en
h
an
ce
th
e
ac
cu
r
ac
y
o
f
th
e
f
aster
R
-
C
NN
m
o
d
el
f
o
r
th
is
s
p
ec
if
ic
task
.
T
r
an
s
f
er
lear
n
i
n
g
allo
wed
th
em
to
lev
er
a
g
e
th
e
lear
n
ed
f
ea
tu
r
es
f
r
o
m
lar
g
e
d
atasets
,
r
esu
ltin
g
in
th
e
b
est
av
er
ag
e
ac
cu
r
ac
y
o
f
8
0
.
6
%
u
s
in
g
R
esNet1
5
2
an
d
a
r
em
a
r
k
ab
le
9
6
.
1
%
ac
c
u
r
ac
y
wh
e
n
c
o
m
b
in
in
g
all
d
ef
ec
t
class
es.
Ad
d
itio
n
ally
,
d
ata
au
g
m
en
tatio
n
was
u
s
ed
to
in
cr
ea
s
e
th
e
d
iv
er
s
ity
o
f
th
e
tr
ain
in
g
d
ataset,
f
u
r
th
er
im
p
r
o
v
in
g
m
o
d
el
p
er
f
o
r
m
a
n
c
e.
T
h
e
f
in
d
in
g
s
s
u
g
g
ested
th
at
th
ese
m
eth
o
d
s
co
u
ld
ap
p
ly
to
o
th
er
in
d
u
s
tr
ial
m
ater
ials
,
s
h
o
wca
s
in
g
th
e
v
er
s
atility
o
f
tr
an
s
f
er
lear
n
in
g
.
H
o
wev
er
,
t
h
e
r
elian
ce
o
n
m
an
u
ally
lab
eled
im
a
g
es
f
o
r
t
r
ain
in
g
p
o
s
ed
a
lim
itatio
n
,
as
lab
elin
g
er
r
o
r
s
c
o
u
ld
im
p
ac
t
r
esu
lts
.
T
h
e
au
th
o
r
s
p
la
n
n
ed
to
ex
p
l
o
r
e
m
o
r
e
co
m
p
lex
tr
an
s
f
er
lear
n
in
g
an
d
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
e
s
in
f
u
tu
r
e
r
esear
ch
an
d
ex
ten
d
th
eir
m
eth
o
d
s
to
an
aly
ze
s
u
r
f
ac
e
d
ef
ec
ts
in
o
th
e
r
ty
p
es o
f
w
o
o
d
p
an
els.
Nex
t,
Gao
et
a
l.
[
6
9
]
d
is
cu
s
s
ed
tr
an
s
f
er
lear
n
in
g
as
an
in
teg
r
al
co
m
p
o
n
e
n
t
o
f
th
eir
p
r
o
p
o
s
ed
m
o
d
el,
Re
s
Net
-
1
8
,
f
o
r
d
etec
tin
g
wo
o
d
k
n
o
t
d
ef
ec
ts
.
I
n
th
is
s
tu
d
y
,
tr
an
s
f
er
lear
n
in
g
was
lev
er
a
g
ed
to
en
h
an
ce
th
e
p
er
f
o
r
m
an
ce
o
f
d
ee
p
lear
n
in
g
tech
n
iq
u
es
in
th
e
c
o
n
tex
t
o
f
wo
o
d
d
ef
ec
t
d
etec
tio
n
,
w
h
ich
h
ad
h
is
to
r
ically
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
r
ev
iew
o
f rece
n
t d
ee
p
lea
r
n
in
g
a
p
p
lica
tio
n
s
in
w
o
o
d
s
u
r
fa
ce
d
efec
t id
en
tifi
ca
tio
n
(
Ma
r
ti
n
a
A
li
)
1703
f
ac
ed
ch
allen
g
es
s
u
ch
as
lo
n
g
tr
ain
in
g
tim
es,
lo
w
r
ec
o
g
n
i
tio
n
ac
cu
r
ac
y
,
an
d
th
e
n
ee
d
f
o
r
m
an
u
al
f
ea
tu
r
e
ex
tr
ac
tio
n
.
I
n
teg
r
atin
g
tr
a
n
s
f
er
lear
n
in
g
with
th
e
"sq
u
ee
ze
-
an
d
-
ex
citatio
n
"
(
SE)
m
o
d
u
le
i
n
to
th
e
r
esid
u
al
b
asic
b
lo
ck
s
tr
u
ctu
r
e,
t
h
e
R
e
s
Net
-
1
8
m
o
d
el
im
p
r
o
v
e
d
f
ea
tu
r
e
e
x
tr
ac
tio
n
in
th
e
ch
a
n
n
el
d
im
en
s
io
n
an
d
ef
f
ec
tiv
ely
f
u
s
ed
f
ea
tu
r
es
ac
r
o
s
s
m
u
ltip
le
s
ca
les.
T
h
e
ex
p
er
im
en
tal
r
esu
lts
in
d
icate
d
th
at
th
e
m
o
d
el
ac
h
iev
ed
an
im
p
r
ess
iv
e
ac
cu
r
ac
y
o
f
9
9
.
0
2
%
in
d
etec
tin
g
wo
o
d
k
n
o
t
d
ef
ec
ts
,
s
ig
n
if
ican
tly
h
ig
h
er
th
an
th
e
class
ical
R
esNet
-
1
8
m
o
d
el'
s
ac
cu
r
ac
y
o
f
9
0
.
8
3
%.
Ad
d
itio
n
ally
,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
r
e
d
u
ce
d
tr
ain
in
g
tim
e
an
d
elim
in
ated
th
e
n
ee
d
f
o
r
ex
t
en
s
iv
e
im
ag
e
p
r
ep
r
o
ce
s
s
in
g
an
d
m
an
u
al
f
ea
tu
r
e
ex
tr
ac
ti
o
n
,
th
er
e
b
y
g
r
ea
tly
en
h
an
cin
g
r
ec
o
g
n
itio
n
ef
f
icie
n
cy
.
Ov
er
all,
th
e
au
th
o
r
co
n
clu
d
ed
th
at
th
e
R
e
s
Net
-
1
8
m
o
d
el,
th
r
o
u
g
h
its
tr
an
s
f
er
lear
n
i
n
g
,
p
r
o
v
id
ed
a
p
r
o
m
is
in
g
s
o
lu
tio
n
f
o
r
wo
o
d
n
o
n
d
estru
ctiv
e
test
in
g
a
n
d
d
e
f
ec
t
id
en
tific
atio
n
,
p
av
in
g
t
h
e
way
f
o
r
m
o
r
e
ef
f
ici
en
t w
o
o
d
k
n
o
t
d
ef
ec
t d
etec
tio
n
in
f
u
t
u
r
e
ap
p
licatio
n
s
.
I
n
a
d
d
i
t
i
o
n
,
H
u
e
t
a
l
.
[
5
0
]
d
i
s
c
u
s
s
e
d
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
a
s
a
c
r
u
c
i
a
l
c
o
m
p
o
n
e
n
t
o
f
t
h
e
i
r
p
r
o
p
o
s
e
d
m
e
t
h
o
d
f
o
r
i
d
e
n
t
i
f
y
i
n
g
w
o
o
d
d
e
f
e
c
t
s
i
n
p
o
p
l
a
r
v
e
n
e
e
r
u
s
i
n
g
a
c
o
m
b
i
n
a
t
i
o
n
o
f
a
p
r
o
g
r
e
s
s
i
v
e
g
r
o
w
i
n
g
g
e
n
e
r
a
t
i
v
e
a
d
v
e
r
s
a
r
i
a
l
n
e
t
w
o
r
k
(
P
G
G
A
N
)
a
n
d
t
h
e
M
a
s
k
R
-
C
N
N
m
o
d
e
l
.
I
n
t
h
i
s
s
t
u
d
y
,
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
w
a
s
e
m
p
l
o
y
e
d
t
o
e
n
h
a
n
c
e
t
h
e
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
e
M
a
s
k
R
-
C
N
N
m
o
d
e
l
,
s
p
e
c
i
f
i
c
a
l
l
y
b
y
i
n
t
e
g
r
a
t
i
n
g
i
t
w
i
t
h
a
c
l
a
s
s
i
f
i
e
r
l
a
y
e
r
.
T
h
i
s
a
p
p
r
o
a
c
h
a
l
l
o
w
e
d
t
h
e
m
o
d
e
l
t
o
l
e
a
r
n
f
r
o
m
a
p
r
e
-
t
r
a
i
n
e
d
n
e
t
w
o
r
k
,
i
m
p
r
o
v
i
n
g
i
t
s
a
b
i
l
i
t
y
t
o
i
d
e
n
t
i
f
y
a
n
d
c
l
a
s
s
i
f
y
d
e
f
e
c
t
s
s
u
c
h
a
s
d
e
a
d
k
n
o
t
s
,
l
i
v
e
k
n
o
t
s
,
a
n
d
i
n
s
e
c
t
h
o
l
e
s
i
n
t
h
e
v
e
n
e
e
r
.
T
h
e
e
x
p
e
r
i
m
e
n
t
a
l
r
e
s
u
l
t
s
i
n
d
i
c
a
t
e
d
t
h
a
t
t
h
e
m
o
d
e
l
a
c
h
i
e
v
e
d
h
i
g
h
a
c
c
u
r
a
c
y
r
a
t
e
s
o
f
9
9
.
0
5
%
f
o
r
l
i
v
e
k
n
o
t
s
,
9
7
.
0
5
%
f
o
r
d
e
a
d
k
n
o
t
s
,
a
n
d
9
9
.
1
0
%
f
o
r
i
n
s
e
c
t
h
o
l
e
s
,
d
e
m
o
n
s
t
r
a
t
i
n
g
t
h
e
e
f
f
e
c
t
i
v
e
n
e
s
s
o
f
t
h
e
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
s
t
r
a
t
e
g
y
i
n
a
c
h
i
e
v
i
n
g
p
r
e
c
i
s
e
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
.
T
h
e
a
u
t
h
o
r
h
i
g
h
l
i
g
h
t
s
t
h
a
t
t
h
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
o
f
t
h
e
M
a
s
k
R
-
C
N
N
m
o
d
e
l
,
w
h
i
c
h
w
a
s
b
a
s
e
d
o
n
R
e
s
N
e
t
5
0
a
n
d
f
u
r
t
h
e
r
e
n
h
a
n
c
e
d
t
h
r
o
u
g
h
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
,
r
e
a
c
h
e
d
9
8
.
4
%
w
h
e
n
t
e
s
t
e
d
o
n
a
n
e
x
p
a
n
d
e
d
d
a
t
a
s
e
t
.
A
d
d
i
t
i
o
n
a
l
l
y
,
t
h
e
u
s
e
o
f
P
G
G
A
N
f
o
r
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
h
e
l
p
e
d
t
o
i
m
p
r
o
v
e
t
h
e
d
i
v
e
r
s
i
t
y
o
f
d
e
f
e
c
t
i
m
a
g
e
s
a
n
d
b
a
l
a
n
c
e
s
a
m
p
l
e
d
i
s
t
r
i
b
u
t
i
o
n
,
c
o
n
t
r
i
b
u
t
i
n
g
t
o
t
h
e
o
v
e
r
a
l
l
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
e
d
e
t
e
c
t
i
o
n
m
o
d
e
l
.
T
h
e
f
i
n
d
i
n
g
s
s
u
g
g
e
s
t
t
h
a
t
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
c
a
n
s
i
g
n
i
f
i
c
a
n
t
l
y
e
n
h
a
n
c
e
d
e
t
e
c
t
i
o
n
a
c
c
u
r
a
c
y
,
e
s
p
e
c
i
a
l
l
y
w
h
e
n
w
o
r
k
i
n
g
w
i
t
h
s
m
a
l
l
e
r
d
a
t
a
s
e
t
s
.
T
h
e
a
u
t
h
o
r
c
o
n
c
l
u
d
e
s
t
h
a
t
t
h
i
s
t
e
c
h
n
o
l
o
g
y
h
a
s
p
o
t
e
n
t
i
a
l
a
p
p
l
i
c
a
t
i
o
n
s
i
n
w
o
o
d
p
r
o
c
e
s
s
i
n
g
e
q
u
i
p
m
e
n
t
,
p
a
r
t
i
c
u
l
a
r
l
y
i
n
w
o
o
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
s
y
s
t
e
m
s
,
i
n
d
i
c
a
t
i
n
g
a
p
r
o
m
i
s
i
n
g
a
v
e
n
u
e
f
o
r
f
u
t
u
r
e
i
m
p
l
e
m
e
n
t
a
t
i
o
n
.
Als
o
,
Din
g
et
a
l.
[
1
]
d
is
cu
s
s
ed
tr
an
s
f
er
lear
n
in
g
as a
k
ey
s
tr
a
teg
y
in
th
eir
ap
p
r
o
ac
h
to
d
etec
tin
g
wo
o
d
d
ef
ec
ts
,
s
p
ec
if
ically
i
n
s
o
lid
wo
o
d
b
o
ar
d
s
.
T
h
is
s
tu
d
y
ap
p
l
ied
th
e
tr
an
s
f
er
lear
n
in
g
m
eth
o
d
to
th
e
s
in
g
le
-
s
h
o
t
m
u
lti
-
b
o
x
d
etec
to
r
(
SS
D)
,
a
ta
r
g
et
d
etec
tio
n
alg
o
r
ith
m
,
u
s
in
g
th
e
Den
s
eNe
t
n
etwo
r
k
to
en
h
an
ce
p
e
r
f
o
r
m
an
ce
.
T
h
e
r
esu
lts
in
d
icate
d
a
m
ea
n
av
er
ag
e
p
r
ec
is
io
n
o
f
9
6
.
1
%
f
o
r
d
etec
tin
g
th
e
th
r
ee
ty
p
es
o
f
d
ef
ec
ts
,
s
h
o
wca
s
in
g
th
e
ef
f
ec
tiv
e
n
ess
o
f
tr
a
n
s
f
er
le
ar
n
in
g
in
im
p
r
o
v
i
n
g
d
etec
tio
n
ac
cu
r
ac
y
.
T
h
e
au
th
o
r
s
em
p
h
a
s
ized
th
e
p
o
ten
tial
o
f
tr
a
n
s
f
er
lear
n
in
g
an
d
ad
v
a
n
ce
d
n
e
u
r
al
n
etwo
r
k
ar
ch
itectu
r
es
in
en
h
an
cin
g
wo
o
d
d
ef
ec
t
d
etec
tio
n
m
eth
o
d
s
.
T
h
ey
also
in
ten
d
ed
to
ex
te
n
d
th
eir
m
o
d
el
to
id
en
tify
o
t
h
er
wo
o
d
s
u
r
f
ac
e
d
ef
ec
ts
,
s
u
ch
as
wo
r
m
h
o
les
an
d
d
is
co
lo
r
atio
n
,
a
n
d
to
im
p
r
o
v
e
tar
g
et
lo
ca
tio
n
p
r
ec
is
io
n
in
f
u
t
u
r
e
r
esear
ch
.
T
h
en
,
E
h
tis
h
am
et
a
l.
[
4
1
]
d
is
cu
s
s
ed
tr
an
s
f
er
lear
n
in
g
a
s
a
cr
u
cial
elem
en
t
in
th
eir
s
tu
d
y
o
f
id
en
tify
in
g
an
d
class
if
y
in
g
d
ef
ec
ts
in
wo
o
d
en
s
tr
u
ctu
r
es.
T
h
e
r
esear
ch
ev
al
u
ated
ten
p
r
e
-
t
r
a
in
ed
C
NN
m
o
d
els
R
esNet1
8
,
R
es
Net5
0
,
R
esN
et1
0
1
,
Sh
u
f
f
leNe
t,
Go
o
g
L
e
Net,
I
n
ce
p
tio
n
-
V3
,
Mo
b
ileNet
-
V2
,
Xce
p
tio
n
,
I
n
ce
p
tio
n
-
R
esNet
-
V2
,
an
d
NASNet
-
Mo
b
ile
an
d
ea
ch
m
o
d
e
l
was
f
u
r
th
er
tr
ain
ed
an
d
v
alid
ated
,
d
em
o
n
s
tr
atin
g
th
e
ef
f
ec
tiv
en
ess
o
f
tr
an
s
f
er
lear
n
in
g
in
ad
ap
tin
g
th
ese
p
r
e
-
tr
ain
ed
m
o
d
els
f
o
r
s
p
ec
if
ic
task
s
.
T
h
e
f
in
d
in
g
s
in
d
icate
d
th
at
th
e
I
n
ce
p
tio
n
-
V
3
m
o
d
el
p
e
r
f
o
r
m
ed
t
h
e
b
est,
a
ch
iev
in
g
an
o
v
er
all
ac
c
u
r
ac
y
r
ate
o
f
9
7
.
3
%
an
d
a
tr
ain
in
g
tim
e
o
f
9
7
m
in
u
tes.
T
h
is
m
o
d
el’
s
a
r
ch
itectu
r
e,
f
ea
tu
r
in
g
4
8
d
ee
p
lay
er
s
,
allo
we
d
it
to
class
if
y
d
ef
ec
ts
in
wo
o
d
en
s
tr
u
ctu
r
es
ef
f
ec
tiv
ely
.
T
h
e
s
tu
d
y
h
ig
h
lig
h
ted
th
e
ef
f
icien
cy
o
f
u
s
in
g
p
r
e
-
tr
ai
n
ed
C
NNs,
as
th
ey
s
ig
n
if
ican
tly
im
p
r
o
v
ed
class
if
icatio
n
ac
cu
r
ac
y
a
n
d
r
ed
u
c
ed
th
e
tim
e
r
eq
u
ir
ed
f
o
r
d
ef
ec
t
id
en
tific
atio
n
co
m
p
ar
ed
to
tr
ad
itio
n
al
m
an
u
a
l in
s
p
ec
tio
n
m
eth
o
d
s
.
Fu
r
th
er
m
o
r
e
,
C
h
u
n
et
a
l.
[
3
3
]
d
is
cu
s
s
ed
tr
an
s
f
er
lear
n
in
g
as a
s
ig
n
if
ican
t c
o
m
p
o
n
en
t in
th
e
ir
s
tu
d
y
o
f
id
en
tify
in
g
tim
b
er
d
ef
ec
ts
ac
r
o
s
s
f
o
u
r
Ma
lay
s
ian
tim
b
er
s
p
ec
ies.
T
h
ey
h
ig
h
lig
h
ted
u
s
in
g
th
e
R
esNet5
0
alg
o
r
ith
m
,
b
ased
o
n
tr
an
s
f
er
lear
n
in
g
,
o
u
tp
e
r
f
o
r
m
in
g
o
th
er
C
NN
m
o
d
els,
ac
h
iev
in
g
a
class
if
icatio
n
ac
cu
r
ac
y
o
f
9
4
.
5
9
%
wh
en
co
m
b
in
ed
with
d
ata
au
g
m
e
n
tatio
n
tec
h
n
iq
u
es.
T
h
e
r
esear
ch
in
d
icate
d
t
h
at
d
ata
a
u
g
m
e
n
tatio
n
ad
d
r
ess
ed
is
s
u
es
r
elate
d
to
lim
ited
d
atasets
an
d
im
p
r
o
v
ed
th
e
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
o
f
C
NNs
b
y
5
.
7
8
%
.
T
h
e
s
tu
d
y
e
v
alu
ated
th
e
ef
f
ec
tiv
en
ess
o
f
tr
an
s
f
er
lear
n
i
n
g
in
c
o
n
ju
n
ctio
n
with
v
ar
i
o
u
s
h
y
p
er
p
ar
a
m
eter
s
ettin
g
s
,
s
u
ch
as
lear
n
in
g
r
ates
an
d
ep
o
ch
s
,
d
em
o
n
s
tr
atin
g
th
at
wh
ile
th
ese
ad
ju
s
tm
en
ts
wer
e
im
p
o
r
tan
t,
h
ig
h
er
v
alu
es
d
id
n
o
t
alwa
y
s
lead
to
b
etter
ac
c
u
r
ac
y
in
C
NN
class
if
icatio
n
.
Ov
er
all,
th
e
f
in
d
in
g
s
s
u
g
g
ested
th
at
d
ata
a
u
g
m
e
n
tatio
n
a
n
d
tr
an
s
f
er
lear
n
in
g
wer
e
e
f
f
ec
ti
v
e
s
tr
ateg
ies
f
o
r
e
n
h
an
cin
g
t
h
e
id
en
tific
atio
n
o
f
tim
b
er
d
ef
ec
ts
,
a
n
d
th
e
au
th
o
r
s
n
o
ted
p
o
ten
tial
f
o
r
f
u
t
u
r
e
r
e
s
ea
r
ch
in
ex
p
lo
r
in
g
m
o
r
e
co
m
p
lex
au
g
m
en
tatio
n
tech
n
iq
u
es a
n
d
ap
p
ly
i
n
g
d
ee
p
lear
n
in
g
to
a
n
aly
ze
d
i
f
f
er
en
t t
y
p
es o
f
tim
b
e
r
d
ef
ec
ts
ac
r
o
s
s
v
ar
io
u
s
s
p
ec
ies.
4.
CO
NCLU
SI
O
N
T
h
e
g
r
o
w
i
n
g
i
m
p
o
r
t
a
n
c
e
o
f
d
e
e
p
l
e
a
r
n
i
n
g
m
et
h
o
d
s
is
e
v
id
e
n
t
,
a
s
r
e
c
e
n
t
r
e
s
ea
r
c
h
h
i
g
h
li
g
h
t
s
t
h
e
i
r
e
f
f
e
c
t
i
v
e
n
es
s
i
n
e
n
h
a
n
c
i
n
g
t
h
e
a
c
c
u
r
a
c
y
a
n
d
e
f
f
i
c
i
e
n
c
y
o
f
w
o
o
d
d
e
f
e
c
t
i
d
e
n
t
i
f
i
c
at
i
o
n
,
s
i
g
n
i
f
i
ca
n
t
l
y
o
u
t
p
e
r
f
o
r
m
i
n
g
t
r
a
d
i
t
i
o
n
a
l
i
n
s
p
e
c
t
i
o
n
m
e
t
h
o
d
s
a
n
d
l
e
a
d
i
n
g
t
o
i
m
p
r
o
v
e
d
q
u
a
l
i
t
y
c
o
n
t
r
o
l
i
n
t
h
e
t
i
m
b
e
r
i
n
d
u
s
t
r
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
3
,
J
u
n
e
2
0
2
5
:
1
6
9
6
-
1
7
0
7
1704
A
d
v
a
n
c
e
m
e
n
t
s
i
n
C
N
N
a
r
c
h
i
t
e
c
t
u
r
es
,
s
u
c
h
a
s
I
n
c
e
p
t
i
o
n
-
R
es
N
e
t
-
V
2
a
n
d
v
a
r
i
o
u
s
R
e
s
N
e
t
m
o
d
e
l
s
,
h
a
v
e
d
e
m
o
n
s
t
r
a
t
e
d
r
e
m
a
r
k
a
b
l
e
i
m
p
r
o
v
e
m
e
n
t
s
i
n
c
l
a
s
s
i
f
ic
a
t
i
o
n
a
c
c
u
r
a
c
y
,
r
e
f
l
e
c
t
i
n
g
a
s
h
i
f
t
t
o
w
a
r
d
s
m
o
r
e
s
o
p
h
i
s
t
i
c
at
e
d
a
n
d
p
r
a
c
t
i
c
a
l
t
e
c
h
n
i
q
u
e
s
f
o
r
d
e
f
e
c
t
i
d
e
n
t
i
f
i
ca
t
i
o
n
.
F
u
r
t
h
e
r
m
o
r
e
,
i
m
p
l
e
m
e
n
t
i
n
g
f
r
a
m
e
w
o
r
k
s
l
i
k
e
F
a
s
t
e
r
R
-
C
NN
a
n
d
M
o
b
i
l
e
N
et
V
3
s
h
o
w
c
as
e
s
t
h
e
a
b
i
li
t
y
o
f
t
h
es
e
a
d
v
a
n
c
e
d
m
o
d
e
l
s
t
o
p
r
o
v
i
d
e
r
e
a
l
-
tim
e
d
e
t
e
c
ti
o
n
a
n
d
c
l
a
s
s
i
f
ic
a
t
i
o
n
,
w
h
i
c
h
i
s
e
s
s
e
n
ti
a
l
f
o
r
i
n
d
u
s
t
r
i
a
l
a
p
p
l
ic
a
t
i
o
n
s
w
h
e
r
e
q
u
i
c
k
d
e
c
is
i
o
n
-
m
a
k
i
n
g
i
s
c
r
u
c
ia
l
.
D
es
p
i
te
t
h
e
s
e
a
d
v
a
n
c
e
m
e
n
t
s
,
c
h
al
l
e
n
g
e
s
r
e
m
a
i
n
,
p
a
r
t
i
c
u
l
a
r
l
y
r
e
g
a
r
d
i
n
g
t
h
e
l
i
m
it
e
d
a
v
a
il
a
b
i
li
t
y
o
f
l
a
b
e
l
e
d
d
a
t
as
e
ts
.
H
o
w
e
v
e
r
,
i
n
n
o
v
a
t
i
v
e
a
p
p
r
o
a
c
h
e
s
l
i
k
e
t
r
a
n
s
f
e
r
le
a
r
n
i
n
g
a
n
d
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
h
a
v
e
p
r
o
v
e
n
e
f
f
e
c
t
i
v
e
i
n
e
n
h
a
n
c
i
n
g
m
o
d
e
l
p
e
r
f
o
r
m
a
n
c
e
,
e
v
e
n
w
h
e
n
w
o
r
k
i
n
g
wi
t
h
s
m
al
l
e
r
d
a
t
as
e
ts
.
T
h
e
e
c
o
n
o
m
i
c
a
n
d
p
r
a
c
t
i
c
a
l
b
e
n
e
f
i
ts
o
f
i
n
t
e
g
r
a
t
i
n
g
t
h
e
s
e
a
d
v
a
n
c
e
d
d
e
t
e
c
t
i
o
n
s
y
s
t
e
m
s
a
r
e
s
u
b
s
ta
n
t
ia
l
,
i
n
c
l
u
d
i
n
g
r
e
d
u
c
e
d
l
a
b
o
r
c
o
s
t
s
,
i
m
p
r
o
v
e
d
s
a
f
e
t
y
,
a
n
d
m
o
r
e
e
f
f
i
c
i
e
n
t
r
e
s
o
u
r
c
e
u
t
i
l
i
z
at
i
o
n
i
n
t
i
m
b
e
r
m
a
n
u
f
a
ct
u
r
i
n
g
,
u
l
t
i
m
a
t
e
l
y
l
ea
d
i
n
g
t
o
s
i
g
n
i
f
i
c
a
n
t
p
r
o
c
e
s
s
o
p
t
i
m
i
z
a
ti
o
n
s
i
n
t
h
e
w
o
o
d
i
n
d
u
s
t
r
y
.
W
h
il
e
c
u
r
r
e
n
t
s
t
u
d
i
es
p
r
e
d
o
m
i
n
a
n
t
l
y
f
o
c
u
s
o
n
s
p
e
c
i
f
i
c
d
e
f
e
c
t
t
y
p
e
s
,
s
u
c
h
as
k
n
o
t
s
,
t
h
e
r
e
is
a
n
u
r
g
e
n
t
n
e
e
d
f
o
r
m
o
r
e
c
o
m
p
r
e
h
e
n
s
i
v
e
m
o
d
e
l
s
c
a
p
a
b
le
o
f
m
u
l
t
i
c
la
s
s
i
d
e
n
ti
f
i
c
a
t
i
o
n
t
o
i
m
p
r
o
v
e
t
h
e
d
e
t
e
c
t
i
o
n
o
f
a
b
r
o
a
d
e
r
r
a
n
g
e
o
f
d
e
f
e
c
t
s
a
n
d
e
n
h
a
n
c
e
o
v
e
r
a
l
l
q
u
a
l
i
t
y
c
o
n
t
r
o
l
i
n
w
o
o
d
p
r
o
c
e
s
s
i
n
g
.
F
u
t
u
r
e
r
e
s
e
a
r
c
h
s
h
o
u
l
d
p
r
i
o
r
i
t
i
z
e
t
h
e
c
r
e
a
t
i
o
n
o
f
l
a
r
g
e
r
,
h
i
g
h
-
q
u
a
l
i
t
y
d
a
t
a
s
et
s
a
n
d
d
e
v
e
l
o
p
i
n
g
c
o
m
p
a
c
t
,
e
f
f
i
c
i
e
n
t
C
N
N
m
o
d
e
l
s
t
h
a
t
ca
n
f
u
n
c
t
i
o
n
o
n
l
o
w
-
p
o
w
e
r
c
o
m
p
u
t
i
n
g
r
e
s
o
u
r
ce
s
.
E
x
p
l
o
r
i
n
g
m
o
r
e
c
o
m
p
l
e
x
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
a
n
d
a
p
p
l
y
i
n
g
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
a
c
r
o
s
s
v
a
r
i
o
u
s
w
o
o
d
t
y
p
e
s
a
n
d
d
e
f
e
c
t
c
a
t
e
g
o
r
i
e
s
co
u
l
d
f
u
r
t
h
e
r
p
r
o
p
e
l
a
d
v
a
n
c
e
m
e
n
t
s
i
n
t
h
i
s
f
i
el
d
.
I
n
s
u
m
m
a
r
y
,
t
h
e
r
e
s
e
a
r
c
h
e
m
p
h
a
s
i
z
es
t
h
a
t
s
i
g
n
i
f
i
c
a
n
t
s
t
r
i
d
es
h
a
v
e
b
e
e
n
m
a
d
e
i
n
u
t
i
l
i
zi
n
g
d
e
e
p
l
ea
r
n
i
n
g
f
o
r
w
o
o
d
d
e
f
e
c
t
i
d
e
n
t
i
f
i
c
at
i
o
n
;
h
o
we
v
er
,
o
n
g
o
i
n
g
i
n
n
o
v
a
t
i
o
n
a
n
d
a
d
a
p
t
a
t
i
o
n
a
r
e
c
r
u
c
i
al
t
o
o
v
e
r
c
o
m
i
n
g
e
x
i
s
t
i
n
g
c
h
a
ll
e
n
g
es
a
n
d
m
a
x
i
m
i
zi
n
g
t
h
e
b
e
n
e
f
i
ts
o
f
t
h
e
s
e
t
e
c
h
n
o
l
o
g
i
es
i
n
r
e
al
-
wo
r
l
d
a
p
p
l
i
c
a
t
i
o
n
s
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
is
s
u
p
p
o
r
ted
b
y
th
e
Min
is
tr
y
o
f
Hig
h
er
E
d
u
ca
tio
n
(
MO
HE
)
,
Ma
lay
s
ia,
t
h
r
o
u
g
h
t
h
e
Fu
n
d
am
en
tal
R
esear
ch
Gr
a
n
t
Sch
em
e
(
FR
GS/1
/2
0
2
2
/I
C
T
0
2
/UT
E
M/0
2
/2
)
a
n
d
Fak
u
lti
T
ek
n
o
lo
g
i
Ma
k
lu
m
at
d
an
Ko
m
u
n
ik
asi,
Un
iv
er
s
iti T
ek
n
ik
al
Ma
lay
s
ia
Me
lak
a,
Me
l
ak
a,
Ma
lay
s
ia.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Ma
r
tin
a
Ali
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Um
m
i Rab
a’
ah
Hash
im
✓
✓
✓
✓
✓
✓
✓
Kastu
r
i K
an
ch
y
m
alay
✓
✓
✓
Aji Pr
asety
a
W
ib
awa
✓
✓
L
izaw
ati
Salah
u
d
d
in
✓
✓
R
ah
illd
a
Nad
h
ir
ah
No
r
izza
ty
R
ah
id
d
in
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
Data
a
v
aila
b
i
lit
y
is
n
o
t
a
p
p
li
ca
b
le
t
o
t
h
is
p
ap
er
as
n
o
n
e
w
d
a
t
a
w
er
e
cr
ea
te
d
o
r
an
al
y
z
e
d
i
n
t
h
is
s
t
u
d
y
.
RE
F
E
R
E
NC
E
S
[
1
]
F
.
D
i
n
g
,
Z.
Zh
u
a
n
g
,
Y
.
Li
u
,
D
.
Ji
a
n
g
,
X
.
Y
a
n
,
a
n
d
Z.
W
a
n
g
,
“
D
e
t
e
c
t
i
n
g
d
e
f
e
c
t
s
o
n
s
o
l
i
d
w
o
o
d
p
a
n
e
l
s
b
a
s
e
d
o
n
a
n
i
m
p
r
o
v
e
d
S
S
D
a
l
g
o
r
i
t
h
m,
”
S
e
n
so
rs
,
v
o
l
.
2
0
,
n
o
.
1
8
,
p
p
.
1
–
1
7
,
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
0
1
8
5
3
1
5
.
[
2
]
U
.
R
.
H
a
s
h
i
m
,
S
.
Z.
H
a
s
h
i
m
,
a
n
d
A
.
K
.
M
u
d
a
,
“
I
mag
e
c
o
l
l
e
c
t
i
o
n
f
o
r
n
o
n
-
s
e
g
m
e
n
t
i
n
g
a
p
p
r
o
a
c
h
o
f
t
i
m
b
e
r
su
r
f
a
c
e
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
s
i
n
S
o
f
t
C
o
m
p
u
t
i
n
g
a
n
d
i
t
s
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
7
,
n
o
.
1
,
p
p
.
1
6
–
3
4
,
2
0
1
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
A
r
ev
iew
o
f rece
n
t d
ee
p
lea
r
n
in
g
a
p
p
lica
tio
n
s
in
w
o
o
d
s
u
r
fa
ce
d
efec
t id
en
tifi
ca
tio
n
(
Ma
r
ti
n
a
A
li
)
1705
[
3
]
Y
.
C
h
e
n
,
C
.
S
u
n
,
Z.
R
e
n
,
a
n
d
B
.
N
a
,
“
R
e
v
i
e
w
o
f
t
h
e
c
u
r
r
e
n
t
s
t
a
t
e
o
f
a
p
p
l
i
c
a
t
i
o
n
o
f
w
o
o
d
d
e
f
e
c
t
r
e
c
o
g
n
i
t
i
o
n
t
e
c
h
n
o
l
o
g
y
,
”
Bi
o
Re
s
o
u
rce
s
,
v
o
l
.
1
8
,
n
o
.
1
,
p
p
.
2
2
8
8
–
2
3
0
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
5
3
7
6
/
b
i
o
r
e
s
.
1
8
.
1
.
C
h
e
n
.
[
4
]
M
.
B
u
d
a
k
ç
ı
a
n
d
H
.
Ç
ı
n
a
r
,
“
C
o
l
o
u
r
e
f
f
e
c
t
s
o
f
st
a
i
n
s
o
n
w
o
o
d
w
i
t
h
k
n
o
t
s,
c
r
a
c
k
s
a
n
d
r
o
t
s,”
Pro
g
r
e
ss
i
n
O
r
g
a
n
i
c
C
o
a
t
i
n
g
s
,
v
o
l
.
5
1
,
p
p
.
1
–
5
,
2
0
0
4
.
[
5
]
U
.
R
.
H
a
sh
i
m,
S
.
Z
.
H
a
s
h
i
m
,
a
n
d
A
.
K
.
M
u
d
a
,
“
A
u
t
o
ma
t
e
d
v
i
s
i
o
n
i
n
s
p
e
c
t
i
o
n
o
f
t
i
mb
e
r
s
u
r
f
a
c
e
d
e
f
e
c
t
:
A
r
e
v
i
e
w
,
”
J
u
r
n
a
l
T
e
k
n
o
l
o
g
i
,
v
o
l
.
7
7
,
n
o
.
2
0
,
p
p
.
1
2
7
–
1
3
5
,
2
0
1
5
,
d
o
i
:
1
0
.
1
1
1
1
3
/
j
t
.
v
7
7
.
6
5
6
2
.
[
6
]
Y
.
T
a
n
,
X
.
Y
a
n
g
,
X
.
B
a
i
,
H
.
D
o
n
g
,
J.
L
i
u
,
a
n
d
L
.
Zh
a
n
g
,
“
I
n
sp
e
c
t
i
o
n
a
n
d
e
v
a
l
u
a
t
i
o
n
o
f
w
o
o
d
c
o
m
p
o
n
e
n
t
s
o
f
a
n
c
i
e
n
t
b
u
i
l
d
i
n
g
s
i
n
t
h
e
s
o
u
t
h
-
t
h
r
e
e
c
o
u
r
t
s
o
f
t
h
e
f
o
r
b
i
d
d
e
n
c
i
t
y
,
”
B
i
o
Re
s
o
u
rce
s
,
v
o
l
.
1
7
,
n
o
.
1
,
p
p
.
9
6
2
–
9
7
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
5
3
7
6
/
b
i
o
r
e
s.1
7
.
1
.
9
6
2
-
9
7
4
.
[
7
]
P
.
X
u
e
t
a
l
.
,
“
A
p
p
l
i
c
a
t
i
o
n
o
f
n
o
n
d
e
st
r
u
c
t
i
v
e
t
e
s
t
i
n
g
t
e
c
h
n
o
l
o
g
i
e
s
i
n
p
r
e
s
e
r
v
i
n
g
h
i
st
o
r
i
c
t
r
e
e
s
a
n
d
a
n
c
i
e
n
t
t
i
m
b
e
r
st
r
u
c
t
u
r
e
s
i
n
C
h
i
n
a
,
”
F
o
res
t
s
,
v
o
l
.
1
2
,
n
o
.
3
,
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
f
1
2
0
3
0
3
1
8
.
[
8
]
C
.
L.
T
o
d
o
r
o
k
i
,
E
.
C
.
L
o
w
e
l
l
,
a
n
d
D
.
D
y
k
s
t
r
a
,
“
A
u
t
o
m
a
t
e
d
k
n
o
t
d
e
t
e
c
t
i
o
n
w
i
t
h
v
i
su
a
l
p
o
s
t
-
p
r
o
c
e
ssi
n
g
o
f
D
o
u
g
l
a
s
-
f
i
r
v
e
n
e
e
r
i
ma
g
e
s,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
r
o
n
i
c
s i
n
A
g
ri
c
u
l
t
u
r
e
,
v
o
l
.
7
0
,
n
o
.
1
,
p
p
.
1
6
3
–
1
7
1
,
2
0
1
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
a
g
.
2
0
0
9
.
1
0
.
0
0
2
.
[
9
]
W
.
Z
h
o
u
,
M
.
F
e
i
,
H
.
Zh
o
u
,
a
n
d
K
.
L
i
,
“
A
s
p
a
r
se
r
e
p
r
e
s
e
n
t
a
t
i
o
n
b
a
s
e
d
f
a
st
d
e
t
e
c
t
i
o
n
m
e
t
h
o
d
f
o
r
s
u
r
f
a
c
e
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
o
f
b
o
t
t
l
e
c
a
p
s,”
N
e
u
r
o
c
o
m
p
u
t
i
n
g
,
v
o
l
.
1
2
3
,
p
p
.
4
0
6
–
4
1
4
,
2
0
1
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.2
0
1
3
.
0
7
.
0
3
8
.
[
1
0
]
C
.
C
l
é
m
e
n
t
,
T
.
Li
h
r
a
,
R
.
G
a
z
o
,
a
n
d
R
.
B
e
a
u
r
e
g
a
r
d
,
“
M
a
x
i
m
i
z
i
n
g
l
u
m
b
e
r
u
s
e
:
T
h
e
e
f
f
e
c
t
o
f
ma
n
u
f
a
c
t
u
r
i
n
g
d
e
f
e
c
t
s
o
n
y
i
e
l
d
,
a
c
a
s
e
st
u
d
y
,
”
F
o
res
t
Pr
o
d
u
c
t
s
J
o
u
r
n
a
l
,
v
o
l
.
5
6
,
n
o
.
1
,
p
p
.
6
0
–
6
5
,
2
0
0
6
.
[
1
1
]
M
.
M
o
h
s
i
n
,
O
.
S
.
B
a
l
o
g
u
n
,
K
.
H
a
a
t
a
j
a
,
a
n
d
P
.
T
o
i
v
a
n
e
n
,
“
R
e
a
l
-
t
i
m
e
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
n
w
o
o
d
s
u
r
f
a
c
e
s
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
”
I
S
a
n
d
T
I
n
t
e
r
n
a
t
i
o
n
a
l
S
y
m
p
o
si
u
m
o
n
E
l
e
c
t
r
o
n
i
c
I
m
a
g
i
n
g
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
3
4
,
n
o
.
1
0
,
2
0
2
2
,
d
o
i
:
1
0
.
2
3
5
2
/
EI
.
2
0
2
2
.
3
4
.
1
0
.
I
P
A
S
-
382.
[
1
2
]
Y
.
F
a
n
g
,
L.
L
i
n
,
H
.
F
e
n
g
,
Z.
L
u
,
a
n
d
G
.
W
.
E
mm
s,
“
R
e
v
i
e
w
o
f
t
h
e
u
se
o
f
a
i
r
-
c
o
u
p
l
e
d
u
l
t
r
a
s
o
n
i
c
t
e
c
h
n
o
l
o
g
i
e
s
f
o
r
n
o
n
d
e
st
r
u
c
t
i
v
e
t
e
st
i
n
g
o
f
w
o
o
d
a
n
d
w
o
o
d
p
r
o
d
u
c
t
s,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
r
o
n
i
c
s
i
n
A
g
ri
c
u
l
t
u
r
e
,
v
o
l
.
1
3
7
,
p
p
.
7
9
–
8
7
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mp
a
g
.
2
0
1
7
.
0
3
.
0
1
5
.
[
1
3
]
B
.
Y
u
c
e
,
E
.
M
a
st
r
o
c
i
n
q
u
e
,
M
.
S
.
P
a
c
k
i
a
n
a
t
h
e
r
,
D
.
P
h
a
m
,
A
.
La
m
b
i
a
se
,
a
n
d
F
.
F
r
u
g
g
i
e
r
o
,
“
N
e
u
r
a
l
n
e
t
w
o
r
k
d
e
si
g
n
a
n
d
f
e
a
t
u
r
e
sel
e
c
t
i
o
n
u
s
i
n
g
p
r
i
n
c
i
p
a
l
c
o
mp
o
n
e
n
t
a
n
a
l
y
si
s
a
n
d
Ta
g
u
c
h
i
me
t
h
o
d
f
o
r
i
d
e
n
t
i
f
y
i
n
g
w
o
o
d
v
e
n
e
e
r
d
e
f
e
c
t
s
,
”
Pr
o
d
u
c
t
i
o
n
a
n
d
Ma
n
u
f
a
c
t
u
r
i
n
g
Re
se
a
r
c
h
,
v
o
l
.
2
,
n
o
.
1
,
p
p
.
2
9
1
–
3
0
8
,
2
0
1
4
,
d
o
i
:
1
0
.
1
0
8
0
/
2
1
6
9
3
2
7
7
.
2
0
1
4
.
8
9
2
4
4
2
.
[
1
4
]
A
.
U
r
b
o
n
a
s,
V
.
R
a
u
d
o
n
i
s,
R
.
M
a
s
k
e
l
i
u
n
a
s,
a
n
d
R
.
D
a
m
a
še
v
i
č
i
u
s,
“
A
u
t
o
m
a
t
e
d
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
w
o
o
d
v
e
n
e
e
r
s
u
r
f
a
c
e
d
e
f
e
c
t
s
u
si
n
g
f
a
st
e
r
r
e
g
i
o
n
-
b
a
se
d
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
w
i
t
h
d
a
t
a
a
u
g
me
n
t
a
t
i
o
n
a
n
d
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
9
,
n
o
.
2
2
,
2
0
1
9
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
9
2
2
4
8
9
8
.
[
1
5
]
M
.
K
r
y
l
,
L
.
D
a
n
y
s
,
R
.
Jar
o
s,
R
.
M
a
r
t
i
n
e
k
,
P
.
K
o
d
y
t
e
k
,
a
n
d
P
.
B
i
l
i
k
,
“
W
o
o
d
r
e
c
o
g
n
i
t
i
o
n
a
n
d
q
u
a
l
i
t
y
i
ma
g
i
n
g
i
n
sp
e
c
t
i
o
n
s
y
st
e
ms,
”
J
o
u
rn
a
l
o
f
S
e
n
s
o
rs
,
v
o
l
.
2
0
2
0
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
0
/
3
2
1
7
1
2
6
.
[
1
6
]
I
.
Y
.
H
.
G
u
,
H
.
A
n
d
e
r
ss
o
n
,
a
n
d
R
.
V
i
c
e
n
,
“
A
u
t
o
m
a
t
i
c
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
w
o
o
d
d
e
f
e
c
t
s
u
s
i
n
g
su
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
s,”
i
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
G
r
a
p
h
i
c
s
,
2
0
0
9
,
p
p
.
3
5
6
–
3
6
7
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
642
-
0
2
3
4
5
-
3
_
3
5
.
[
1
7
]
T.
H
.
C
h
u
n
e
t
a
l
.
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
w
o
o
d
d
e
f
e
c
t
u
s
i
n
g
p
a
t
t
e
r
n
r
e
c
o
g
n
i
t
i
o
n
t
e
c
h
n
i
q
u
e
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
A
d
v
a
n
c
e
s
i
n
I
n
t
e
l
l
i
g
e
n
t
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
7
,
n
o
.
2
,
p
p
.
1
6
3
–
1
7
6
,
2
0
2
1
,
d
o
i
:
1
0
.
2
6
5
5
5
/
i
j
a
i
n
.
v
7
i
2
.
5
8
8
.
[
1
8
]
N
.
D
.
A
b
d
u
l
l
a
h
,
U
.
R
.
H
a
sh
i
m,
S
.
A
h
ma
d
,
a
n
d
L
.
S
a
l
a
h
u
d
d
i
n
,
“
A
n
a
l
y
s
i
s
o
f
t
e
x
t
u
r
e
f
e
a
t
u
r
e
s
f
o
r
w
o
o
d
d
e
f
e
c
t
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
Bu
l
l
e
t
i
n
o
f
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
9
,
n
o
.
1
,
p
p
.
1
2
1
–
1
2
8
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
5
9
1
/
e
e
i
.
v
9
i
1
.
1
5
5
3
.
[
1
9
]
H
.
A
.
H
u
b
e
r
,
C
.
W
.
M
c
M
i
l
l
i
n
,
a
n
d
J.
P
.
M
c
K
i
n
n
e
y
,
“
L
u
m
b
e
r
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
a
b
i
l
i
t
i
e
s
o
f
f
u
r
n
i
t
u
r
e
r
o
u
g
h
m
i
l
l
e
m
p
l
o
y
e
e
s
,
”
F
o
res
t
Pro
d
u
c
t
s
J
o
u
r
n
a
l
,
v
o
l
.
3
5
,
n
o
.
1
1
–
1
2
,
p
p
.
7
9
–
8
2
,
1
9
8
5
.
[
2
0
]
D
.
Y
a
d
a
v
,
A
k
a
n
k
s
h
a
,
a
n
d
A
.
K
.
Y
a
d
a
v
,
“
A
n
o
v
e
l
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
b
a
se
d
mo
d
e
l
f
o
r
r
e
c
o
g
n
i
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
a
p
p
l
e
l
e
a
f
d
i
sea
s
e
s,
”
T
r
a
i
t
e
m
e
n
t
d
u
S
i
g
n
a
l
,
v
o
l
.
3
7
,
n
o
.
6
,
p
p
.
1
0
9
3
–
1
1
0
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
8
2
8
0
/
TS.
3
7
0
6
2
2
.
[
2
1
]
X
.
Zh
u
,
M
.
Z
h
u
,
a
n
d
H
.
R
e
n
,
“
M
e
t
h
o
d
o
f
p
l
a
n
t
l
e
a
f
r
e
c
o
g
n
i
t
i
o
n
b
a
s
e
d
o
n
i
mp
r
o
v
e
d
d
e
e
p
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
,
”
C
o
g
n
i
t
i
v
e
S
y
s
t
e
m
s
Re
se
a
r
c
h
,
v
o
l
.
5
2
,
p
p
.
2
2
3
–
2
3
3
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
g
s
y
s.
2
0
1
8
.
0
6
.
0
0
8
.
[
2
2
]
J.
F
a
n
,
Y
.
L
i
u
,
Z.
H
u
,
Q
.
Z
h
a
o
,
L.
S
h
e
n
,
a
n
d
X
.
Zh
o
u
,
“
S
o
l
i
d
w
o
o
d
p
a
n
e
l
d
e
f
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
r
e
c
o
g
n
i
t
i
o
n
s
y
st
e
m
b
a
s
e
d
o
n
F
a
st
e
r
R
-
C
N
N
,
”
J
o
u
r
n
a
l
o
f
F
o
res
t
ry
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
4
,
n
o
.
3
,
p
p
.
1
1
2
–
1
1
7
,
2
0
1
9
,
d
o
i
:
1
0
.
1
3
3
6
0
/
j
.
i
ssn
.
2
0
9
6
-
1
3
5
9
.
2
0
1
9
.
0
3
.
0
1
7
.
[
2
3
]
T.
G
e
r
g
e
l
,
M
.
S
e
d
l
i
a
k
,
T.
B
u
c
h
a
,
M
.
O
r
a
v
e
c
,
M
.
S
l
a
m
k
a
,
a
n
d
M
.
P
á
st
o
r
,
“
P
r
e
d
i
c
t
i
o
n
m
o
d
e
l
o
f
w
o
o
d
e
n
l
o
g
s c
u
t
t
i
n
g
p
a
t
t
e
r
n
s
a
n
d
i
t
s
e
f
f
i
c
i
e
n
c
y
i
n
p
r
a
c
t
i
c
e
,
”
Ap
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
0
,
n
o
.
9
,
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
0
0
9
3
0
0
3
.
[
2
4
]
A
.
R
.
D.
G
e
u
s,
A
.
R
.
B
a
c
k
e
s
,
A
.
B
.
G
o
n
t
i
j
o
,
G
.
H
.
Q
.
A
l
b
u
q
u
e
r
q
u
e
,
a
n
d
J.
R
.
S
o
u
z
a
,
“
A
maz
o
n
w
o
o
d
s
p
e
c
i
e
s
c
l
a
ss
i
f
i
c
a
t
i
o
n
:
a
c
o
m
p
a
r
i
so
n
b
e
t
w
e
e
n
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
p
r
e
-
d
e
si
g
n
e
d
f
e
a
t
u
r
e
s,”
W
o
o
d
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
5
5
,
n
o
.
3
,
p
p
.
8
5
7
–
8
7
2
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
2
2
6
-
021
-
0
1
2
8
2
-
w.
[
2
5
]
K
.
Z
h
a
n
g
,
B
.
Y
u
a
n
,
a
n
d
Y
.
L
i
,
“
Ef
f
i
c
i
e
n
c
y
a
n
a
l
y
s
i
s
o
f
w
o
o
d
p
r
o
c
e
s
si
n
g
i
n
d
u
s
t
r
y
i
n
C
h
i
n
a
d
u
r
i
n
g
2
0
0
6
-
2
0
1
5
,
”
I
O
P
C
o
n
f
e
re
n
c
e
S
e
r
i
e
s:
Ma
t
e
r
i
a
l
s
S
c
i
e
n
c
e
a
n
d
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
3
2
2
,
n
o
.
5
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
5
7
-
8
9
9
X
/
3
2
2
/
5
/
0
5
2
0
6
2
.
[
2
6
]
R
.
Q
a
y
y
u
m
,
K
.
K
a
mal
,
T
.
Za
f
a
r
,
a
n
d
S
.
M
a
t
h
a
v
a
n
,
“
W
o
o
d
d
e
f
e
c
t
s
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
G
L
C
M
b
a
se
d
f
e
a
t
u
r
e
s
a
n
d
P
S
O
t
r
a
i
n
e
d
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
i
n
2
0
1
6
2
2
n
d
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
A
u
t
o
m
a
t
i
o
n
a
n
d
C
o
m
p
u
t
i
n
g
(
I
C
A
C
)
,
2
0
1
6
,
p
p
.
2
7
3
–
2
7
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
o
n
A
C
.
2
0
1
6
.
7
6
0
4
9
3
1
.
[
2
7
]
R
.
N
.
N
.
R
a
h
i
d
d
i
n
,
U
.
R
.
H
a
s
h
i
m
,
N
.
H
.
I
smai
l
,
L.
S
a
l
a
h
u
d
d
i
n
,
N
.
H
.
C
h
o
o
n
,
a
n
d
S
.
N
.
Za
b
r
i
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
w
o
o
d
d
e
f
e
c
t
i
ma
g
e
s
u
si
n
g
l
o
c
a
l
b
i
n
a
r
y
p
a
t
t
e
r
n
v
a
r
i
a
n
t
s,”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
s
i
n
I
n
t
e
l
l
i
g
e
n
t
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
6
,
n
o
.
1
,
p
p
.
3
6
–
4
5
,
2
0
2
0
,
d
o
i
:
1
0
.
2
6
5
5
5
/
i
j
a
i
n
.
v
6
i
1
.
3
9
2
.
[
2
8
]
Y
.
H
u
a
n
g
,
J.
J
i
n
g
,
a
n
d
Z
.
W
a
n
g
,
“
F
a
b
r
i
c
d
e
f
e
c
t
s
e
g
m
e
n
t
a
t
i
o
n
m
e
t
h
o
d
b
a
s
e
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
st
r
u
m
e
n
t
a
t
i
o
n
a
n
d
M
e
a
s
u
rem
e
n
t
,
v
o
l
.
7
0
,
p
p
.
1
–
1
5
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
M
.
2
0
2
0
.
3
0
4
7
1
9
0
.
[
2
9
]
J.
S
h
i
,
Z.
L
i
,
T
.
Z
h
u
,
D
.
W
a
n
g
,
a
n
d
C
.
N
i
,
“
D
e
f
e
c
t
d
e
t
e
c
t
i
o
n
o
f
i
n
d
u
st
r
y
w
o
o
d
v
e
n
e
e
r
b
a
s
e
d
o
n
N
A
S
a
n
d
mu
l
t
i
-
c
h
a
n
n
e
l
mas
k
R
-
C
N
N
,
”
S
e
n
s
o
rs
,
v
o
l
.
2
0
,
n
o
.
1
6
,
p
p
.
1
–
1
7
,
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
0
1
6
4
3
9
8
.
[
3
0
]
T.
H
e
,
Y
.
L
i
u
,
Y
.
Y
u
,
Q
.
Z
h
a
o
,
a
n
d
Z
.
H
u
,
“
A
p
p
l
i
c
a
t
i
o
n
o
f
d
e
e
p
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
o
n
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
a
n
d
d
e
t
e
c
t
i
on
o
f
w
o
o
d
d
e
f
e
c
t
s,
”
Me
a
s
u
rem
e
n
t
:
J
o
u
rn
a
l
o
f
t
h
e
I
n
t
e
rn
a
t
i
o
n
a
l
Me
a
su
r
e
m
e
n
t
C
o
n
f
e
d
e
ra
t
i
o
n
,
v
o
l
.
1
5
2
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
me
a
su
r
e
me
n
t
.
2
0
1
9
.
1
0
7
3
5
7
.
[
3
1
]
U
.
B
u
e
h
l
m
a
n
n
a
n
d
R
.
E
.
T
h
o
m
a
s,
“
R
e
l
a
t
i
o
n
s
h
i
p
b
e
t
w
e
e
n
l
u
m
b
e
r
y
i
e
l
d
a
n
d
b
o
a
r
d
m
a
r
k
e
r
a
c
c
u
r
a
c
y
i
n
r
i
p
-
f
i
r
st
r
o
u
g
h
m
i
l
l
s,”
Eu
r
o
p
e
a
n
J
o
u
rn
a
l
o
f
Wo
o
d
a
n
d
Wo
o
d
Pro
d
u
c
t
s:
H
o
l
z
a
l
s
R
o
h
-
u
n
d
We
r
k
st
o
f
f
,
v
o
l
.
6
5
,
n
o
.
1
,
p
p
.
4
3
–
4
8
,
2
0
0
7
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
1
0
7
-
0
0
6
-
0
0
9
9
-
7.
[
3
2
]
D
.
E.
K
l
i
n
e
,
C
.
S
u
r
a
k
,
a
n
d
P
.
A
.
A
r
a
ma
n
,
“
A
u
t
o
m
a
t
e
d
h
a
r
d
w
o
o
d
l
u
m
b
e
r
g
r
a
d
i
n
g
u
t
i
l
i
z
i
n
g
a
m
u
l
t
i
p
l
e
s
e
n
s
o
r
m
a
c
h
i
n
e
v
i
s
i
o
n
t
e
c
h
n
o
l
o
g
y
,
”
C
o
m
p
u
t
e
rs
a
n
d
E
l
e
c
t
r
o
n
i
c
s
i
n
A
g
r
i
c
u
l
t
u
r
e
,
v
o
l
.
4
1
,
n
o
.
1
–
3
,
p
p
.
1
3
9
–
1
5
5
,
2
0
0
3
,
d
o
i
:
1
0
.
1
0
1
6
/
S
0
1
6
8
-
1
6
9
9
(
0
3
)
0
0
0
4
8
-
6.
[
3
3
]
T.
H
.
C
h
u
n
,
U
.
R
.
H
a
s
h
i
m
,
S
.
A
h
m
a
d
,
L
.
S
a
l
a
h
u
d
d
i
n
,
N
.
H
.
C
h
o
o
n
,
a
n
d
K
.
K
a
n
c
h
y
ma
l
a
y
,
“
Ef
f
i
c
a
c
y
o
f
t
h
e
i
m
a
g
e
a
u
g
m
e
n
t
a
t
i
o
n
met
h
o
d
u
si
n
g
C
N
N
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
i
n
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
t
i
m
b
e
r
d
e
f
e
c
t
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
A
d
v
a
n
c
e
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
3
,
n
o
.
5
,
p
p
.
1
0
7
–
1
1
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
4
5
6
9
/
I
JA
C
S
A
.
2
0
2
2
.
0
1
3
0
5
1
4
.
[
3
4
]
M
.
H
.
H
a
r
u
n
e
t
a
l
.
,
“
T
h
e
i
n
v
e
st
i
g
a
t
i
o
n
o
n
d
e
f
e
c
t
r
e
c
o
g
n
i
t
i
o
n
s
y
st
e
m
u
s
i
n
g
G
a
u
ss
i
a
n
sm
o
o
t
h
i
n
g
a
n
d
t
e
mp
l
a
t
e
ma
t
c
h
i
n
g
a
p
p
r
o
a
c
h
,
”
I
n
d
o
n
e
si
a
n
J
o
u
r
n
a
l
o
f
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
8
,
n
o
.
2
,
p
p
.
8
1
2
–
8
2
0
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
e
c
s.
v
1
8
.
i
2
.
p
p
8
1
2
-
8
2
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.