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.
1
4
,
No
.
5
,
Octo
b
er
2
0
2
5
,
p
p
.
4
1
2
3
~
4
1
3
6
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
5
.
p
p
4
1
2
3
-
4
1
3
6
4123
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Enha
nced clas
sifi
ca
tion o
f
a
ro
ma
ti
c herbs
using
Eff
i
cientNe
t
a
nd t
ra
nsfer
lear
ning
Sa
m
ira
Na
s
cim
ent
o
Ant
un
es,
M
a
da
lena
De
O
liv
eira
B
a
rbo
s
a
Div
ino
,
L
ua
na
Do
s
Sa
n
t
o
s
Co
rdeiro
,
F
er
na
nd
a
P
er
eira
L
eit
e
Ag
uia
r,
M
a
rc
elo
T
s
ug
uio
O
k
a
no
P
r
o
g
r
a
m
o
f
P
r
o
d
u
c
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
P
a
u
l
i
st
a
U
n
i
v
e
r
s
i
t
y
,
S
ã
o
P
a
u
l
o
,
B
r
a
z
i
l
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
2
1
,
2
0
2
4
R
ev
is
ed
J
u
n
2
0
,
2
0
2
5
Acc
ep
ted
J
u
l 1
0
,
2
0
2
5
He
rb
s h
a
v
e
lo
n
g
b
e
e
n
u
se
d
fo
r
c
u
li
n
a
ry
a
n
d
m
e
d
icin
a
l
p
u
rp
o
se
s,
a
s we
ll
a
s in
re
li
g
io
u
s
ri
tu
a
ls,
d
u
e
t
o
t
h
e
ir
e
ss
e
n
ti
a
l
o
il
s
a
n
d
a
ro
m
a
ti
c
p
ro
p
e
rti
e
s.
Ho
we
v
e
r,
d
isti
n
g
u
is
h
in
g
b
e
twe
e
n
a
ro
m
a
ti
c
a
n
d
m
e
d
icin
a
l
h
e
rb
s
b
a
se
d
o
n
v
is
u
a
l
c
h
a
ra
c
teristics
a
lo
n
e
c
a
n
b
e
c
h
a
ll
e
n
g
in
g
.
Wi
th
re
c
e
n
t
a
d
v
a
n
c
e
s
in
c
o
m
p
u
ter
v
isio
n
,
p
lan
t
i
d
e
n
ti
f
ica
ti
o
n
fr
o
m
i
m
a
g
e
s
h
a
s
se
e
n
si
g
n
ifi
c
a
n
t
g
ro
wt
h
,
o
ffe
rin
g
p
ro
m
isin
g
a
p
p
li
c
a
ti
o
n
s
in
se
v
e
ra
l
d
o
m
a
in
s.
Th
is
a
rti
c
le
a
ims
t
o
e
v
a
lu
a
te
th
e
c
las
sifica
ti
o
n
o
f
a
ro
m
a
ti
c
h
e
rb
s
u
sin
g
th
e
Eff
icie
n
tNe
t
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
(CNN
)
tec
h
n
iq
u
e
with
t
ra
n
sfe
r
lea
rn
in
g
.
T
h
e
m
e
th
o
d
o
l
o
g
y
u
se
d
is
e
x
p
e
rime
n
tal
re
se
a
rc
h
,
sy
ste
m
a
ti
c
a
ll
y
m
a
n
ip
u
latin
g
v
a
riab
les
to
o
b
se
rv
e
th
e
ir
e
ffe
c
ts
o
n
th
e
o
b
jec
t
o
f
st
u
d
y
.
Th
e
re
se
a
rc
h
e
r
p
la
y
s
a
n
a
c
ti
v
e
r
o
le
i
n
th
is
p
ro
c
e
ss
,
ra
th
e
r
th
a
n
b
e
in
g
a
p
a
ss
iv
e
o
b
se
rv
e
r.
Ba
se
d
o
n
th
e
re
su
lt
s
a
n
d
th
e
li
tera
tu
re
re
v
iew
,
it
is
e
v
id
e
n
t
th
a
t
th
e
o
b
jec
ti
v
e
o
f
t
h
is
re
se
a
rc
h
wa
s
a
c
h
iev
e
d
,
a
s
d
e
sp
it
e
th
e
o
p
p
o
r
tu
n
it
ies
fo
r
imp
ro
v
e
m
e
n
t
i
n
train
i
n
g
to
a
c
h
iev
e
a
c
c
u
ra
c
y
a
b
o
v
e
0
.
8
,
it
wa
s
p
o
ss
ib
le
t
o
e
v
a
lu
a
te
th
e
c
las
sifica
ti
o
n
o
f
a
ro
m
a
ti
c
h
e
rb
s u
si
n
g
Eff
icie
n
tNe
t
CNN
t
h
r
o
u
g
h
t
h
e
tran
sfe
r
lea
rn
i
n
g
tec
h
n
i
q
u
e
.
K
ey
w
o
r
d
s
:
Ar
o
m
atic
h
er
b
s
C
las
s
if
icatio
n
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Dee
p
lear
n
in
g
E
f
f
icien
t
N
et
I
m
ag
e
id
en
tific
atio
n
Plan
t id
en
tific
atio
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Ma
r
ce
lo
T
s
u
g
u
io
Ok
an
o
Pro
g
r
am
o
f
Pro
d
u
ctio
n
E
n
g
in
e
er
in
g
,
Pau
lis
ta
Un
iv
er
s
ity
(
UN
I
P)
R
ua
Dr
.
B
ac
elar
,
N
o.
1
2
1
2
,
Vi
la
C
lem
en
tin
o
,
São
Pau
l
o
, B
r
a
zil
E
m
ail: m
ar
ce
lo
.
o
k
a
n
o
@
u
n
ip
.
b
r
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
id
en
tific
atio
n
a
n
d
class
if
icatio
n
o
f
v
is
u
ally
s
im
ilar
h
er
b
s
p
r
esen
t
a
s
ig
n
if
ican
t
ch
alle
n
g
e
d
u
e
to
th
eir
r
esem
b
lan
ce
in
c
h
ar
ac
t
er
is
tics
s
u
ch
as
co
lo
r
,
s
h
a
p
e,
an
d
tex
tu
r
e.
T
r
ad
itio
n
al
m
e
th
o
d
s
o
f
te
n
f
ail
t
o
d
is
tin
g
u
is
h
th
ese
s
p
ec
ies
with
h
ig
h
ac
c
u
r
ac
y
,
h
ig
h
lig
h
tin
g
t
h
e
n
ee
d
f
o
r
m
o
r
e
ad
v
an
ce
d
tech
n
o
lo
g
ies.
Stu
d
ies
h
av
e
s
h
o
wn
th
at
m
o
d
els
b
ase
d
o
n
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
wo
r
k
s
(
C
NNs)
ca
n
en
h
a
n
ce
th
is
p
r
o
ce
s
s
,
en
ab
lin
g
au
to
m
ated
an
d
m
o
r
e
p
r
ec
is
e
id
en
tific
atio
n
.
Her
b
s
p
lay
a
cr
u
cial
r
o
le
in
b
o
th
n
u
tr
itio
n
an
d
p
h
y
t
o
th
er
ap
y
,
p
o
s
s
ess
in
g
h
is
to
r
ical
an
d
cu
ltu
r
al
s
ig
n
if
ican
ce
.
T
h
ey
h
a
v
e
b
ee
n
wid
ely
u
s
ed
in
r
it
u
als
an
d
r
elig
io
u
s
ce
r
em
o
n
ies d
u
e
to
th
eir
a
r
o
m
a
tic
an
d
m
ed
icin
al
p
r
o
p
er
ties
[
1
]
.
T
h
e
wid
e
v
ar
iety
o
f
a
r
o
m
at
ic
an
d
m
ed
icin
al
h
er
b
s
wit
h
s
im
ilar
s
h
ap
e,
co
lo
r
,
an
d
g
eo
m
etr
y
ch
ar
ac
ter
is
tics
m
ak
es
th
eir
v
is
u
al
d
if
f
e
r
en
tiatio
n
d
if
f
icu
lt.
H
o
wev
er
,
r
ec
en
t
a
d
v
an
ce
m
en
ts
in
co
m
p
u
ter
v
is
io
n
h
av
e
s
p
u
r
r
ed
r
ap
id
g
r
o
wth
i
n
p
lan
t
i
d
en
tific
atio
n
r
esear
c
h
,
with
p
r
o
m
is
in
g
r
esu
lts
th
at
d
em
o
n
s
tr
ate
h
ig
h
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
a
n
d
p
r
a
ctica
l
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
[
2
]
.
W
ith
d
ig
ital
tech
n
o
lo
g
ies
lik
e
ar
tific
ial
in
tellig
en
ce
(
AI
)
,
tr
an
s
f
er
le
ar
n
in
g
,
d
ee
p
lear
n
in
g
(
DL
)
,
C
NN
s
,
an
d
m
o
b
ile
d
ev
ice
s
,
au
to
m
atin
g
th
e
class
if
icatio
n
p
r
o
ce
s
s
o
f
ar
o
m
atic
an
d
m
ed
icin
al
h
er
b
s
h
as
b
ec
o
m
e
f
ea
s
ib
le.
Her
b
class
if
icatio
n
h
o
ld
s
p
ar
ticu
lar
im
p
o
r
tan
ce
in
m
ed
i
cin
e,
p
lan
t
s
cien
ce
,
an
d
th
e
f
o
o
d
in
d
u
s
tr
y
.
W
h
ile
v
ar
io
u
s
p
ar
ts
o
f
p
lan
ts
,
s
u
ch
as
leav
es,
f
lo
wer
s
,
f
r
u
its
,
s
ee
d
s
,
an
d
r
o
o
ts
,
ca
n
b
e
u
s
ed
f
o
r
s
p
ec
ies
id
e
n
tific
atio
n
,
lea
v
es
ar
e
esp
ec
ially
ad
v
an
tag
e
o
u
s
d
u
e
t
o
th
eir
a
v
ailab
ilit
y
th
r
o
u
g
h
o
u
t
m
u
ch
o
f
th
e
p
lan
t’
s
life
cy
cle
[
3
]
.
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
.
5
,
Octo
b
er
2
0
2
5
:
4
1
2
3
-
4
1
3
6
4124
T
h
is
s
tu
d
y
aim
s
to
ev
alu
ate
s
p
ec
if
ic
class
if
icatio
n
tech
n
iq
u
es
u
s
in
g
th
e
E
f
f
icien
tNet
C
NN
an
d
tr
an
s
f
er
lear
n
in
g
to
o
v
e
r
co
m
e
th
ese
ch
allen
g
es.
T
h
e
r
esear
ch
ad
o
p
ts
an
ex
p
er
im
en
tal
m
eth
o
d
o
lo
g
y
th
at
allo
ws
f
o
r
th
e
co
n
t
r
o
lled
m
a
n
ip
u
latio
n
o
f
s
tu
d
y
v
ar
iab
les
to
an
aly
z
e
th
eir
im
p
ac
t
o
n
th
e
o
b
ject
o
f
in
v
esti
g
atio
n
.
T
h
is
ap
p
r
o
ac
h
en
a
b
les
a
co
m
p
r
eh
e
n
s
iv
e
ass
ess
m
en
t
o
f
th
e
s
y
s
tem
’
s
b
eh
av
io
r
u
n
d
er
d
if
f
er
e
n
t
co
n
d
itio
n
s
,
e
n
s
u
r
in
g
th
at
th
e
ef
f
ec
ts
o
f
ea
ch
v
ar
ia
b
le
ar
e
s
y
s
tem
atica
lly
ev
alu
ated
.
T
h
e
v
alid
atio
n
o
f
th
e
m
eth
o
d
o
lo
g
y
is
ca
r
r
ied
o
u
t
th
r
o
u
g
h
ex
p
e
r
im
en
ts
th
at
s
im
u
late
v
ar
io
u
s
o
p
er
atio
n
al
co
n
d
it
io
n
s
,
allo
win
g
th
e
r
esear
ch
er
t
o
ac
tiv
ely
o
b
s
er
v
e
th
e
in
f
lu
en
ce
o
f
ea
c
h
v
ar
ia
b
le
[
4
]
.
T
h
e
s
tr
u
ctu
r
e
o
f
th
is
p
ap
er
i
s
as
f
o
llo
ws:
s
ec
t
io
n
2
p
r
es
en
ts
th
e
th
eo
r
etica
l
f
r
am
ewo
r
k
f
o
r
th
e
p
r
o
p
o
s
ed
to
p
ics
in
cl
u
d
in
g
th
e
m
eth
o
d
o
lo
g
y
.
S
ec
tio
n
4
d
eta
ils
th
e
r
esu
lts
an
d
d
is
cu
s
s
io
n
.
Fin
ally
,
s
ec
tio
n
5
co
n
clu
d
es with
f
in
al
o
b
s
er
v
ati
o
n
s
an
d
d
ir
ec
tio
n
s
f
o
r
f
u
tu
r
e
r
esear
ch
.
2.
T
H
E
O
R
E
T
I
CA
L
F
RAM
E
WO
RK
2
.
1
.
Cla
s
s
if
ica
t
io
n v
s
.
i
m
a
g
e
identif
ica
t
io
n wit
h E
f
f
icient
Net
co
nv
o
lutio
na
l neura
l net
wo
rk
T
h
e
d
is
tin
ctio
n
b
etwe
en
im
ag
e
class
if
icatio
n
an
d
i
m
ag
e
id
e
n
tific
atio
n
is
f
u
n
d
am
en
tal
t
o
u
n
d
er
s
tan
d
i
n
g
co
m
p
u
ter
v
is
io
n
task
s
.
I
m
ag
e
class
if
icat
io
n
r
ef
er
s
to
ass
ig
n
in
g
an
im
ag
e
a
s
in
g
le
class
o
r
lab
el.
I
n
c
o
n
tr
ast,
im
a
g
e
id
e
n
tific
atio
n
in
v
o
lv
es
id
en
tify
in
g
o
n
e
o
r
m
o
r
e
o
b
jects
th
r
o
u
g
h
b
o
u
n
d
in
g
b
o
x
es
ass
o
ciate
d
with
o
b
ject
class
es.
Acc
o
r
d
in
g
to
R
u
s
s
ak
o
v
s
k
y
et
a
l
.
[
5
]
,
i
m
ag
e
class
if
icatio
n
is
a
p
r
o
b
l
em
wh
er
e
th
e
g
o
al
is
to
ass
ig
n
a
s
in
g
le
clas
s
to
an
en
tire
im
ag
e.
T
h
is
p
r
o
ce
s
s
in
v
o
lv
es
d
eter
m
in
in
g
th
e
g
e
n
er
a
l
ca
teg
o
r
y
to
wh
ich
th
e
im
ag
e
b
elo
n
g
s
with
o
u
t
c
o
n
s
id
er
in
g
th
e
p
r
ec
is
e
lo
ca
tio
n
o
f
o
b
jects
with
in
th
e
im
ag
e.
On
th
e
o
th
er
h
an
d
,
im
ag
e
id
en
tific
atio
n
,
as
d
ef
in
ed
b
y
B
o
ch
k
o
v
s
k
iy
et
a
l
.
[
6
]
,
in
v
o
lv
es
th
e
lo
ca
lizatio
n
an
d
id
en
tific
atio
n
o
f
s
p
ec
if
ic
o
b
jects
o
r
in
d
iv
id
u
al
s
with
in
a
co
m
p
lex
im
ag
e.
I
n
th
is
co
n
tex
t,
th
e
task
is
to
r
ec
o
g
n
ize
o
b
jects
'
p
r
esen
ce
an
d
d
eter
m
i
n
e
th
eir
lo
ca
tio
n
s
an
d
o
f
ten
th
eir
s
p
ec
if
ic
id
en
titi
es.
T
h
e
s
wif
t
ad
v
a
n
ce
m
en
t
o
f
d
ig
ital
tech
n
o
lo
g
y
,
co
m
b
i
n
ed
with
th
e
g
r
o
win
g
in
teg
r
atio
n
o
f
AI
in
ag
r
icu
ltu
r
e,
h
as
o
p
en
e
d
n
ew
p
ath
s
f
o
r
in
n
o
v
ativ
e
s
o
lu
tio
n
s
in
p
lan
t
id
en
tific
atio
n
[
7
]
.
T
h
is
p
r
o
ce
s
s
ty
p
ically
in
v
o
lv
es
u
s
in
g
o
b
ject
d
etec
tio
n
alg
o
r
ith
m
s
,
s
u
ch
as
y
o
u
o
n
ly
lo
o
k
o
n
ce
(
YOL
O
)
o
r
f
aster
r
eg
io
n
-
b
ased
co
n
v
o
lu
t
io
n
al
n
eu
r
al
n
etwo
r
k
(
Fas
ter
R
-
C
NN
)
,
to
id
en
tify
r
eg
io
n
s
o
f
in
te
r
est
in
th
e
im
ag
e.
W
h
ile
im
ag
e
class
if
icatio
n
f
o
cu
s
es
o
n
ass
ig
n
in
g
lab
els
to
en
tire
im
ag
es,
im
ag
e
id
en
tific
atio
n
in
v
o
lv
es
d
etec
tin
g
an
d
id
en
tify
in
g
s
p
ec
if
ic
o
b
jects
with
in
th
e
im
ag
e.
T
h
ese
d
is
tin
ctio
n
s
ar
e
ess
en
tial f
o
r
u
n
d
er
s
tan
d
in
g
th
e
d
if
f
er
e
n
t a
p
p
r
o
ac
h
es a
n
d
t
ec
h
n
iq
u
es u
s
ed
i
n
m
ac
h
in
e
lear
n
in
g
.
T
h
i
s
r
e
s
e
a
r
c
h
f
o
c
u
s
e
d
o
n
c
l
a
s
s
i
f
i
c
a
t
i
o
n
u
s
i
n
g
t
h
e
E
f
f
i
c
i
e
n
t
N
e
t
C
N
N
.
A
c
c
o
r
d
i
n
g
t
o
T
a
n
a
n
d
L
e
[
8
]
,
t
h
e
E
f
f
i
c
i
e
n
t
N
e
t
C
N
N
p
r
o
p
o
s
e
s
a
n
i
n
n
o
v
a
t
i
v
e
a
p
p
r
o
a
c
h
t
o
d
e
s
i
g
n
i
n
g
s
c
a
l
a
b
l
e
a
n
d
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
e
f
f
i
c
i
e
n
t
C
N
N
s
.
T
h
e
y
i
n
t
r
o
d
u
c
e
a
s
c
a
l
i
n
g
m
e
t
h
o
d
c
o
n
s
i
s
t
i
n
g
o
f
d
e
p
t
h
,
w
i
d
t
h
,
a
n
d
r
e
s
o
l
u
t
i
o
n
s
c
a
l
e
s
,
a
l
l
o
w
i
n
g
t
h
e
m
o
d
e
l
t
o
b
e
r
e
s
o
u
r
c
e
-
e
f
f
i
c
i
e
n
t
w
h
i
l
e
m
a
i
n
t
a
i
n
i
n
g
o
r
i
m
p
r
o
v
i
n
g
c
l
a
s
s
i
f
i
c
a
t
i
o
n
p
e
r
f
o
r
m
a
n
c
e
.
F
u
r
t
h
e
r
m
o
r
e
,
C
h
e
n
n
a
[
9
]
e
m
p
h
a
s
i
z
e
t
h
a
t
t
h
e
E
f
f
i
c
i
e
n
t
N
e
t
C
N
N
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
i
n
t
h
e
d
e
s
i
g
n
o
f
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
s
.
T
h
e
y
n
o
t
e
t
h
a
t
t
h
e
s
c
a
l
i
n
g
m
e
t
h
o
d
p
r
o
p
o
s
e
d
b
y
E
f
f
i
c
i
e
n
t
N
e
t
a
l
l
o
w
s
f
o
r
e
f
f
e
c
t
i
v
e
a
d
a
p
t
a
t
i
o
n
t
o
d
i
f
f
e
r
e
n
t
d
a
t
a
s
e
t
s
i
z
e
s
a
n
d
c
o
m
p
u
t
a
t
i
o
n
a
l
c
o
n
s
t
r
a
i
n
t
s
,
m
a
k
i
n
g
i
t
a
n
a
t
t
r
a
c
t
i
v
e
c
h
o
i
c
e
f
o
r
v
a
r
i
o
u
s
p
r
a
c
t
i
c
a
l
a
p
p
l
i
c
a
t
i
o
n
s
.
T
h
e
E
f
f
i
c
i
e
n
t
N
e
t
C
N
N
u
t
i
l
i
z
e
s
c
o
n
v
e
n
t
i
o
n
a
l
m
e
t
r
i
c
s
f
o
r
e
v
a
l
u
a
t
i
n
g
c
l
a
s
s
i
f
i
c
a
t
i
o
n
m
o
d
e
l
s
,
s
u
c
h
a
s
a
c
c
u
r
a
c
y
,
t
o
m
e
a
s
u
r
e
i
t
s
e
f
f
e
c
t
i
v
e
n
e
s
s
i
n
i
m
a
g
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
.
A
c
c
o
r
d
i
n
g
t
o
T
a
n
a
n
d
L
e
[
8
]
,
t
h
e
i
m
a
g
e
l
e
v
e
l
s
i
n
E
f
f
i
c
i
e
n
t
N
e
t
r
e
f
e
r
t
o
d
i
f
f
e
r
e
n
t
v
a
r
i
a
n
t
s
o
f
t
h
e
E
f
f
i
c
i
e
n
t
N
e
t
a
r
c
h
i
t
e
c
t
u
r
e
t
h
a
t
a
r
e
s
c
a
l
e
d
i
n
d
e
p
t
h
,
w
i
d
t
h
,
a
n
d
i
m
a
g
e
r
e
s
o
l
u
t
i
o
n
.
T
h
e
s
e
l
e
v
e
l
s
a
r
e
r
e
p
r
e
s
e
n
t
e
d
b
y
l
e
t
t
e
r
s
a
n
d
n
u
m
b
e
r
s
,
s
u
c
h
a
s
B
0
,
B
1
,
B
2
,
a
n
d
B
7
,
w
h
e
r
e
B
0
r
e
f
e
r
s
t
o
t
h
e
l
o
w
e
s
t
r
e
s
o
l
u
t
i
o
n
,
a
n
d
B
7
r
e
f
e
r
s
t
o
t
h
e
h
i
g
h
e
s
t
r
e
s
o
l
u
t
i
o
n
a
n
d
r
e
q
u
i
r
e
s
g
r
e
a
t
e
r
c
o
m
p
l
e
x
i
t
y
a
n
d
c
o
m
p
u
t
a
t
i
o
n
a
l
c
a
p
a
c
i
t
y
.
E
a
c
h
i
m
a
g
e
l
e
v
e
l
h
a
s
a
s
p
e
c
i
f
i
c
c
o
m
b
i
n
a
t
i
o
n
o
f
n
e
t
w
o
r
k
d
e
p
t
h
,
c
h
a
n
n
e
l
w
i
d
t
h
,
a
n
d
i
n
p
u
t
r
e
s
o
l
u
t
i
o
n
.
T
h
e
h
i
g
h
e
r
t
h
e
l
e
v
e
l
n
u
m
b
e
r
(
e
.
g
.
,
B
7
c
o
m
p
a
r
e
d
t
o
B
0
)
,
t
h
e
g
r
e
a
t
e
r
t
h
e
i
m
a
g
e
'
s
d
e
p
t
h
,
w
i
d
t
h
,
a
n
d
r
e
s
o
l
u
t
i
o
n
,
r
e
s
u
l
t
i
n
g
i
n
a
m
o
r
e
c
o
m
p
l
e
x
a
n
d
p
o
w
e
r
f
u
l
n
e
t
w
o
r
k
c
a
p
a
b
l
e
o
f
h
a
n
d
l
i
n
g
m
o
r
e
d
e
m
a
n
d
i
n
g
c
o
m
p
u
t
e
r
v
i
s
i
o
n
t
a
s
k
s
.
T
a
b
l
e
1
s
u
m
m
a
r
i
z
e
s
t
h
i
s
d
y
n
a
m
i
c
.
T
ab
le
1
.
E
f
f
icien
tNet
m
o
d
el
b
ase
ac
co
r
d
in
g
t
o
im
ag
e
r
eso
l
u
tio
n
M
o
d
e
l
b
a
s
e
R
e
s
o
l
u
t
i
o
n
Ef
f
i
c
i
e
n
t
N
e
t
B
0
2
2
4
Ef
f
i
c
i
e
n
t
N
e
t
B
1
2
4
0
Ef
f
i
c
i
e
n
t
N
e
t
B
2
2
6
0
Ef
f
i
c
i
e
n
t
N
e
t
B
3
3
0
0
Ef
f
i
c
i
e
n
t
N
e
t
B
4
3
8
0
Ef
f
i
c
i
e
n
t
N
e
t
B
5
4
5
6
Ef
f
i
c
i
e
n
t
N
e
t
B
6
5
2
8
Ef
f
i
c
i
e
n
t
N
e
t
B
7
6
0
0
T
h
is
s
ca
la
b
il
it
y
a
p
p
r
o
a
ch
a
llo
ws
E
f
f
ici
en
tNe
t
C
NNs
t
o
b
e
a
d
a
p
t
e
d
t
o
d
if
f
e
r
e
n
t
a
v
aila
b
le
co
m
p
u
tati
o
n
a
l r
es
o
u
r
ce
s
a
n
d
p
r
e
cisi
o
n
r
e
q
u
i
r
e
m
e
n
ts
,
p
r
o
v
i
d
i
n
g
a
f
le
x
i
b
l
e
an
d
e
f
f
ec
t
iv
e
s
o
l
u
tio
n
f
o
r
v
a
r
i
o
u
s
D
L
ap
p
li
ca
ti
o
n
s
in
c
o
m
p
u
te
r
v
is
i
o
n
.
Acc
o
r
d
i
n
g
t
o
T
an
a
n
d
L
e
[
8
]
,
ac
c
u
r
a
cy
is
a
n
ess
e
n
ti
al
m
etr
ic
f
o
r
ev
al
u
a
ti
n
g
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
E
f
f
ic
ien
tNe
t
C
NN
i
n
o
b
je
ct
c
lass
i
f
ic
ati
o
n
t
ask
s
.
T
h
ey
s
tat
e
t
h
a
t
E
f
f
i
ci
en
tNet
was
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
E
n
h
a
n
ce
d
cla
s
s
ifica
tio
n
o
f a
r
o
ma
tic
h
erb
s
u
s
in
g
E
fficien
tN
et
a
n
d
tr
a
n
s
fer
…
(
S
a
mir
a
N
a
s
cime
n
to
A
n
tu
n
es
)
4125
d
esi
g
n
e
d
t
o
m
a
x
i
m
iz
e
a
cc
u
r
a
cy
o
n
r
ef
er
e
n
ce
d
atas
ets
s
u
c
h
as
I
m
a
g
eN
et
,
e
n
s
u
r
i
n
g
t
h
a
t
t
h
e
m
o
d
el
c
an
m
a
k
e
ac
c
u
r
at
e
a
n
d
r
eli
ab
le
p
r
e
d
ic
ti
o
n
s
ac
r
o
s
s
v
a
r
i
o
u
s
o
b
je
ct
cla
s
s
es.
F
u
r
th
e
r
m
o
r
e
,
A
n
t
u
n
es
e
t
a
l
.
[
7
]
n
o
te
t
h
at
ac
c
u
r
a
cy
is
a
c
r
u
c
ial
m
e
tr
ic
f
o
r
d
et
er
m
i
n
i
n
g
E
f
f
ici
e
n
tNe
t'
s
p
r
ac
ti
ca
l
u
tili
ty
in
d
i
f
f
e
r
e
n
t
ap
p
l
i
ca
ti
o
n
s
.
T
h
e
y
a
ls
o
em
p
h
asi
ze
t
h
a
t
a
p
r
o
m
i
n
e
n
t
le
v
el
o
f
a
cc
u
r
ac
y
is
ess
e
n
ti
al
t
o
en
s
u
r
e
t
h
a
t
t
h
e
m
o
d
el
is
d
e
p
e
n
d
a
b
l
e
e
n
o
u
g
h
t
o
b
e
u
s
e
d
in
r
e
al
-
w
o
r
l
d
e
n
v
i
r
o
n
m
e
n
ts
,
wh
er
e
p
r
e
d
i
cti
o
n
ac
cu
r
a
c
y
i
s
p
ar
am
o
u
n
t
.
2
.
2
.
Dee
p lea
rning
,
co
nv
o
lutio
na
l neura
l net
wo
rk
,
a
nd
deep
t
ra
ns
f
er
lea
rning
T
h
e
c
o
n
c
e
p
t
s
o
f
D
L
,
C
N
N
s
,
a
n
d
d
e
e
p
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
(
D
T
L
)
h
a
v
e
b
e
e
n
c
r
u
c
i
a
l
i
n
t
h
e
f
i
e
l
d
o
f
A
I
.
T
h
e
y
d
e
m
o
n
s
t
r
a
t
e
t
h
e
a
b
i
l
i
t
y
t
o
m
a
n
a
g
e
c
o
m
p
l
e
x
d
a
t
a
a
n
d
i
n
c
r
e
a
s
i
n
g
l
y
s
o
p
h
i
s
t
i
c
a
t
e
d
t
a
s
k
s
,
r
e
s
u
l
t
i
n
g
i
n
a
p
o
s
i
t
i
v
e
i
m
p
a
c
t
d
u
e
t
o
t
h
e
i
r
w
i
d
e
r
a
n
g
e
o
f
a
p
p
l
i
c
a
t
i
o
n
s
i
n
v
a
r
i
o
u
s
f
i
e
l
d
s
.
T
h
e
g
r
o
w
i
n
g
p
o
p
u
l
a
r
i
t
y
o
f
c
l
a
s
s
i
f
i
c
a
t
i
o
n
m
e
t
h
o
d
s
s
u
c
h
a
s
d
e
e
p
C
N
N
s
e
n
h
a
n
c
e
s
AI
'
s
p
o
t
e
n
t
i
a
l
i
n
m
u
l
t
i
p
l
e
d
i
s
c
i
p
l
i
n
e
s
,
p
a
r
t
i
c
u
l
a
r
l
y
c
o
m
p
u
t
e
r
v
i
s
i
o
n
[
1
0
]
.
DL
is
a
co
m
p
u
ter
-
im
p
lem
e
n
ted
alg
o
r
ith
m
th
at
u
tili
ze
s
a
n
eu
r
al
n
etwo
r
k
t
o
s
elec
t
d
at
a
f
ea
tu
r
es
au
to
m
atica
lly
.
AI
d
o
es
n
o
t
r
e
q
u
ir
e
an
y
ex
tr
a
e
n
h
an
ce
m
en
ts
.
I
t
co
m
b
i
n
es
lo
w
-
lev
el
attr
ib
u
tes
to
cr
ea
te
h
ig
h
-
lev
el
attr
ib
u
tes
th
at
d
escr
ib
e
t
h
e
d
is
tr
ib
u
ted
n
atu
r
e
o
f
th
e
d
a
ta
an
d
s
am
p
le
attr
ib
u
tes.
C
o
m
p
ar
ed
t
o
tr
ad
itio
n
al
im
ag
e
r
ec
o
g
n
itio
n
an
d
d
etec
ti
o
n
m
eth
o
d
s
,
DL
is
m
o
r
e
ef
f
ec
tiv
e
in
im
ag
e
r
ec
o
g
n
itio
n
an
d
tar
g
et
d
etec
tio
n
.
T
o
d
ay
,
th
e
m
ain
ty
p
es o
f
n
etw
o
r
k
s
ar
e
m
u
ltil
ay
er
p
e
r
ce
p
tr
o
n
’
s
,
C
NNs,
an
d
r
ec
u
r
r
e
n
t
n
eu
r
a
l
n
etwo
r
k
s
(
R
NNs)
[
1
1
]
.
T
h
e
m
o
s
t
co
m
m
o
n
ty
p
e
o
f
DL
f
o
r
s
p
atial
p
atter
n
an
al
y
s
is
is
a
C
NN
(
also
ca
l
led
C
o
n
v
Nets).
C
NNs
a
im
to
ac
q
u
ir
e
s
p
atial
s
k
ills
,
s
u
ch
as
ed
g
es,
c
o
r
n
er
s
,
tex
tu
r
es,
o
r
o
th
er
s
h
ap
es
th
at
m
o
r
e
ac
cu
r
ately
d
escr
ib
e
t
h
e
in
ten
d
ed
class
o
r
v
o
lu
m
e
[
1
1
]
.
DL
m
eth
o
d
s
,
esp
ec
ially
th
o
s
e
b
ased
o
n
C
NNs,
ar
e
p
o
p
u
lar
in
th
e
ag
r
icu
ltu
r
al
d
o
m
ain
f
o
r
d
etec
tio
n
an
d
cla
s
s
if
icatio
n
p
u
r
p
o
s
es
[
1
1
]
.
T
h
e
co
r
e
to
lear
n
in
g
th
ese
f
ea
tu
r
es
is
a
s
er
ies
o
f
tr
an
s
f
o
r
m
atio
n
s
o
f
th
e
in
p
u
t
d
ata
th
at
o
cc
u
r
at
d
if
f
er
e
n
t
s
p
at
ial
s
ca
les
(
e.
g
.
,
th
r
o
u
g
h
p
o
o
li
n
g
)
.
T
h
is
f
ac
ilit
ates
id
en
tify
in
g
a
n
d
c
o
m
b
in
in
g
lo
w
-
lev
el
p
r
o
p
er
ties
an
d
co
n
ce
p
tu
al
id
ea
s
[
1
1
]
.
Acc
o
r
d
in
g
to
L
u
et
a
l
.
[
1
1
]
,
C
NNs
ty
p
ically
co
m
p
r
is
e
co
n
v
o
lu
tio
n
al,
p
o
o
lin
g
,
an
d
f
u
lly
co
n
n
ec
te
d
lay
er
s
.
T
h
e
c
o
n
v
o
lu
tio
n
al
lay
e
r
co
n
s
id
er
s
th
e
lo
ca
l
ass
o
ciatio
n
o
f
im
ag
e
co
n
te
n
t
to
d
er
iv
e
f
ea
tu
r
es.
L
ik
e
o
th
er
ty
p
ical
n
e
u
r
al
n
etwo
r
k
m
o
d
els,
C
NNs
co
n
s
is
t
o
f
n
e
u
r
o
n
s
o
r
g
an
ized
in
lay
er
s
an
d
th
u
s
ca
p
ab
le
o
f
lear
n
in
g
co
m
p
lex
r
ep
r
esen
tatio
n
s
,
with
n
eu
r
o
n
s
b
etwe
en
la
y
er
s
co
n
n
ec
ted
th
r
o
u
g
h
weig
h
ts
an
d
b
ia
s
es.
T
h
e
f
ir
s
t
la
y
er
is
th
e
in
p
u
t
lay
e
r
,
f
o
r
ex
a
m
p
l
e,
r
em
o
te
s
en
s
in
g
d
ata,
a
n
d
t
h
e
f
in
al
lay
e
r
is
th
e
o
u
tp
u
t,
s
u
ch
as
a
p
r
ed
icted
class
if
icatio
n
o
f
p
lan
t
s
p
ec
ies.
B
etwe
en
th
e
lay
er
s
ar
e
h
id
d
e
n
tr
an
s
f
o
r
m
atio
n
s
th
at
alter
h
o
w
th
e
in
p
u
t
s
p
ac
e
is
tr
an
s
f
o
r
m
ed
i
n
to
o
n
e
t
h
at
alig
n
s
with
th
e
o
u
tp
u
t.
C
NNs h
av
e
at
least o
n
e
co
n
v
o
lu
tio
n
al
lay
er
as a
h
id
d
en
la
y
e
r
th
at
ex
p
lo
r
es
p
atter
n
s
.
Oth
er
n
o
n
-
c
o
n
v
o
lu
tio
n
al
lay
er
s
ca
n
also
b
e
in
co
r
p
o
r
ated
.
T
h
e
c
o
n
v
o
lu
ti
o
n
al
lay
e
r
s
in
clu
d
e
v
ar
i
o
u
s
f
ilter
s
th
at
ca
n
b
e
o
p
tim
ize
d
;
th
ese
lay
er
s
tr
an
s
f
o
r
m
th
e
in
p
u
t
o
r
p
r
e
v
io
u
s
h
id
d
e
n
lay
er
in
f
o
r
m
atio
n
.
T
h
e
n
u
m
b
e
r
o
f
f
ilter
s
d
eter
m
in
es
th
e
th
ic
k
n
ess
o
f
a
co
n
v
o
lu
tio
n
al
lay
er
.
T
h
e
r
esu
ltin
g
tr
an
s
f
o
r
m
atio
n
s
aim
to
ex
p
o
s
e
s
ig
n
if
ican
t p
atter
n
s
th
at
ad
d
r
e
s
s
th
e
p
r
o
b
lem
at
h
a
n
d
[
1
2
]
.
D
T
L
is
a
m
e
t
h
o
d
t
h
a
t
a
t
t
e
m
p
t
s
t
o
r
e
d
u
c
e
d
e
p
e
n
d
e
n
c
e
a
n
d
c
o
s
t
s
b
y
u
t
i
li
z
i
n
g
k
n
o
w
le
d
g
e
f
r
o
m
a
s
o
u
r
c
e
t
a
s
k
t
o
t
r
a
i
n
a
ta
r
g
e
t
t
as
k
[
1
3
]
.
T
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
r
e
l
a
x
e
s
t
h
e
r
e
q
u
i
r
e
m
e
n
t
t
h
a
t
t
r
ai
n
i
n
g
a
n
d
t
e
s
t
i
n
g
d
a
t
a
b
e
i
n
d
e
p
e
n
d
e
n
t
a
n
d
i
d
e
n
t
i
c
a
l
l
y
d
i
s
t
r
i
b
u
t
e
d
,
w
h
i
c
h
m
o
t
i
v
at
e
s
t
h
e
u
s
e
o
f
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
t
o
a
d
d
r
e
s
s
t
h
e
p
r
o
b
l
e
m
o
f
i
n
s
u
f
f
i
c
i
e
n
t
t
r
ai
n
i
n
g
d
a
t
a
.
I
n
tr
a
n
s
f
e
r
l
ea
r
n
i
n
g
,
t
h
e
p
r
o
c
es
s
o
f
t
r
a
i
n
i
n
g
a
n
d
t
e
s
t
i
n
g
d
a
ta
d
o
e
s
n
o
t
n
e
e
d
t
o
b
e
i
d
e
n
t
i
c
a
ll
y
d
i
s
t
r
i
b
u
t
e
d
,
a
n
d
t
h
e
m
o
d
e
l
i
n
t
h
e
t
a
r
g
e
t
d
o
m
a
i
n
d
o
e
s
n
o
t
n
e
e
d
t
o
b
e
t
r
a
i
n
e
d
f
r
o
m
s
c
r
a
t
c
h
,
w
h
i
c
h
c
a
n
s
i
g
n
i
f
i
c
a
n
t
l
y
r
e
d
u
c
e
t
h
e
d
e
m
a
n
d
f
o
r
t
r
a
i
n
i
n
g
d
at
a
a
n
d
t
h
e
t
i
m
e
r
e
q
u
i
r
e
d
t
o
t
r
a
i
n
t
h
e
m
o
d
el
in
t
h
e
ta
r
g
e
t
d
o
m
a
in
[
1
4
]
,
[
1
5
]
.
T
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
a
l
l
o
w
s
y
o
u
t
o
l
e
v
e
r
a
g
e
w
h
a
t
y
o
u
h
a
v
e
l
e
a
r
n
e
d
i
n
o
n
e
s
i
t
u
at
i
o
n
t
o
l
e
a
r
n
m
o
r
e
r
a
p
i
d
l
y
i
n
a
n
o
t
h
e
r
.
I
t
i
s
c
o
m
m
o
n
l
y
u
s
e
d
f
o
r
o
b
j
e
c
t
r
e
c
o
g
n
i
t
i
o
n
a
n
d
i
m
a
g
e
c
l
a
s
s
i
f
i
c
a
ti
o
n
t
h
r
o
u
g
h
p
r
e
-
t
r
a
i
n
e
d
C
N
N
m
o
d
e
l
s
.
D
i
f
f
e
r
e
n
t
m
e
t
h
o
d
s
o
f
a
l
t
e
r
i
n
g
a
p
r
e
-
t
r
a
i
n
e
d
C
N
N
a
n
d
w
h
e
n
t
o
u
s
e
e
a
c
h
m
e
t
h
o
d
a
r
e
o
f
g
r
e
a
t
i
n
t
e
r
e
s
t
f
o
r
r
e
s
e
a
r
c
h
.
T
h
e
e
f
f
ec
t
o
f
e
a
c
h
m
e
t
h
o
d
o
n
t
r
a
i
n
i
n
g
t
i
m
e
a
n
d
t
est
s
et
a
c
c
u
r
a
c
y
a
f
f
ec
t
s
t
h
e
wa
y
t
r
a
n
s
f
e
r
l
ea
r
n
i
n
g
i
s
e
m
p
l
o
y
e
d
.
T
h
e
h
e
r
b
s
s
e
l
ec
t
e
d
f
o
r
a
n
a
l
y
s
is
i
n
cl
u
d
e
d
L
a
u
r
u
s
n
o
b
i
l
i
s
,
R
o
s
m
a
r
i
n
u
s
o
f
fi
c
i
n
a
l
is
,
an
d
M
e
n
t
h
a
s
p
i
c
a
t
a
.
T
h
e
b
a
y
l
e
a
f
(
L
a
u
r
u
s
n
o
b
i
l
is
)
i
s
c
h
a
r
a
c
t
e
r
i
z
e
d
b
y
l
a
n
ce
o
l
a
te
l
e
a
v
es
wi
t
h
a
n
al
t
e
r
n
a
te
a
r
r
a
n
g
e
m
e
n
t
a
n
d
s
h
o
r
t
p
e
t
i
o
l
es
,
m
e
a
s
u
r
i
n
g
5
-
8
c
m
i
n
le
n
g
t
h
a
n
d
3
-
4
c
m
i
n
w
i
d
t
h
.
T
h
e
s
e
l
e
a
v
es
a
r
e
l
ea
t
h
e
r
y
,
p
u
n
c
t
a
te
,
w
i
t
h
r
e
v
o
l
u
t
e
a
n
d
w
a
v
y
e
n
t
i
r
e
m
a
r
g
i
n
s
.
T
h
e
u
p
p
e
r
s
u
r
f
a
c
e
is
g
l
a
b
r
o
u
s
a
n
d
g
l
o
s
s
y
,
r
a
n
g
i
n
g
f
r
o
m
o
li
v
e
g
r
e
e
n
to
b
r
o
w
n
,
w
h
i
le
t
h
e
l
o
w
e
r
s
u
r
f
a
c
e
e
x
h
i
b
i
t
s
a
d
u
l
l
o
li
v
e
g
r
e
e
n
t
o
b
r
o
w
n
c
o
l
o
r
at
i
o
n
w
i
t
h
a
p
r
o
m
i
n
e
n
t
m
i
d
r
i
b
a
n
d
v
e
i
n
s
.
L
a
u
r
u
s
n
o
b
i
li
s
i
s
wi
d
e
l
y
u
s
e
d
i
n
t
h
e
f
o
o
d
a
n
d
p
h
a
r
m
a
c
e
u
t
i
c
a
l i
n
d
u
s
t
r
i
e
s
d
u
e
to
i
t
s
a
n
t
i
o
x
i
d
a
n
t
a
n
d
a
n
t
i
m
ic
r
o
b
i
a
l
p
r
o
p
e
r
t
i
es
[
1
6
]
.
R
o
s
m
a
r
i
n
u
s
o
f
f
ic
i
n
a
l
is
,
f
r
o
m
t
h
e
L
a
m
i
a
c
e
a
e
f
a
m
i
l
y
,
is
a
n
a
r
o
m
a
t
i
c
p
e
r
e
n
n
i
a
l
p
l
a
n
t
w
it
h
e
r
e
c
t
s
t
e
m
s
a
n
d
b
l
u
e
-
w
h
i
t
e
f
l
o
w
e
r
s
.
C
o
m
m
o
n
l
y
k
n
o
w
n
a
s
r
o
s
e
m
a
r
y
a
n
d
n
a
t
i
v
e
t
o
t
h
e
M
e
d
i
t
e
r
r
a
n
e
a
n
r
e
g
i
o
n
,
i
t
s
f
r
e
s
h
a
n
d
d
r
i
e
d
l
e
a
v
e
s
a
r
e
w
i
d
e
l
y
u
s
e
d
as
a
s
e
as
o
n
i
n
g
a
n
d
f
o
r
h
e
r
b
a
l
t
e
a
p
r
e
p
a
r
a
ti
o
n
.
I
t
i
s
p
a
r
t
i
c
u
l
a
r
l
y
n
o
t
e
d
f
o
r
i
ts
a
n
t
i
-
i
n
f
l
a
m
m
a
t
o
r
y
p
r
o
p
e
r
t
i
e
s
d
e
m
o
n
s
t
r
at
e
d
i
n
p
r
e
c
l
i
n
i
c
al
i
n
v
i
v
o
m
o
d
e
l
s
[
1
7
]
.
F
i
n
a
l
l
y
,
s
p
e
a
r
m
i
n
t
(
M
e
n
t
h
a
s
p
i
c
a
t
a
)
i
s
d
i
s
t
i
n
g
u
i
s
h
ed
b
y
i
t
s
o
p
p
o
s
i
t
e
le
a
v
e
s
wi
t
h
s
h
o
r
t
p
e
t
i
o
l
e
s
a
n
d
o
b
l
o
n
g
t
o
o
v
a
l
s
h
a
p
e
s
wi
t
h
s
e
r
r
at
e
d
m
a
r
g
i
n
s
.
T
h
i
s
h
e
r
b
p
l
a
y
s
a
c
r
u
c
i
a
l
r
o
l
e
d
u
e
t
o
i
t
s
r
i
c
h
c
o
m
p
o
s
i
t
i
o
n
o
f
p
h
e
n
o
l
i
c
c
o
m
p
o
u
n
d
s
,
w
h
i
c
h
a
r
e
e
f
f
e
c
t
i
v
e
i
n
t
r
ea
t
i
n
g
c
a
r
d
i
o
v
a
s
c
u
l
a
r
d
i
s
e
as
e
s
.
A
d
d
i
t
i
o
n
al
l
y
,
i
t
i
s
f
r
eq
u
e
n
t
l
y
u
s
e
d
a
s
a
s
e
as
o
n
i
n
g
a
n
d
i
n
i
n
f
u
s
i
o
n
s
[
1
8
]
.
2
.
3
.
M
et
ho
ds
T
h
is
r
esear
ch
f
o
llo
wed
a
s
tr
u
ctu
r
ed
e
x
p
er
im
e
n
tal
p
r
o
ce
d
u
r
e
to
ass
ess
th
e
class
if
icatio
n
o
f
ar
o
m
atic
h
er
b
s
u
s
in
g
th
e
E
f
f
icien
tNet
C
NN
wi
th
tr
an
s
f
er
lear
n
in
g
.
T
h
e
m
eth
o
d
o
lo
g
y
is
p
r
esen
ted
in
two
m
ain
p
h
ases
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
.
5
,
Octo
b
er
2
0
2
5
:
4
1
2
3
-
4
1
3
6
4126
to
allo
w
f
o
r
r
ep
licatio
n
b
y
o
th
er
r
esear
ch
er
s
:
i)
liter
atu
r
e
r
ev
iew
an
d
d
ataset
p
r
ep
ar
atio
n
,
a
n
d
ii)
ex
p
er
im
e
n
tal
m
o
d
el
tr
ain
in
g
an
d
ev
alu
atio
n
.
2
.
3
.
1
.
L
it
er
a
t
ure
re
v
iew
a
nd
da
t
a
s
et
prepa
ra
t
io
n
I
n
i
t
i
a
l
l
y
,
a
c
o
m
p
r
e
h
e
n
s
i
v
e
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
w
a
s
c
o
n
d
u
c
t
e
d
t
o
p
r
o
v
i
d
e
a
s
o
l
i
d
t
h
e
o
r
e
t
i
c
a
l
f
o
u
n
d
a
t
i
o
n
o
n
r
e
l
e
v
a
n
t
t
o
p
i
c
s
,
i
n
c
l
u
d
i
n
g
D
L
,
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
s
,
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
,
a
n
d
E
f
f
i
c
i
e
n
t
N
e
t
’
s
s
p
e
c
i
f
i
c
a
p
p
l
i
c
a
t
i
o
n
t
o
p
l
a
n
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
.
T
h
i
s
r
e
v
i
e
w
a
i
m
e
d
t
o
i
d
e
n
t
i
f
y
k
n
o
w
l
e
d
g
e
g
a
p
s
a
n
d
v
a
l
i
d
a
t
e
E
f
f
i
c
i
e
n
t
N
e
t
’
s
s
u
i
t
a
b
i
l
i
t
y
f
o
r
t
h
i
s
t
a
s
k
.
T
h
e
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
i
s
e
s
s
e
n
t
i
a
l
f
o
r
u
n
d
e
r
s
t
a
n
d
i
n
g
t
h
e
e
v
o
l
u
t
i
o
n
o
f
C
N
N
s
,
e
n
a
b
l
i
n
g
t
h
e
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
a
d
v
a
n
c
e
m
e
n
t
s
,
c
h
a
l
l
e
n
g
e
s
,
a
n
d
r
e
s
e
a
r
c
h
o
p
p
o
r
t
u
n
i
t
i
e
s
.
R
e
c
e
n
t
s
t
u
d
i
e
s
e
m
p
h
a
s
i
z
e
t
h
e
i
r
i
m
p
o
r
t
a
n
c
e
i
n
a
n
a
l
y
z
i
n
g
a
r
c
h
i
t
e
c
t
u
r
e
s
,
o
p
t
i
m
i
z
i
n
g
m
o
d
e
l
s
,
a
n
d
e
x
p
l
o
r
i
n
g
p
r
a
c
t
i
c
a
l
a
p
p
l
i
c
a
t
i
o
n
s
[
1
9
]
.
C
o
n
t
e
m
p
o
r
a
r
y
a
u
t
h
o
r
s
e
m
p
h
a
s
i
z
e
t
h
e
r
e
l
e
v
a
n
c
e
o
f
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
s
i
n
c
o
n
s
t
r
u
c
t
i
n
g
s
c
i
e
n
t
i
f
i
c
k
n
o
w
l
e
d
g
e
.
L
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
s
i
s
a
s
t
u
d
y
t
h
a
t
e
x
a
m
i
n
e
s
a
n
d
s
y
n
t
h
e
s
i
z
e
s
a
n
e
x
i
s
t
i
n
g
b
o
d
y
o
f
l
i
t
e
r
a
t
u
r
e
b
y
i
d
e
n
t
i
f
y
i
n
g
,
q
u
e
s
t
i
o
n
i
n
g
,
a
n
d
d
e
v
e
l
o
p
i
n
g
t
h
e
f
o
u
n
d
a
t
i
o
n
s
o
f
a
t
h
e
o
r
y
t
h
r
o
u
g
h
t
h
e
a
n
a
l
y
s
i
s
o
f
a
b
o
d
y
(
o
r
m
u
l
t
i
p
l
e
b
o
d
i
e
s
)
o
f
p
r
i
o
r
w
o
r
k
[
2
0
]
.
A
d
d
i
t
i
o
n
a
l
l
y
,
E
b
i
d
o
r
a
n
d
I
k
h
i
d
e
[
2
1
]
h
i
g
h
l
i
g
h
t
t
h
e
i
m
p
o
r
t
a
n
c
e
o
f
i
n
c
l
u
d
i
n
g
d
i
v
e
r
s
e
s
o
u
r
c
e
s
,
s
u
c
h
a
s
t
e
c
h
n
i
c
a
l
r
e
p
o
r
t
s
a
n
d
s
p
e
c
i
a
l
i
z
e
d
j
o
u
r
n
a
l
s
,
t
o
e
n
s
u
r
e
a
c
o
m
p
r
e
h
e
n
s
i
v
e
a
p
p
r
o
a
c
h
i
n
t
h
e
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
.
T
h
e
a
i
m
o
f
t
h
i
s
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
w
a
s
t
o
s
e
e
k
t
h
e
m
o
s
t
r
e
c
e
n
t
c
o
n
t
r
i
b
u
t
i
o
n
s
t
o
u
n
d
e
r
s
t
a
n
d
i
n
g
t
h
e
t
o
p
i
c
.
A
c
c
o
r
d
i
n
g
t
o
B
a
r
r
y
e
t
a
l
.
[
2
2
]
,
r
e
g
u
l
a
r
l
y
u
p
d
a
t
i
n
g
t
h
e
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
i
s
c
r
u
c
i
a
l
t
o
c
o
n
t
e
x
t
u
a
l
i
z
i
n
g
t
h
e
s
t
u
d
y
w
i
t
h
i
n
t
h
e
s
t
a
t
e
o
f
t
h
e
a
r
t
.
T
h
e
c
a
r
e
f
u
l
s
e
l
e
c
t
i
o
n
o
f
s
o
u
r
c
e
e
n
s
u
r
e
s
t
h
e
r
e
l
i
a
b
i
l
i
t
y
a
n
d
r
e
l
e
v
a
n
c
e
o
f
t
h
e
i
n
f
o
r
m
a
t
i
o
n
,
w
h
i
c
h
a
r
e
e
s
s
e
n
t
i
a
l
f
o
r
b
u
i
l
d
i
n
g
a
c
o
n
s
i
s
t
e
n
t
t
h
e
o
r
e
t
i
c
a
l
f
r
a
m
e
w
o
r
k
[
2
3
]
.
Z
o
p
h
e
t
a
l
.
[
2
4
]
i
n
t
r
o
d
u
c
e
d
g
r
o
u
n
d
b
r
e
a
k
i
n
g
a
p
p
r
o
a
c
h
t
o
s
c
a
l
a
b
l
e
i
m
a
g
e
r
e
c
o
g
n
i
t
i
o
n
b
y
d
e
v
e
l
o
p
i
n
g
t
r
a
n
s
f
e
r
a
b
l
e
a
r
c
h
i
t
e
c
t
u
r
e
s
t
h
r
o
u
g
h
a
u
t
o
m
a
t
e
d
n
e
u
r
a
l
a
r
c
h
i
t
e
c
t
u
r
e
s
e
a
r
c
h
,
s
i
g
n
i
f
i
c
a
n
t
l
y
i
m
p
r
o
v
i
n
g
e
f
f
i
c
i
e
n
c
y
a
n
d
a
c
c
u
r
a
c
y
i
n
i
m
a
g
e
r
e
c
o
g
n
i
t
i
o
n
.
T
h
i
s
w
o
r
k
l
a
i
d
f
o
u
n
d
a
t
i
o
n
f
o
r
f
u
r
t
h
e
r
a
d
v
a
n
c
e
m
e
n
t
s
i
n
e
f
f
i
c
i
e
n
t
m
o
d
e
l
d
e
s
i
g
n
,
s
u
c
h
a
s
E
f
f
i
c
i
e
n
t
N
e
t
,
p
a
r
t
i
c
u
l
a
r
l
y
i
n
r
e
s
o
u
r
c
e
-
c
o
n
s
t
r
a
i
n
e
d
e
n
v
i
r
o
n
m
e
n
t
s
.
T
h
e
s
e
i
m
p
r
o
v
e
m
e
n
t
s
i
n
n
e
t
w
o
r
k
a
r
c
h
i
t
e
c
t
u
r
e
s
h
a
v
e
f
a
c
i
l
i
t
a
t
e
d
t
h
e
d
e
v
e
l
o
p
m
e
n
t
o
f
m
o
r
e
s
p
e
c
i
a
l
i
z
e
d
t
e
c
h
n
i
q
u
e
s
,
s
u
c
h
a
s
s
e
m
a
n
t
i
c
s
e
g
m
e
n
t
a
t
i
o
n
,
w
h
i
c
h
a
r
e
e
s
s
e
n
t
i
a
l
f
o
r
a
p
p
l
i
c
a
t
i
o
n
s
r
e
q
u
i
r
i
n
g
h
i
g
h
p
r
e
c
i
s
i
o
n
i
n
v
i
s
u
a
l
d
e
t
a
i
l
.
C
o
n
s
e
q
u
e
n
t
l
y
,
t
h
e
a
d
o
p
t
i
o
n
o
f
a
d
v
a
n
c
e
d
s
e
m
a
n
t
i
c
s
e
g
m
e
n
t
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
t
o
e
n
s
u
r
e
t
h
e
p
r
e
s
e
r
v
a
t
i
o
n
o
f
f
i
n
e
d
e
t
a
i
l
s
i
n
i
m
a
g
e
s
i
s
c
r
u
c
i
a
l
f
o
r
d
i
f
f
e
r
e
n
t
i
a
t
i
n
g
v
i
s
u
a
l
l
y
s
i
m
i
l
a
r
h
e
r
b
s
.
T
h
e
u
s
e
o
f
m
e
t
h
o
d
o
l
o
g
i
e
s
l
i
k
e
t
h
o
s
e
p
r
e
s
e
n
t
e
d
b
y
S
h
i
v
a
p
r
a
s
a
d
a
n
d
W
a
d
h
a
w
a
n
[
2
5
]
h
i
g
h
l
i
g
h
t
s
s
e
g
m
e
n
t
a
t
i
o
n
s
t
r
a
t
e
g
i
e
s
t
h
a
t
e
m
p
h
a
s
i
z
e
t
h
e
p
r
e
s
e
r
v
a
t
i
o
n
o
f
s
u
b
t
l
e
a
n
d
e
s
s
e
n
t
i
a
l
f
e
a
t
u
r
e
s
.
T
h
i
s
a
p
p
r
o
a
c
h
i
s
v
i
t
a
l
f
o
r
m
a
i
n
t
a
i
n
i
n
g
t
h
e
i
n
t
e
g
r
i
t
y
o
f
v
i
s
u
a
l
i
n
f
o
r
m
a
t
i
o
n
d
u
r
i
n
g
i
m
a
g
e
p
r
o
c
e
s
s
i
n
g
,
e
n
a
b
l
i
n
g
m
o
r
e
a
c
c
u
r
a
t
e
a
n
d
e
f
f
e
c
t
i
v
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
b
y
e
n
h
a
n
c
i
n
g
t
h
e
c
r
i
t
i
c
a
l
d
e
t
a
i
l
s
t
h
a
t
d
i
s
t
i
n
g
u
i
s
h
e
a
c
h
s
p
e
c
i
e
s
.
T
h
e
d
a
t
a
s
e
t
s
e
l
e
c
t
e
d
f
o
r
m
o
d
e
l
t
r
a
i
n
i
n
g
c
o
n
s
i
s
t
e
d
o
f
i
m
a
g
e
s
o
f
t
h
r
e
e
h
e
r
b
s
p
e
c
i
e
s
:
r
o
s
e
m
a
r
y
,
m
i
n
t
,
a
n
d
b
a
y
l
e
a
f
.
T
h
e
h
e
r
b
s
w
e
r
e
c
h
o
s
e
n
b
a
s
e
d
o
n
t
h
e
i
r
v
i
s
u
a
l
s
i
m
i
l
a
r
i
t
i
e
s
i
n
c
o
l
o
r
,
t
e
x
t
u
r
e
,
a
n
d
s
h
a
p
e
,
w
h
i
c
h
p
r
e
s
e
n
t
c
h
a
l
l
e
n
g
e
s
f
o
r
i
m
a
g
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
.
T
h
e
i
m
a
g
e
s
w
e
r
e
g
a
t
h
e
r
e
d
f
r
o
m
o
p
e
n
s
o
u
r
c
e
s
a
n
d
c
a
t
e
g
o
r
i
z
e
d
b
a
s
e
d
o
n
t
h
e
i
r
v
i
s
u
a
l
q
u
a
l
i
t
y
a
n
d
r
e
l
e
v
a
n
c
e
,
c
o
n
t
a
i
n
i
n
g
i
n
d
i
v
i
d
u
a
l
l
e
a
f
i
m
a
g
e
s
a
s
w
e
l
l
a
s
i
n
s
t
a
n
c
e
s
o
f
l
e
a
v
e
s
i
n
b
u
n
d
l
e
s
a
n
d
m
i
x
e
d
w
i
t
h
o
t
h
e
r
o
b
j
e
c
t
s
.
T
h
i
s
d
i
v
e
r
s
i
t
y
w
a
s
i
n
t
e
n
d
e
d
t
o
r
e
p
l
i
c
a
t
e
r
e
a
l
-
w
o
r
l
d
c
o
n
d
i
t
i
o
n
s
b
u
t
a
l
s
o
l
e
d
t
o
s
o
m
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
c
h
a
l
l
e
n
g
e
s
.
G
i
v
e
n
t
h
e
c
h
a
ll
e
n
g
e
s
o
f
t
e
n
i
m
p
o
s
e
d
b
y
t
h
e
d
a
t
as
e
t
s
i
z
e
a
n
d
th
e
n
e
e
d
t
o
i
n
c
r
e
as
e
m
o
d
el
r
o
b
u
s
t
n
e
s
s
,
t
h
e
i
n
c
o
r
p
o
r
a
t
i
o
n
o
f
s
el
f
-
s
u
p
e
r
v
i
s
e
d
l
e
a
r
n
i
n
g
(
SS
L
)
t
e
c
h
n
i
q
u
e
s
c
a
n
b
e
h
i
g
h
l
y
e
f
f
ec
t
i
v
e
.
A
b
d
u
l
r
a
z
z
a
q
e
t
a
l
.
[
2
6
]
h
i
g
h
l
i
g
h
t
e
d
m
e
t
h
o
d
s
li
k
e
m
i
s
s
i
n
g
p
a
r
t
p
r
e
d
i
c
t
i
o
n
a
n
d
a
u
t
o
m
a
t
i
c
c
o
l
o
r
i
za
t
i
o
n
o
f
b
l
a
c
k
-
a
n
d
-
w
h
i
t
e
i
m
a
g
es
o
f
f
e
r
p
r
o
m
i
s
i
n
g
a
p
p
r
o
a
c
h
e
s
.
T
h
e
s
e
s
e
l
f
-
s
u
p
e
r
v
i
s
e
d
m
e
t
h
o
d
s
e
n
a
b
le
t
h
e
m
o
d
e
l
t
o
d
e
v
e
l
o
p
a
d
e
e
p
e
r
u
n
d
e
r
s
t
a
n
d
i
n
g
o
f
t
h
e
i
n
t
r
i
n
s
i
c
f
e
a
t
u
r
es
o
f
h
e
r
b
s
w
i
t
h
o
u
t
r
el
y
i
n
g
e
x
c
l
u
s
i
v
e
l
y
o
n
l
a
b
e
l
e
d
d
at
a
.
S
S
L
h
a
s
s
h
o
wn
g
r
e
a
t
p
o
t
e
n
ti
a
l
i
n
e
x
t
r
a
c
t
i
n
g
d
i
s
c
r
i
m
i
n
a
ti
v
e
r
e
p
r
e
s
e
n
t
a
ti
o
n
s
,
w
h
i
c
h
a
r
e
c
r
u
c
i
a
l
f
o
r
t
h
e
p
r
e
c
i
s
e
d
i
s
t
i
n
ct
i
o
n
b
e
t
we
e
n
v
i
s
u
a
l
l
y
s
i
m
i
l
a
r
a
r
o
m
a
t
i
c
h
e
r
b
s
,
al
l
o
wi
n
g
f
o
r
m
o
r
e
e
f
f
i
c
i
e
n
t
le
a
r
n
i
n
g
a
n
d
m
o
r
e
a
c
c
u
r
a
t
e
c
la
s
s
i
f
i
ca
t
i
o
n
.
I
n
p
a
r
al
l
e
l
,
t
h
e
p
r
e
c
is
i
o
n
i
n
n
e
u
r
a
l
n
e
t
w
o
r
k
s
e
le
c
t
i
o
n
a
n
d
f
e
a
t
u
r
e
o
p
t
i
m
i
z
at
i
o
n
p
l
a
y
s
a
k
e
y
r
o
l
e
i
n
e
n
h
a
n
c
i
n
g
c
l
a
s
s
i
f
i
c
a
ti
o
n
p
e
r
f
o
r
m
a
n
c
e
.
T
h
i
s
i
s
w
el
l
il
l
u
s
t
r
a
t
e
d
in
[
2
7
]
,
w
h
i
ch
d
e
m
o
n
s
t
r
a
t
e
s
t
h
e
c
r
o
s
s
-
d
i
s
c
ip
l
i
n
a
r
y
a
p
p
l
i
c
a
b
i
li
t
y
o
f
t
h
e
s
e
t
ec
h
n
i
q
u
e
s
.
T
h
e
s
t
u
d
y
s
h
o
w
s
h
o
w
s
t
r
a
te
g
i
c
a
d
j
u
s
t
m
en
t
s
i
n
m
o
d
e
l
s
e
le
c
t
i
o
n
a
n
d
d
a
ta
t
r
e
a
t
m
e
n
t
ca
n
s
i
g
n
i
f
i
c
a
n
tl
y
im
p
r
o
v
e
c
l
a
s
s
i
f
ic
a
t
i
o
n
o
u
t
c
o
m
e
s
,
o
f
f
e
r
i
n
g
v
a
l
u
a
b
l
e
i
n
s
i
g
h
ts
f
o
r
a
d
d
r
e
s
s
i
n
g
s
i
m
i
l
a
r
c
h
a
l
l
e
n
g
e
s
i
n
o
t
h
e
r
t
e
c
h
n
i
ca
l
d
o
m
a
i
n
s
.
Mo
r
e
o
v
e
r
,
i
m
a
g
e
q
u
a
lit
y
is
f
u
n
d
a
m
e
n
ta
l
t
o
th
e
ac
cu
r
a
c
y
o
f
t
h
e
cl
ass
i
f
ic
ati
o
n
p
r
o
ce
s
s
.
T
h
e
in
t
eg
r
ati
o
n
o
f
e
n
h
a
n
c
em
e
n
t
m
et
h
o
d
s
,
s
u
c
h
as
d
e
b
l
u
r
r
i
n
g
an
d
as
y
m
m
e
tr
i
c
s
p
ati
al
at
te
n
ti
o
n
,
as
p
r
es
en
te
d
i
n
t
h
e
s
tu
d
y
[
2
8
]
,
ca
n
s
i
g
n
i
f
i
ca
n
tl
y
im
p
r
o
v
e
th
e
s
h
a
r
p
n
ess
an
d
cl
a
r
it
y
o
f
h
er
b
im
a
g
es
.
T
h
es
e
en
h
an
ce
m
e
n
t
t
ec
h
n
iq
u
es,
a
p
p
li
ed
d
u
r
i
n
g
t
h
e
p
r
e
p
r
o
ce
s
s
in
g
p
h
as
e,
a
r
e
ess
e
n
t
ial
f
o
r
a
d
d
r
ess
i
n
g
is
s
u
es
r
ela
te
d
t
o
b
l
u
r
r
y
o
r
l
o
w
-
q
u
ali
ty
im
ag
es,
en
s
u
r
i
n
g
t
h
at
t
h
e
m
o
d
el
r
ec
ei
v
es
h
ig
h
e
r
-
q
u
ali
ty
v
is
u
al
d
ata
.
B
y
im
p
r
o
v
i
n
g
d
a
ta
i
n
p
u
t
q
u
a
lit
y
,
t
h
ese
m
et
h
o
d
s
en
h
a
n
ce
cl
ass
i
f
ic
ati
o
n
ac
cu
r
a
c
y
a
n
d
r
e
d
u
ce
e
r
r
o
r
s
ca
u
s
ed
b
y
in
a
d
eq
u
ate
i
m
a
g
e
c
o
n
d
iti
o
n
s
.
2
.
3
.
2
.
E
x
perim
ent
a
l mo
del t
r
a
ini
ng
a
nd
ev
a
lua
t
io
n
T
h
e
m
o
d
el
tr
ai
n
in
g
was c
o
n
d
u
cted
o
n
Go
o
g
le
C
o
lab
,
u
s
in
g
a
v
ailab
le
g
r
ap
h
ics p
r
o
ce
s
s
in
g
u
n
it
(
GPU
)
r
eso
u
r
ce
s
to
f
ac
ilit
ate
ef
f
icien
t
p
r
o
ce
s
s
in
g
.
Py
th
o
n
p
r
o
g
r
am
m
in
g
lan
g
u
ag
e
was
u
tili
ze
d
,
alo
n
g
s
id
e
k
e
y
lib
r
ar
ies
s
u
ch
as
T
en
s
o
r
Flo
w,
Nu
m
Py
,
Pan
d
as
,
Ma
tp
lo
tlib
,
an
d
Op
en
C
V.
T
h
ese
to
o
ls
allo
wed
f
o
r
p
r
ec
is
e
m
an
ip
u
latio
n
a
n
d
a
n
aly
s
is
o
f
th
e
im
ag
e
d
ata.
‒
T
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
w
i
t
h
E
f
f
i
c
i
e
n
t
N
e
t
:
t
h
e
E
f
f
i
c
i
e
n
t
N
e
t
C
N
N
w
a
s
p
r
e
-
t
r
a
i
n
e
d
w
i
t
h
I
m
a
g
e
N
e
t
w
e
i
g
h
t
s
t
o
l
e
v
e
r
a
g
e
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
,
w
h
i
c
h
a
c
c
e
l
e
r
a
t
e
s
t
r
a
i
n
i
n
g
b
y
a
d
a
p
t
i
n
g
l
e
a
r
n
e
d
f
e
a
t
u
r
e
s
f
r
o
m
a
l
a
r
g
e
,
g
e
n
e
r
a
l
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
E
n
h
a
n
ce
d
cla
s
s
ifica
tio
n
o
f a
r
o
ma
tic
h
erb
s
u
s
in
g
E
fficien
tN
et
a
n
d
tr
a
n
s
fer
…
(
S
a
mir
a
N
a
s
cime
n
to
A
n
tu
n
es
)
4127
d
a
t
a
s
e
t
t
o
s
p
e
c
i
f
i
c
h
e
r
b
c
l
a
s
s
i
f
i
c
a
t
i
o
n
t
a
s
k
.
T
h
e
m
o
d
e
l
w
a
s
f
i
n
e
-
t
u
n
e
d
f
o
r
5
0
e
p
o
c
h
s
,
a
n
d
v
a
r
i
a
n
t
s
B
0
t
h
r
o
u
g
h
B
7
w
e
r
e
t
e
s
t
e
d
w
i
t
h
r
e
s
o
l
u
t
i
o
n
s
r
a
n
g
i
n
g
f
r
o
m
2
2
4
t
o
6
0
0
p
i
x
e
l
s
.
E
a
c
h
v
a
r
i
a
n
t
’
s
c
o
n
f
i
g
u
r
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
d
e
p
t
h
,
w
i
d
t
h
,
a
n
d
r
e
s
o
l
u
t
i
o
n
s
c
a
l
i
n
g
,
w
a
s
a
d
j
u
s
t
e
d
t
o
a
c
h
i
e
v
e
o
p
t
i
m
a
l
a
c
c
u
r
a
c
y
a
n
d
c
o
m
p
u
t
a
t
i
o
n
a
l
e
f
f
i
c
i
e
n
c
y
.
‒
C
u
s
to
m
m
o
d
el
tr
ain
in
g
:
i
n
ad
d
itio
n
to
tr
an
s
f
e
r
lear
n
in
g
,
a
c
u
s
to
m
m
o
d
el
was
tr
ain
ed
f
r
o
m
s
cr
atch
u
s
in
g
o
n
ly
th
e
d
ataset
o
f
r
o
s
em
a
r
y
,
m
in
t,
an
d
b
ay
leaf
im
ag
es.
T
h
is
ap
p
r
o
ac
h
aim
e
d
to
a
d
d
r
ess
th
e
ch
allen
g
e
s
en
co
u
n
ter
e
d
in
tr
an
s
f
er
lear
n
in
g
d
u
e
to
d
ataset
lim
itatio
n
s
an
d
v
er
if
y
th
e
m
o
d
el’
s
ab
ilit
y
to
lear
n
ex
clu
s
iv
ely
f
r
o
m
th
is
s
p
ec
if
i
c
d
ataset.
Fo
r
co
m
p
ar
a
b
ilit
y
,
th
e
cu
s
to
m
m
o
d
el
was
also
tr
ain
ed
f
o
r
5
0
ep
o
ch
s
,
an
d
v
ar
ia
n
ts
B
0
th
r
o
u
g
h
B
4
wer
e
test
ed
.
‒
Me
tr
ics
f
o
r
ev
alu
atio
n
:
th
e
p
r
im
ar
y
ev
al
u
atio
n
m
etr
ic
w
as
ac
cu
r
ac
y
,
d
ef
in
e
d
as
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tly
class
if
ied
im
ag
es
o
u
t
o
f
th
e
to
tal
class
if
icatio
n
s
.
L
o
s
s
was
a
ls
o
m
o
n
ito
r
ed
to
ass
es
s
m
o
d
el
co
n
v
er
g
en
ce
a
n
d
d
etec
t
p
o
ten
tial
o
v
er
f
itti
n
g
.
T
h
ese
m
etr
ics
wer
e
r
ec
o
r
d
e
d
th
r
o
u
g
h
o
u
t
tr
ain
in
g
a
n
d
o
n
test
s
et
s
to
p
r
o
v
id
e
a
co
m
p
r
eh
en
s
iv
e
v
iew
o
f
m
o
d
el
p
e
r
f
o
r
m
an
ce
.
T
h
e
m
o
d
el’
s
h
y
p
er
p
a
r
am
eter
s
,
s
u
ch
as
lear
n
in
g
r
ate,
b
atch
s
ize,
an
d
n
u
m
b
e
r
o
f
ep
o
ch
s
,
wer
e
k
ep
t
co
n
s
is
ten
t a
cr
o
s
s
all
tr
ain
in
g
s
ess
io
n
s
to
en
s
u
r
e
co
m
p
ar
ab
ilit
y
.
T
h
e
im
ag
e
d
ata
wer
e
s
p
lit in
to
tr
ain
in
g
a
n
d
test
s
ets
u
s
in
g
an
au
to
m
ated
alg
o
r
i
th
m
,
m
ain
tain
in
g
b
alan
ce
d
r
ep
r
esen
tatio
n
f
o
r
ea
c
h
h
e
r
b
ca
te
g
o
r
y
.
T
h
is
s
etu
p
is
in
ten
d
ed
t
o
en
a
b
le
o
t
h
er
r
esear
ch
er
s
to
r
ep
licate
th
e
ex
p
e
r
im
en
ts
an
d
o
b
s
er
v
e
s
im
ilar
r
es
u
lts
u
n
d
er
co
n
tr
o
lled
co
n
d
itio
n
s
.
T
h
is
s
tr
u
ctu
r
ed
m
eth
o
d
o
lo
g
y
allo
ws
f
o
r
r
ep
lica
tio
n
b
y
d
etailin
g
ea
ch
ex
p
er
i
m
en
tal
s
tep
,
f
r
o
m
d
ataset
p
r
ep
ar
atio
n
to
m
o
d
el
ev
alu
atio
n
,
en
a
b
lin
g
f
u
tu
r
e
s
tu
d
ies
to
b
u
ild
o
n
t
h
ese
f
in
d
in
g
s
an
d
r
ef
in
e
h
e
r
b
class
if
icatio
n
tech
n
iq
u
es u
s
in
g
C
NNs a
n
d
tr
an
s
f
er
lear
n
in
g
.
2
.
4
.
Co
ns
o
lid
a
t
io
n o
f
AI t
ec
hn
iq
ues
a
pp
lie
d t
o
re
s
ea
rc
h
T
h
e
ap
p
licatio
n
o
f
ad
v
an
ce
d
AI
tech
n
iq
u
es
h
as
b
ec
o
m
e
a
r
o
b
u
s
t
an
d
ef
f
icien
t
ap
p
r
o
ac
h
f
o
r
co
m
p
lex
im
ag
e
class
if
icatio
n
task
s
,
esp
ec
ially
in
ar
ea
s
wh
er
e
s
u
b
tle
v
is
u
al
d
if
f
er
en
ce
s
m
ak
e
r
ec
o
g
n
itio
n
ch
allen
g
in
g
.
I
n
th
is
s
tu
d
y
,
we
u
s
ed
th
e
E
f
f
icien
tNet
ar
ch
itectu
r
e
co
m
b
i
n
ed
with
tr
an
s
f
er
lear
n
in
g
a
n
d
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es
to
class
if
y
ar
o
m
a
tic
h
er
b
s
,
m
ee
tin
g
th
e
d
em
a
n
d
f
o
r
a
n
au
to
m
ate
d
an
d
p
r
ec
is
e
s
o
lu
tio
n
.
T
h
e
E
f
f
icien
tNet
ar
ch
itectu
r
e
r
ep
r
e
s
en
ts
an
in
n
o
v
atio
n
in
th
e
d
esi
g
n
o
f
C
NNs
b
y
i
n
tr
o
d
u
cin
g
a
co
m
p
o
u
n
d
s
ca
lin
g
m
eth
o
d
th
at
au
to
m
atica
lly
a
d
ju
s
ts
th
e
n
etwo
r
k
'
s
d
ep
th
,
wid
th
,
an
d
r
eso
lu
tio
n
b
ased
o
n
th
e
av
ailab
le
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
an
d
task
co
m
p
lex
ity
[
8
]
.
T
h
i
s
ap
p
r
o
ac
h
is
p
ar
ticu
lar
l
y
v
alu
ab
le
f
o
r
p
lan
t
class
if
icatio
n
p
r
o
b
lem
s
,
wh
er
e
th
e
m
o
d
el
n
ee
d
s
to
ca
p
tu
r
e
s
u
b
tle
v
ar
iatio
n
s
i
n
tex
tu
r
e
an
d
co
l
o
r
b
etwe
en
d
if
f
er
en
t
s
p
ec
ies.
R
ec
en
t
s
tu
d
ies
h
av
e
h
ig
h
lig
h
ted
th
at
c
o
m
p
o
u
n
d
s
ca
lin
g
allo
ws
f
o
r
h
ig
h
ac
cu
r
ac
y
with
o
p
tim
ized
r
eso
u
r
ce
u
s
e,
wh
ich
is
ess
en
tial in
lar
g
e
-
s
ca
le
im
ag
e
p
r
o
ce
s
s
in
g
e
n
v
ir
o
n
m
en
ts
[
2
5
]
,
[
2
6
]
.
T
r
an
s
f
e
r
le
ar
n
i
n
g
was
a
p
p
lie
d
to
l
ev
er
a
g
e
p
r
i
o
r
k
n
o
wle
d
g
e
f
r
o
m
m
o
d
els
t
r
ai
n
ed
o
n
la
r
g
e
d
at
asets
,
s
u
c
h
as
I
m
a
g
eNe
t
,
an
d
a
d
a
p
t
th
e
m
f
o
r
s
p
ec
i
f
i
c
h
e
r
b
cl
ass
i
f
i
ca
ti
o
n
t
ask
s
.
T
r
a
n
s
f
er
le
ar
n
i
n
g
is
a
t
ec
h
n
i
q
u
e
i
n
wh
i
ch
a
n
al
g
o
r
it
h
m
l
ea
r
n
s
i
m
a
g
e
f
ea
t
u
r
es
wit
h
i
n
o
n
e
d
o
m
ai
n
an
d
ca
n
th
en
a
p
p
l
y
t
h
is
ac
q
u
i
r
e
d
k
n
o
wle
d
g
e
to
a
n
ew
d
o
m
ai
n
,
e
v
e
n
wit
h
a
s
m
alle
r
d
a
tas
et
[
2
7
]
.
T
h
e
te
ch
n
i
q
u
e
o
f
t
r
a
n
s
f
e
r
le
a
r
n
in
g
w
it
h
f
i
n
e
-
t
u
n
i
n
g
e
n
a
b
l
es
s
tar
ti
n
g
wi
th
a
m
o
d
el
p
r
e
-
tr
ai
n
ed
f
o
r
a
s
p
e
ci
f
ic
tas
k
a
n
d
t
h
e
n
ad
j
u
s
t
in
g
o
n
l
y
ce
r
t
ai
n
la
y
e
r
s
o
f
th
e
n
eu
r
a
l
n
e
tw
o
r
k
to
ad
a
p
t
it
f
o
r
a
s
im
i
la
r
y
et
d
is
t
in
ct
t
ar
g
et
t
ask
[
2
8
]
.
T
h
is
tec
h
n
i
q
u
e
is
wi
d
el
y
r
e
c
o
g
n
i
z
ed
f
o
r
ac
ce
le
r
at
in
g
tr
a
in
in
g
a
n
d
i
m
p
r
o
v
i
n
g
ac
cu
r
a
cy
,
esp
ec
ial
ly
w
h
e
n
a
p
p
li
ed
t
o
lim
i
te
d
o
r
h
i
g
h
l
y
v
is
u
all
y
s
im
il
ar
d
at
asets
[
2
9
]
.
Ad
d
itio
n
ally
,
to
im
p
r
o
v
e
g
en
er
aliza
tio
n
an
d
r
e
d
u
ce
o
v
er
f
itt
in
g
,
d
ata
au
g
m
en
tatio
n
tec
h
n
i
q
u
es
wer
e
ap
p
lied
,
in
clu
d
in
g
tr
an
s
f
o
r
m
at
io
n
s
s
u
ch
as
r
o
tatio
n
,
b
r
ig
h
tn
e
s
s
ad
ju
s
tm
en
t,
an
d
cr
o
p
p
in
g
.
Data
au
g
m
en
tatio
n
cr
ea
tes
ar
tific
ial
v
ar
iatio
n
s
o
f
th
e
tr
ain
in
g
im
ag
es,
s
im
u
latin
g
a
m
o
r
e
r
o
b
u
s
t
d
ataset
an
d
h
elp
in
g
th
e
m
o
d
el
h
an
d
le
v
ar
iatio
n
s
in
an
g
le
an
d
lig
h
tin
g
[
3
0
]
,
[
3
1
]
.
Stu
d
ies
h
av
e
s
h
o
wn
th
at
d
ata
au
g
m
e
n
t
atio
n
is
ef
f
ec
tiv
e
in
p
lan
t
class
if
icatio
n
d
o
m
ain
s
,
wh
er
e
th
e
m
o
d
el
n
ee
d
s
to
a
d
ap
t
to
s
u
b
tle
c
h
an
g
es
in
lea
f
ap
p
ea
r
an
ce
[
3
2
]
.
Fin
ally
,
u
s
in
g
ac
c
u
r
ac
y
an
d
lo
s
s
m
etr
ics
was
ess
en
tial
f
o
r
ev
alu
atin
g
a
n
d
a
d
ju
s
tin
g
th
e
m
o
d
el
th
r
o
u
g
h
o
u
t
th
e
tr
ain
in
g
p
r
o
ce
s
s
,
h
el
p
in
g
to
d
etec
t
is
s
u
es
lik
e
o
v
er
f
itt
in
g
o
r
u
n
d
er
f
itti
n
g
.
I
n
s
u
m
m
ar
y
,
co
m
b
i
n
in
g
E
f
f
icien
tNet,
tr
an
s
f
er
lear
n
i
n
g
,
an
d
d
ata
au
g
m
en
tatio
n
p
r
o
v
id
es
a
s
o
lid
f
r
am
ewo
r
k
f
o
r
co
m
p
lex
v
is
u
al
r
ec
o
g
n
itio
n
task
s
,
s
u
ch
as
th
e
class
if
icatio
n
o
f
ar
o
m
atic
h
er
b
s
,
d
em
o
n
s
tr
atin
g
th
e
p
o
ten
tial o
f
th
ese
tech
n
iq
u
es
to
en
h
an
ce
th
e
ac
cu
r
ac
y
an
d
a
p
p
licab
ilit
y
o
f
AI
m
o
d
els in
c
h
allen
g
in
g
c
o
n
tex
ts
.
2
.
5
.
Co
m
pa
ra
t
iv
e
ev
a
lua
t
io
n o
f
co
nv
o
lutio
na
l a
rc
hite
ct
u
re
s
f
o
r
a
ro
m
a
t
ic
herb
cla
s
s
if
ica
t
io
n
T
h
i
s
s
t
u
d
y
f
o
c
u
s
e
d
e
x
c
l
u
s
i
v
e
l
y
o
n
u
s
i
n
g
E
f
f
i
ci
e
n
t
N
et
f
o
r
t
h
e
c
l
a
s
s
i
f
ic
a
t
i
o
n
o
f
a
r
o
m
at
i
c
h
e
r
b
s
d
u
e
t
o
i
ts
c
o
m
p
o
u
n
d
s
c
a
l
i
n
g
a
b
i
li
t
y
,
w
h
ic
h
a
u
t
o
m
a
t
i
c
al
l
y
a
d
j
u
s
ts
t
h
e
m
o
d
e
l
’
s
d
e
p
t
h
,
wi
d
t
h
,
a
n
d
r
es
o
l
u
t
i
o
n
a
c
c
o
r
d
i
n
g
t
o
t
h
e
t
a
s
k
'
s
r
e
q
u
i
r
e
m
e
n
t
s
.
T
h
is
c
h
a
r
a
c
t
e
r
is
t
i
c
m
a
k
es
E
f
f
i
c
i
e
n
t
N
e
t
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
e
f
f
i
ci
e
n
t
a
n
d
e
s
p
e
c
i
al
l
y
u
s
e
f
u
l
i
n
r
e
s
o
u
r
c
e
-
l
i
m
it
e
d
e
n
v
i
r
o
n
m
e
n
t
s
.
H
o
w
e
v
e
r
,
a
c
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
is
w
i
t
h
o
t
h
e
r
m
o
d
e
r
n
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
s
c
o
u
l
d
p
r
o
v
i
d
e
a
m
o
r
e
c
o
m
p
r
e
h
e
n
s
i
v
e
e
v
a
l
u
a
t
i
o
n
o
f
t
h
e
m
o
d
e
l
’
s
e
f
f
e
c
t
i
v
e
n
e
s
s
a
n
d
i
d
e
n
t
i
f
y
a
l
t
e
r
n
a
tiv
e
s
t
h
a
t
m
a
y
b
r
i
n
g
d
i
s
t
i
n
ct
a
d
v
a
n
t
a
g
es
f
o
r
h
e
r
b
c
la
s
s
i
f
i
c
at
i
o
n
.
R
es
N
e
t
,
f
o
r
i
n
s
t
a
n
c
e
,
s
t
a
n
d
s
o
u
t
f
o
r
it
s
u
s
e
o
f
r
es
i
d
u
a
l
c
o
n
n
e
c
t
i
o
n
s
,
w
h
i
c
h
f
a
c
i
l
it
a
t
e
g
r
a
d
i
e
n
t
f
l
o
w
in
d
e
e
p
n
e
t
w
o
r
k
s
a
n
d
m
i
t
i
g
a
t
e
th
e
v
a
n
i
s
h
i
n
g
g
r
a
d
i
e
n
t
p
r
o
b
l
e
m
.
T
h
e
m
o
d
e
l
h
a
s
o
n
e
M
a
x
P
o
o
l
l
a
y
e
r
,
4
8
c
o
n
v
o
l
u
t
i
o
n
a
l
l
a
y
e
r
s
,
a
n
d
a
s
t
a
n
d
a
r
d
p
o
o
l
l
a
y
e
r
.
I
n
e
a
c
h
i
d
e
n
t
i
f
i
c
at
i
o
n
b
l
o
c
k
,
a
s
w
e
ll
a
s
i
n
e
a
c
h
c
o
n
v
o
l
u
t
i
o
n
b
l
o
c
k
,
t
h
e
r
e
a
r
e
t
h
r
e
e
c
o
n
v
o
l
u
t
i
o
n
l
a
y
e
r
s
[
3
3
]
.
C
o
m
p
a
r
i
n
g
R
e
s
N
e
t
wi
t
h
E
f
f
i
c
i
e
n
t
N
et
w
o
u
l
d
h
e
l
p
a
s
s
e
s
s
w
h
e
t
h
e
r
t
h
e
g
e
n
e
r
al
i
z
a
ti
o
n
c
a
p
a
c
i
t
y
o
f
r
es
i
d
u
a
l
c
o
n
n
e
c
t
i
o
n
s
b
e
n
e
f
i
ts
t
h
e
cl
a
s
s
i
f
i
ca
t
i
o
n
o
f
c
l
a
s
s
es
w
it
h
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
.
5
,
Octo
b
er
2
0
2
5
:
4
1
2
3
-
4
1
3
6
4128
h
i
g
h
v
i
s
u
a
l
s
i
m
i
l
a
r
i
t
y
,
s
u
c
h
as
a
r
o
m
a
t
i
c
h
e
r
b
s
.
De
n
s
eN
e
t
,
i
n
tu
r
n
,
a
d
o
p
t
s
a
d
e
n
s
e
l
y
c
o
n
n
e
c
t
ed
l
a
y
e
r
a
r
c
h
i
t
e
c
t
u
r
e,
p
r
o
m
o
t
i
n
g
f
e
a
t
u
r
e
r
e
u
s
e
a
n
d
i
m
p
r
o
v
i
n
g
i
n
f
o
r
m
a
t
i
o
n
p
r
o
p
a
g
a
t
i
o
n
t
h
r
o
u
g
h
o
u
t
t
h
e
n
e
tw
o
r
k
.
T
h
i
s
ca
n
b
e
p
a
r
t
i
c
u
l
a
r
l
y
a
d
v
a
n
t
a
g
e
o
u
s
i
n
d
a
t
a
s
et
s
w
h
e
r
e
f
i
n
e
v
is
u
a
l
d
et
a
i
ls
,
s
u
c
h
as
t
e
x
t
u
r
e
a
n
d
e
d
g
e
s
,
a
r
e
c
r
u
c
i
al
f
o
r
c
l
a
s
s
i
f
ic
a
t
i
o
n
.
T
h
e
a
b
i
l
it
y
t
o
r
e
u
s
e
i
n
t
e
r
m
e
d
i
at
e
i
n
f
o
r
m
a
t
i
o
n
a
l
l
o
w
s
De
n
s
eN
e
t
t
o
c
a
p
t
u
r
e
s
u
b
t
l
e
v
i
s
u
al
n
u
a
n
c
e
s
,
o
f
f
e
r
i
n
g
a
n
i
n
t
e
r
e
s
ti
n
g
a
l
t
e
r
n
at
iv
e
t
o
E
f
f
i
c
i
e
n
tN
e
t
i
n
te
r
m
s
o
f
d
i
f
f
e
r
e
n
t
i
a
ti
n
g
v
i
s
u
a
ll
y
s
i
m
i
l
a
r
s
p
e
c
i
e
s
[
3
4
]
.
Mo
b
ileNet,
a
lig
h
tweig
h
t
ar
c
h
itectu
r
e,
was
s
p
ec
if
ically
d
e
s
ig
n
ed
f
o
r
e
f
f
icien
cy
o
n
m
o
b
i
le
d
ev
ices,
m
ak
in
g
it
i
d
ea
l
f
o
r
ap
p
licatio
n
s
r
eq
u
ir
in
g
lo
w
r
eso
u
r
ce
co
n
s
u
m
p
tio
n
,
s
u
c
h
as
f
ield
-
b
ased
p
lan
t
class
if
icatio
n
.
I
n
p
r
ac
tical
u
s
e
s
itu
atio
n
s
,
s
u
ch
as
id
en
tify
in
g
h
er
b
s
in
r
e
m
o
te
lo
ca
tio
n
s
,
Mo
b
ileNet
co
u
ld
b
e
a
b
en
e
f
icial
ch
o
ice,
b
alan
cin
g
ac
cu
r
ac
y
a
n
d
co
m
p
u
tatio
n
al
ef
f
icien
cy
o
n
d
ev
ices
with
lo
wer
p
r
o
ce
s
s
in
g
p
o
wer
.
Stu
d
ies
in
d
icate
th
at
Mo
b
ileNet
p
er
f
o
r
m
s
well
o
n
m
o
b
ile
d
e
v
ices,
m
ain
tain
in
g
co
m
p
etitiv
e
ac
cu
r
a
cy
lev
els
co
m
p
ar
ed
to
m
o
r
e
co
m
p
le
x
n
etwo
r
k
s
[
3
5
]
.
Ar
ch
itectu
r
es
s
u
ch
as
I
n
c
ep
tio
n
,
wh
ic
h
co
m
b
in
es
co
n
v
o
lu
tio
n
s
o
f
v
ar
io
u
s
s
izes
to
ca
p
tu
r
e
a
wid
e
r
an
g
e
o
f
v
is
u
al
f
ea
tu
r
es,
a
n
d
Xce
p
tio
n
,
wh
ic
h
en
h
an
ce
s
th
is
a
p
p
r
o
ac
h
with
s
ep
a
r
ab
le
co
n
v
o
l
u
tio
n
s
f
o
r
g
r
ea
ter
e
f
f
i
cien
cy
,
c
o
u
ld
also
en
r
ic
h
th
is
an
aly
s
is
.
I
n
ce
p
tio
n
is
ef
f
e
ctiv
e
with
im
ag
es
ex
h
ib
itin
g
h
ig
h
p
atter
n
v
ar
iab
ilit
y
,
wh
ile
Xce
p
tio
n
co
m
b
in
e
s
ac
cu
r
ac
y
an
d
ef
f
icien
c
y
,
m
a
k
in
g
it
s
u
itab
le
f
o
r
co
n
tex
ts
wh
er
e
b
ala
n
cin
g
c
o
m
p
u
tatio
n
al
co
s
t w
ith
f
ea
tu
r
e
ex
tr
ac
tio
n
ca
p
ac
ity
is
ess
en
tial
[
3
6
]
.
3.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
3
.
1
.
Resul
t
s
T
a
b
l
e
2
p
r
e
s
e
n
t
s
t
h
e
te
s
t
s
e
t'
s
r
e
s
u
l
ts
r
e
g
a
r
d
i
n
g
t
h
e
m
o
d
e
l
t
r
a
in
i
n
g
t
h
r
o
u
g
h
t
h
e
t
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
p
r
o
c
e
s
s
,
c
o
n
s
i
d
e
r
i
n
g
t
h
e
C
N
N
E
f
f
ic
i
en
t
N
e
t
p
r
e
-
t
r
a
i
n
e
d
w
i
t
h
t
h
e
we
i
g
h
t
s
o
f
t
h
e
I
m
a
g
e
N
et
d
a
t
ase
t
.
T
a
b
l
e
3
(
s
e
e
i
n
A
p
p
e
n
d
i
x
)
p
r
e
s
e
n
t
s
t
h
e
e
v
o
l
u
ti
o
n
o
f
a
c
c
u
r
a
c
y
a
n
d
l
o
s
s
t
h
r
o
u
g
h
o
u
t
t
h
e
t
r
a
i
n
i
n
g
p
r
o
c
e
s
s
,
p
r
o
v
i
d
i
n
g
a
d
e
t
a
i
l
e
d
v
i
e
w
o
f
t
h
e
m
o
d
el
’
s
p
e
r
f
o
r
m
a
n
c
e
u
n
d
e
r
d
i
f
f
e
r
e
n
t
c
o
n
f
i
g
u
r
a
t
i
o
n
s
.
A
n
a
l
y
zi
n
g
t
h
e
s
e
d
a
t
a
i
s
e
s
s
e
n
t
i
a
l
t
o
u
n
d
e
r
s
t
a
n
d
h
o
w
v
a
r
i
a
t
i
o
n
s
i
n
n
e
t
w
o
r
k
a
r
ch
i
t
e
c
t
u
r
e
a
n
d
t
r
ai
n
i
n
g
p
a
r
a
m
e
te
r
s
i
m
p
a
ct
t
h
e
f
i
n
al
r
e
s
u
l
t
s
.
T
h
i
s
a
n
a
l
y
s
is
is
c
r
u
c
ial
f
o
r
o
p
t
i
m
i
z
i
n
g
t
h
e
m
ac
h
i
n
e
l
ea
r
n
i
n
g
p
r
o
c
e
s
s
,
e
n
s
u
r
i
n
g
t
h
a
t
th
e
m
o
d
e
l
i
s
n
o
t
o
n
l
y
e
f
f
i
c
ie
n
t
b
u
t
a
l
s
o
ca
p
a
b
l
e
o
f
a
c
h
i
e
v
i
n
g
t
h
e
h
i
g
h
e
s
t
p
o
s
s
i
b
le
a
c
c
u
r
a
c
y
.
D
u
e
t
o
t
h
e
m
o
d
e
l
s
'
a
c
c
u
r
a
c
y
b
e
i
n
g
c
l
o
s
e
t
o
0
.
5
,
t
h
e
p
r
o
n
o
u
n
c
e
d
f
l
u
c
t
u
a
t
i
o
n
s
o
f
a
c
c
u
r
ac
y
t
h
r
o
u
g
h
o
u
t
t
h
e
t
r
a
i
n
i
n
g
,
t
h
e
a
v
e
r
a
g
e
a
c
c
u
r
a
c
y
b
e
tw
e
e
n
0
.
6
a
n
d
0
.
7
c
o
n
s
i
d
e
r
i
n
g
t
h
e
te
s
t
s
e
ts
,
a
n
d
t
h
e
c
o
n
f
u
s
i
o
n
o
f
r
e
s
u
l
t
s
w
h
e
n
t
es
t
i
n
g
wi
t
h
d
i
f
f
e
r
en
t
i
m
a
g
e
s
,
i
t
w
a
s
d
e
c
i
d
e
d
t
o
tr
a
i
n
t
h
e
E
f
f
i
ci
e
n
t
N
et
C
N
N
wi
t
h
o
u
t
w
e
i
g
h
ts
.
T
r
a
i
n
i
n
g
a
m
o
d
e
l
c
o
n
s
i
d
e
r
e
d
c
u
s
t
o
m
i
z
e
d
,
m
e
a
n
i
n
g
t
h
e
m
o
d
e
l
l
e
a
r
n
s
o
n
l
y
f
r
o
m
t
h
e
i
n
p
u
t
d
a
t
a
s
et
a
n
d
w
i
t
h
n
u
l
l
i
n
p
u
t
w
e
ig
h
t
,
a
i
m
e
d
t
o
i
d
e
n
t
i
f
y
t
h
e
p
r
o
b
a
b
l
e
c
a
u
s
es
f
o
r
t
h
e
a
c
c
u
r
a
c
y
v
a
l
u
e
s
u
n
d
e
r
0
.
8
a
n
d
t
h
e
d
i
f
f
e
r
e
n
t
c
o
n
f
u
s
i
o
n
s
d
u
r
i
n
g
t
e
s
ti
n
g
.
T
a
b
l
e
4
p
r
e
s
e
n
t
s
t
h
e
t
e
s
t
s
et'
s
r
e
s
u
lt
s
c
o
n
c
e
r
n
i
n
g
t
h
e
m
o
d
e
l'
s
t
r
a
i
n
i
n
g
w
i
t
h
o
u
t
t
h
e
t
r
a
n
s
f
e
r
l
ea
r
n
i
n
g
p
r
o
c
e
s
s
,
c
o
n
s
i
d
e
r
i
n
g
t
h
e
E
f
f
i
c
i
e
n
tN
e
t
C
NN
wi
t
h
o
u
t
i
n
p
u
t
w
e
i
g
h
ts
a
n
d
wi
t
h
o
n
l
y
t
h
e
s
e
l
e
ct
e
d
a
r
o
m
a
t
i
c
h
e
r
b
s
d
a
t
as
et
.
N
o
t
a
b
l
y
,
d
u
e
t
o
c
o
m
p
u
t
a
t
i
o
n
a
l
c
a
p
a
c
i
t
y
,
t
h
e
m
o
d
e
l
s
w
e
r
e
t
r
a
i
n
e
d
t
o
c
o
n
s
i
d
e
r
r
e
s
o
l
u
t
i
o
n
s
B
0
t
h
r
o
u
g
h
B
4
.
T
ab
l
e
5
s
h
o
w
s
t
h
e
e
v
o
l
u
t
i
o
n
o
f
a
cc
u
r
a
c
y
a
n
d
l
o
s
s
t
h
r
o
u
g
h
o
u
t
t
h
e
t
r
a
i
n
i
n
g
s
es
s
i
o
n
s
.
R
eg
ar
d
in
g
th
e
tr
ain
in
g
tim
e,
it
is
n
o
ted
th
at
it in
cr
ea
s
ed
ac
co
r
d
in
g
to
th
e
r
eso
lu
tio
n
.
Ad
d
itio
n
ally
,
th
e
tim
e
is
co
n
s
id
er
ab
l
y
lo
n
g
er
f
o
r
a
m
o
d
el
lear
n
i
n
g
with
o
u
t
i
n
p
u
t
weig
h
ts
.
T
h
is
d
y
n
am
ic
i
s
ex
p
ec
ted
b
ec
au
s
e,
ac
co
r
d
in
g
to
T
a
n
an
d
L
e
[
8
]
,
th
e
im
ag
e
lev
els
in
E
f
f
icien
tNet
r
ef
er
to
d
if
f
er
en
t
v
ar
ian
ts
o
f
th
e
E
f
f
icien
tNet
ar
ch
itectu
r
e,
an
d
th
e
h
ig
h
er
t
h
e
lev
el
n
u
m
b
er
,
th
e
g
r
ea
ter
th
e
d
ep
th
,
wid
th
,
a
n
d
r
eso
l
u
tio
n
o
f
t
h
e
im
ag
e,
r
esu
ltin
g
in
a
m
o
r
e
c
o
m
p
lex
an
d
p
o
wer
f
u
l
n
etwo
r
k
ca
p
ab
l
e
o
f
h
an
d
lin
g
m
o
r
e
d
em
a
n
d
i
n
g
c
o
m
p
u
ter
v
is
io
n
task
s
,
h
en
ce
th
e
l
o
n
g
er
tr
ain
i
n
g
tim
e.
W
h
en
an
aly
zin
g
th
e
lo
s
s
,
it
is
n
o
ticea
b
le
th
at
b
o
th
in
t
h
e
tr
an
s
f
er
lear
n
in
g
p
r
o
ce
s
s
an
d
i
n
th
e
c
u
s
to
m
p
r
o
ce
s
s
,
it
in
cr
ea
s
es
as
th
e
r
eso
lu
tio
n
in
cr
ea
s
es.
T
h
is
i
s
also
ex
p
ec
ted
d
u
e
to
th
e
b
etter
r
eso
lu
tio
n
o
f
th
e
i
m
ag
e
p
ix
els,
wh
ich
e
n
ab
les o
b
jects to
b
e
id
en
tifie
d
m
o
r
e
ac
c
u
r
ately
.
T
ab
le
2
.
Su
m
m
a
r
y
o
f
E
f
f
icien
tNet
m
o
d
el
r
esu
lts
u
s
in
g
tr
an
s
f
er
lear
n
in
g
M
o
d
e
l
v
a
r
i
a
n
t
R
e
s
o
l
u
t
i
o
n
Ep
o
c
h
s
Tr
a
i
n
i
n
g
t
i
me
Lo
ss
A
c
c
u
r
a
c
y
B0
2
2
4
50
1
2
m
3
0
s
0
.
7
9
0
.
7
1
B1
2
4
0
50
1
8
m
3
3
s
0
.
7
7
0
.
5
1
B2
2
6
0
50
2
4
m
3
2
s
0
.
8
0
0
.
7
7
B3
3
0
0
50
4
2
m
3
5
s
0
.
7
2
0
.
7
7
B4
3
8
0
50
0
1
h
3
9
m
0
.
7
3
0
.
8
0
B5
4
5
6
50
0
3
h
3
2
m
0
.
8
4
0
.
5
7
B6
5
2
8
50
0
6
h
0
9
m
0
.
6
9
0
.
7
4
B7
6
0
0
50
0
9
h
4
4
m
0
.
7
7
0
.
6
5
T
ab
le
4
.
C
u
s
to
m
m
o
d
el
tr
ain
in
g
r
esu
lts
(
with
o
u
t tr
an
s
f
er
lea
r
n
in
g
)
M
o
d
e
l
v
a
r
i
a
n
t
R
e
s
o
l
u
t
i
o
n
Ep
o
c
h
s
Ti
me
Lo
ss
A
c
c
u
r
a
c
y
B0
2
2
4
50
51
m
42
s
0
.
9
4
0
.
8
8
B1
2
4
0
50
01
h
11
m
1
.
0
9
0
.
8
0
B2
2
6
0
50
01
h
37
m
2
5
4
2
0
.
4
0
B3
3
0
0
50
02
h
50
m
0
.
9
6
0
.
8
0
B4
3
8
0
50
06
h
36
m
1
.
5
3
0
.
6
8
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
E
n
h
a
n
ce
d
cla
s
s
ifica
tio
n
o
f a
r
o
ma
tic
h
erb
s
u
s
in
g
E
fficien
tN
et
a
n
d
tr
a
n
s
fer
…
(
S
a
mir
a
N
a
s
cime
n
to
A
n
tu
n
es
)
4129
T
ab
le
5
.
E
v
o
lu
tio
n
o
f
tr
ai
n
in
g
co
n
s
id
er
in
g
t
h
e
p
r
o
ce
s
s
o
f
tr
an
s
f
er
lear
n
in
g
R
e
s
o
l
u
t
i
o
n
Lo
ss
A
c
c
u
r
a
c
y
B0
B1
B2
B3
B4
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
.
5
,
Octo
b
er
2
0
2
5
:
4
1
2
3
-
4
1
3
6
4130
T
h
e
ac
cu
r
ac
y
v
ar
ie
d
ac
co
r
d
i
n
g
to
th
e
r
eso
lu
tio
n
,
m
ea
n
in
g
th
at
th
e
h
ig
h
er
th
e
r
eso
lu
tio
n
,
th
e
h
ig
h
er
th
e
ac
cu
r
ac
y
.
Ad
d
itio
n
ally
,
th
e
v
ar
iatio
n
in
ac
cu
r
ac
y
in
th
e
tr
an
s
f
er
lear
n
in
g
p
r
o
ce
s
s
h
a
d
m
o
r
e
s
ig
n
if
ican
t
o
s
cillatio
n
th
an
th
at
in
th
e
c
u
s
to
m
m
o
d
el,
wh
ic
h
s
h
o
wed
a
g
r
o
win
g
lear
n
in
g
tr
en
d
o
v
er
th
e
ep
o
c
h
s
.
T
h
is
s
u
g
g
ests
th
at
th
e
cu
s
to
m
m
o
d
el
was
ab
le
to
lear
n
m
o
r
e
e
f
f
ec
tiv
ely
f
r
o
m
t
h
e
a
v
ailab
le
d
a
taset,
r
esu
ltin
g
in
a
m
o
r
e
s
tab
le
p
er
f
o
r
m
a
n
ce
o
v
e
r
tim
e.
3.
2
.
Dis
cus
s
io
n
T
h
is
s
tu
d
y
ad
d
r
ess
es
s
ig
n
if
ican
t
g
ap
s
o
b
s
er
v
ed
in
p
r
ev
i
o
u
s
r
esear
ch
o
n
th
e
class
if
icatio
n
o
f
ar
o
m
ati
c
h
er
b
s
u
s
in
g
co
m
p
u
ter
v
is
io
n
tech
n
iq
u
es.
W
h
ile
p
r
io
r
s
tu
d
ies
h
av
e
ex
p
lo
r
ed
th
e
u
s
e
o
f
C
NNs
an
d
tr
an
s
f
er
lear
n
in
g
f
o
r
p
lan
t
id
en
tific
ati
o
n
,
t
h
ey
p
r
im
ar
ily
f
o
cu
s
ed
o
n
d
is
tin
ct
p
la
n
t
s
p
ec
ies
with
v
is
u
ally
a
p
p
ar
e
n
t
d
if
f
er
en
ce
s
.
T
h
ese
s
tu
d
ies
o
f
ten
em
p
h
asized
g
en
er
al
p
lan
t
class
if
icatio
n
r
ath
er
t
h
an
th
e
n
u
an
ce
d
d
if
f
er
en
tiatio
n
r
eq
u
ir
ed
f
o
r
v
is
u
ally
s
im
ilar
ar
o
m
atic
h
er
b
s
,
s
u
ch
as
r
o
s
em
ar
y
,
m
i
n
t,
an
d
b
a
y
leaf
.
C
o
n
s
eq
u
en
tly
,
th
ey
d
id
n
o
t
e
x
p
licitly
ad
d
r
ess
th
e
in
f
lu
e
n
c
e
o
f
s
u
b
tle
v
ar
iatio
n
s
in
co
lo
r
,
s
h
ap
e,
an
d
tex
tu
r
e
th
at
ar
e
cr
itical
f
o
r
ac
c
u
r
ately
class
if
y
in
g
th
ese
h
er
b
s
.
T
h
is
r
esear
ch
s
p
ec
if
ically
in
v
e
s
tig
ates
th
e
ap
p
licatio
n
o
f
t
h
e
E
f
f
icien
tNet
C
NN
ar
ch
itectu
r
e
to
tack
le
th
ese
ch
allen
g
es,
e
x
am
in
in
g
h
o
w
tr
an
s
f
e
r
lear
n
i
n
g
ca
n
b
e
f
in
e
-
tu
n
e
d
to
im
p
r
o
v
e
class
if
icatio
n
ac
cu
r
ac
y
f
o
r
v
is
u
ally
s
im
ilar
p
lan
t
s
p
ec
ies,
th
er
eb
y
f
illi
n
g
a
k
ey
g
a
p
in
th
e
ex
is
tin
g
liter
atu
r
e.
T
h
e
u
s
e
o
f
tr
an
s
f
e
r
lear
n
i
n
g
allo
wed
th
e
E
f
f
icien
tNet
m
o
d
el,
p
r
e
-
tr
ai
n
ed
o
n
th
e
I
m
ag
eN
et
d
ataset,
to
b
e
s
u
cc
ess
f
u
lly
ad
ap
ted
f
o
r
s
p
ec
if
ic
ar
o
m
atic
h
er
b
class
if
icatio
n
.
T
h
is
tech
n
iq
u
e
p
r
o
v
ed
ef
f
icien
t
f
o
r
ac
ce
ler
atin
g
tr
ain
in
g
,
r
ed
u
cin
g
co
m
p
u
tatio
n
al
co
s
ts
,
an
d
m
ain
tain
in
g
ac
cu
r
ac
y
.
No
tab
ly
,
lo
wer
-
r
eso
lu
tio
n
p
r
e
-
tr
ain
ed
m
o
d
els,
s
u
ch
as
E
f
f
ici
en
tNet
B
0
,
s
h
o
wed
s
u
b
s
tan
tially
s
h
o
r
ter
tr
ain
in
g
tim
es
with
o
u
t
s
ig
n
if
ican
tly
co
m
p
r
o
m
is
in
g
ac
cu
r
ac
y
.
Hig
h
e
r
-
r
eso
lu
tio
n
m
o
d
els,
in
clu
d
in
g
E
f
f
icien
tNet
B
3
an
d
B
4
,
d
e
m
o
n
s
tr
ated
in
cr
ea
s
ed
ac
c
u
r
a
cy
,
ac
h
iev
in
g
v
alu
es
o
v
er
0
.
8
in
ce
r
tain
ca
s
es.
Ho
wev
er
,
th
e
co
m
p
u
tatio
n
al
co
s
t
as
s
o
ciate
d
with
h
ig
h
er
r
eso
lu
tio
n
s
,
f
r
o
m
E
f
f
icien
t
Net
B
5
to
B
7
,
was
co
n
s
id
er
ab
ly
h
ig
h
e
r
,
with
tr
ain
in
g
tim
es
u
p
to
f
iv
e
tim
es
lo
n
g
er
,
wh
ich
lim
its
th
eir
p
r
ac
tical
f
ea
s
ib
ilit
y
f
o
r
r
eso
u
r
ce
-
co
n
s
tr
ain
ed
d
ev
ices.
T
h
is
o
b
s
er
v
atio
n
u
n
d
er
s
co
r
es
th
e
im
p
o
r
tan
ce
o
f
b
alan
ci
n
g
r
eso
lu
tio
n
with
co
m
p
u
tatio
n
al
e
f
f
icien
cy
,
esp
e
cially
in
f
ield
ap
p
licatio
n
s
wh
er
e
r
eso
u
r
ce
s
m
ay
b
e
lim
ited
.
Alth
o
u
g
h
cu
s
to
m
-
tr
ai
n
ed
m
o
d
els
(
tr
ain
e
d
f
r
o
m
s
cr
atch
)
ac
h
iev
ed
co
m
p
etitiv
e
ac
c
u
r
ac
y
,
t
h
ey
r
eq
u
ir
ed
s
ig
n
if
ica
n
tly
lo
n
g
e
r
tr
ain
in
g
tim
es
an
d
d
is
p
lay
ed
g
r
ea
ter
s
tab
ilit
y
in
ac
cu
r
ac
y
o
v
er
th
e
co
u
r
s
e
o
f
tr
ain
in
g
.
T
r
a
n
s
f
er
lear
n
in
g
m
o
d
els,
in
co
n
tr
ast,
ex
h
i
b
ited
s
o
m
e
f
lu
ctu
atio
n
s
in
ac
cu
r
ac
y
ac
r
o
s
s
d
if
f
er
en
t
r
eso
lu
tio
n
s
,
s
u
g
g
esti
n
g
th
at
th
e
p
r
e
-
tr
ain
ed
m
o
d
el
m
ig
h
t b
e
s
en
s
itiv
e
to
th
e
s
p
ec
if
ic
ch
ar
ac
ter
is
tics
o
f
th
e
h
er
b
d
ataset.
T
h
e
v
ar
ia
b
ilit
y
o
b
s
e
r
v
ed
in
ac
cu
r
ac
y
an
d
lo
s
s
o
v
er
th
e
e
p
o
ch
’
s
p
o
in
ts
to
p
o
ten
tial
o
v
er
f
itti
n
g
,
p
ar
ticu
lar
ly
with
h
ig
h
er
r
eso
lu
tio
n
s
,
h
ig
h
lig
h
tin
g
th
e
im
p
o
r
t
an
ce
o
f
em
p
lo
y
in
g
d
ata
au
g
m
en
tatio
n
tech
n
i
q
u
es
to
im
p
r
o
v
e
m
o
d
el
r
o
b
u
s
tn
ess
an
d
en
a
b
le
b
etter
g
e
n
er
aliza
tio
n
to
n
ew
im
a
g
es o
f
h
e
r
b
s
.
T
h
e
s
tu
d
y
id
en
tifie
d
th
at
m
o
d
el
ac
cu
r
ac
y
co
u
ld
b
e
im
p
ac
ted
b
y
th
e
p
r
esen
ce
o
f
im
ag
es
o
f
h
e
r
b
b
u
n
ch
es
m
ix
ed
with
im
a
g
es
o
f
in
d
iv
id
u
al
leav
es
in
th
e
d
ataset.
T
h
u
s
,
a
k
ey
r
ec
o
m
m
e
n
d
atio
n
f
o
r
f
u
tu
r
e
r
esear
ch
is
to
r
ef
i
n
e
th
e
d
ataset
to
in
clu
d
e
o
n
ly
is
o
lated
lea
f
im
ag
es,
wh
ich
co
u
ld
h
elp
r
e
d
u
ce
class
if
icatio
n
co
n
f
u
s
io
n
.
Ad
d
itio
n
ally
,
f
u
r
t
h
er
ex
p
lo
r
atio
n
o
f
d
if
f
er
en
t
r
eso
lu
tio
n
co
n
f
ig
u
r
atio
n
s
an
d
im
p
lem
en
tin
g
f
in
e
-
g
r
ain
ed
v
is
u
al
class
if
icatio
n
(
FGVC
)
tech
n
iq
u
es
ar
e
r
e
co
m
m
en
d
e
d
to
im
p
r
o
v
e
t
h
e
m
o
d
el’
s
ab
ilit
y
to
d
if
f
er
en
tiate
s
u
b
tle
v
a
r
iatio
n
s
b
etwe
en
h
er
b
s
p
ec
ies.
I
n
co
m
p
ar
in
g
tr
an
s
f
er
lea
r
n
in
g
-
b
ased
m
o
d
els
with
c
u
s
to
m
-
tr
ain
ed
m
o
d
els,
ea
ch
ap
p
r
o
ac
h
p
r
esen
ts
d
is
tin
ct
ad
v
an
tag
es
.
T
r
an
s
f
er
lear
n
in
g
s
ig
n
if
ica
n
tly
r
ed
u
ce
s
tr
ain
in
g
tim
e
an
d
e
n
ab
les
r
ap
id
r
esu
lts
,
wh
er
ea
s
cu
s
to
m
m
o
d
els
o
f
f
e
r
g
r
ea
ter
s
tab
ilit
y
in
ac
cu
r
ac
y
th
r
o
u
g
h
o
u
t
tr
ain
i
n
g
,
esp
ec
i
ally
at
in
ter
m
ed
iate
r
eso
lu
tio
n
s
.
Alth
o
u
g
h
m
o
r
e
tim
e
-
co
n
s
u
m
in
g
,
cu
s
to
m
m
o
d
els
d
em
o
n
s
tr
ated
h
ig
h
er
r
esil
ien
ce
to
o
v
er
f
itti
n
g
,
s
h
o
win
g
m
o
r
e
co
n
s
is
ten
t
p
er
f
o
r
m
a
n
ce
o
n
u
n
s
ee
n
d
ata
a
v
alu
ab
le
c
h
ar
ac
te
r
is
tic
in
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
.
I
n
t
er
m
s
o
f
ac
cu
r
ac
y
,
m
id
-
r
e
s
o
lu
t
io
n
m
o
d
e
l
s
,
p
ar
ti
cu
l
ar
ly
E
f
f
ic
ien
tN
et
B
2
to
B
4
,
s
e
e
m
to
o
f
f
er
an
o
p
tim
al
b
a
lan
ce
b
et
we
en
c
o
m
p
u
ta
t
io
n
a
l
ef
f
i
ci
en
cy
an
d
g
en
er
a
l
iza
t
io
n
c
ap
ab
i
li
ty
,
m
a
in
ta
in
in
g
co
m
p
e
ti
tiv
e
ac
cu
r
ac
y
wi
th
o
u
t
th
e
h
ig
h
co
m
p
u
t
at
io
n
al
c
o
s
t
s
a
s
s
o
c
ia
t
ed
w
ith
h
ig
h
er
-
r
e
s
o
lu
t
io
n
m
o
d
el
s
.
T
h
i
s
s
tu
d
y
p
r
o
v
id
e
s
a
b
a
lan
ce
d
p
er
s
p
e
c
tiv
e
o
n
th
e
s
t
r
en
g
th
s
an
d
li
m
it
at
io
n
s
o
f
ea
c
h
ap
p
r
o
ac
h
,
o
f
f
er
in
g
p
r
a
ct
ic
al
in
s
ig
h
t
s
f
o
r
th
e
d
e
v
el
o
p
m
en
t
o
f
ef
f
ic
ien
t
m
o
d
e
l
s
in
ar
o
m
at
ic
h
er
b
c
la
s
s
if
i
ca
tio
n
an
d
s
i
m
il
ar
c
la
s
s
if
ic
at
io
n
co
n
t
ex
t
s
.
T
ab
le
6
s
u
m
m
ar
ize
s
th
e
k
ey
f
i
n
d
in
g
s
an
d
o
b
s
e
r
v
a
tio
n
s
f
r
o
m
th
e
s
t
u
d
y
,
h
ig
h
li
g
h
t
in
g
m
o
d
el
ef
f
i
ci
en
cy
,
th
e
im
p
ac
t
o
f
r
eso
lu
tio
n
,
v
ar
iab
il
ity
in
p
er
f
o
r
m
an
c
e
m
e
tr
i
c
s
,
an
d
r
ec
o
m
m
en
d
at
io
n
s
f
o
r
f
u
tu
r
e
en
h
an
c
em
en
t
s
.
T
h
e
ef
f
ec
tiv
e
n
ess
o
f
th
e
ar
o
m
atic
h
er
b
class
if
icatio
n
m
o
d
el
d
ep
en
d
s
s
ig
n
if
ican
tly
o
n
th
e
s
elec
tio
n
an
d
o
p
tim
izatio
n
o
f
th
e
n
eu
r
a
l
n
etwo
r
k
s
em
p
lo
y
ed
.
T
h
e
tech
n
iq
u
es
d
is
cu
s
s
ed
in
[
3
7
]
,
ex
p
lo
r
e
f
ea
tu
r
e
f
u
s
io
n
f
r
o
m
m
u
ltip
le
n
e
u
r
al
n
etwo
r
k
ar
ch
itectu
r
es.
T
h
is
ap
p
r
o
a
ch
aim
s
to
co
m
b
in
e
th
e
s
tr
en
g
th
s
o
f
d
if
f
er
en
t
n
etwo
r
k
s
to
ca
p
tu
r
e
a
b
r
o
ad
er
r
an
g
e
o
f
d
is
cr
im
in
ativ
e
f
ea
tu
r
es.
Netwo
r
k
s
elec
tio
n
c
an
en
h
a
n
ce
m
o
d
el
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
E
n
h
a
n
ce
d
cla
s
s
ifica
tio
n
o
f a
r
o
ma
tic
h
erb
s
u
s
in
g
E
fficien
tN
et
a
n
d
tr
a
n
s
fer
…
(
S
a
mir
a
N
a
s
cime
n
to
A
n
tu
n
es
)
4131
ac
cu
r
ac
y
an
d
o
p
tim
ize
th
e
le
ar
n
in
g
p
r
o
ce
s
s
b
y
le
v
er
ag
in
g
in
f
o
r
m
atio
n
f
u
s
io
n
to
ad
d
r
ess
th
e
s
p
ec
if
ic
v
is
u
al
s
im
ilar
ities
o
f
ar
o
m
atic
h
er
b
s
,
en
s
u
r
in
g
m
o
r
e
r
eliab
le
class
if
icatio
n
r
esu
lts
.
T
ab
le
6
.
Key
f
in
d
in
g
s
an
d
o
b
s
er
v
atio
n
s
A
sp
e
c
t
F
i
n
d
i
n
g
s
M
o
d
e
l
e
f
f
i
c
i
e
n
c
y
Tr
a
n
sf
e
r
l
e
a
r
n
i
n
g
r
e
d
u
c
e
d
t
r
a
i
n
i
n
g
t
i
m
e
b
u
t
a
c
h
i
e
v
e
d
c
o
mp
a
r
a
b
l
e
a
c
c
u
r
a
c
y
t
o
c
u
s
t
o
m
m
o
d
e
l
s.
R
e
s
o
l
u
t
i
o
n
i
m
p
a
c
t
H
i
g
h
e
r
r
e
s
o
l
u
t
i
o
n
s (B
3
,
B
4
)
i
mp
r
o
v
e
d
a
c
c
u
r
a
c
y
,
b
u
t
i
n
c
r
e
a
se
d
c
o
m
p
u
t
a
t
i
o
n
t
i
m
e
si
g
n
i
f
i
c
a
n
t
l
y
.
Lo
ss
a
n
d
a
c
c
u
r
a
c
y
v
a
r
i
a
b
i
l
i
t
y
Tr
a
n
sf
e
r
l
e
a
r
n
i
n
g
mo
d
e
l
s s
h
o
w
e
d
g
r
e
a
t
e
r
v
a
r
i
a
b
i
l
i
t
y
i
n
a
c
c
u
r
a
c
y
a
c
r
o
ss res
o
l
u
t
i
o
n
s.
G
e
n
e
r
a
l
i
z
a
t
i
o
n
i
s
su
e
s
C
u
s
t
o
m m
o
d
e
l
g
e
n
e
r
a
l
i
z
e
d
w
e
l
l
w
i
t
h
l
o
w
e
r
a
c
c
u
r
a
c
y
o
s
c
i
l
l
a
t
i
o
n
b
u
t
r
e
q
u
i
r
e
d
e
x
t
e
n
d
e
d
t
r
a
i
n
i
n
g
.
F
u
t
u
r
e
e
n
h
a
n
c
e
me
n
t
s
S
u
g
g
e
s
t
e
d
r
e
f
i
n
i
n
g
t
h
e
d
a
t
a
b
a
s
e
w
i
t
h
i
so
l
a
t
e
d
l
e
a
f
i
ma
g
e
s
a
n
d
e
x
p
l
o
r
i
n
g
d
i
f
f
e
r
e
n
t
r
e
s
o
l
u
t
i
o
n
s
.
Acc
u
r
a
te
h
er
b
id
en
ti
f
i
ca
tio
n
f
ac
e
s
ch
al
len
g
e
s
d
u
e
to
th
e
h
ig
h
v
i
s
u
al
s
im
i
lar
ity
b
et
we
e
n
d
i
f
f
er
en
t
h
er
b
ty
p
e
s
.
S
tu
d
ie
s
h
av
e
ex
p
l
o
r
ed
co
m
p
l
em
e
n
t
ar
y
s
t
r
a
teg
ie
s
to
a
d
d
r
e
s
s
th
e
s
e
ch
al
len
g
e
s
,
in
c
lu
d
in
g
:
‒
S
SL,
th
i
s
te
ch
n
iq
u
e
r
e
d
u
ce
s
r
el
ian
ce
o
n
lab
el
ed
d
a
ta
s
e
t
s
b
y
al
lo
w
in
g
m
o
d
e
ls
to
lea
r
n
m
o
r
e
r
e
le
v
an
t
r
ep
r
e
s
e
n
t
at
io
n
s
o
f
h
er
b
s
f
r
o
m
u
n
la
b
e
led
d
at
a,
en
h
an
c
in
g
cla
s
s
if
i
ca
ti
o
n
b
y
e
m
p
h
a
s
iz
in
g
th
e
in
tr
in
s
ic
f
ea
tu
r
e
s
o
f
th
e
le
av
e
s
[
2
6
]
.
‒
Ne
two
r
k
s
e
le
ct
io
n
an
d
in
f
o
r
m
at
io
n
f
u
s
io
n
,
b
y
o
p
t
im
iz
in
g
th
e
ch
o
i
ce
o
f
n
eu
r
a
l
n
et
wo
r
k
ar
ch
i
te
ctu
r
e
s
an
d
f
u
s
in
g
in
f
o
r
m
at
io
n
,
th
i
s
ap
p
r
o
a
ch
im
p
r
o
v
e
s
c
la
s
s
i
f
i
ca
t
io
n
a
cc
u
r
ac
y
.
Acc
o
r
d
in
g
to
t
h
e
Z
af
ar
e
t
a
l
.
[
2
7
]
,
th
is
s
t
r
a
teg
y
c
an
b
e
ap
p
li
ed
to
d
e
ter
m
in
e
th
e
b
e
s
t
-
p
er
f
o
r
m
in
g
v
er
s
io
n
o
f
E
f
f
ic
ie
n
tN
et
f
o
r
h
e
r
b
id
en
tif
ic
at
io
n
.
‒
Feat
u
r
e
p
r
ese
r
v
at
i
o
n
in
i
m
a
g
es
,
a
d
v
a
n
c
ed
s
e
g
m
en
tat
io
n
tec
h
n
iq
u
es
h
e
lp
m
ai
n
ta
in
s
u
b
tl
e
d
et
a
ils
i
n
i
m
a
g
es
,
p
r
e
v
e
n
ti
n
g
th
e
lo
s
s
o
f
c
r
u
ci
al
i
n
f
o
r
m
ati
o
n
n
ec
ess
a
r
y
f
o
r
d
is
ti
n
g
u
is
h
i
n
g
v
is
u
all
y
s
im
ila
r
s
p
ec
ies
[
2
5
]
.
‒
I
m
ag
e
q
u
a
li
ty
en
h
a
n
ce
m
en
t,
m
eth
o
d
s
f
o
r
im
p
r
o
v
in
g
t
h
e
s
h
ar
p
n
e
s
s
an
d
cl
ar
i
ty
o
f
m
ed
i
ca
l
im
ag
e
s
ca
n
b
e
ad
ap
ted
to
en
s
u
r
e
h
ig
h
er
cl
a
s
s
if
ic
at
io
n
p
r
e
ci
s
io
n
in
h
e
r
b
an
al
y
s
i
s
b
y
p
r
o
v
id
in
g
h
i
g
h
er
-
q
u
al
it
y
v
is
u
al
d
ata
b
ef
o
r
e
au
to
m
a
ted
an
a
ly
s
i
s
[
2
8
]
.
T
o
g
et
h
er
,
th
e
s
e
t
ec
h
n
iq
u
es
f
o
r
m
a
r
o
b
u
s
t
f
r
a
m
e
wo
r
k
f
o
r
en
h
an
c
in
g
th
e
p
er
f
o
r
m
an
c
e
an
d
r
e
li
ab
il
ity
o
f
au
to
m
at
ed
h
er
b
cl
as
s
if
ica
ti
o
n
m
o
d
el
s
,
ad
d
r
e
s
s
in
g
b
o
th
d
a
ta
q
u
al
ity
an
d
m
o
d
e
l o
p
ti
m
i
za
tio
n
.
4.
CO
NCLU
SI
O
N
T
h
i
s
s
tu
d
y
s
u
cc
e
s
s
f
u
ll
y
d
e
m
o
n
s
tr
a
te
s
th
e
ap
p
l
ica
t
io
n
o
f
E
f
f
ic
ien
tN
et
w
i
th
t
r
an
s
f
er
l
ea
r
n
in
g
f
o
r
ar
o
m
at
ic
h
er
b
c
la
s
s
i
f
i
c
at
io
n
,
a
ch
i
ev
i
n
g
n
o
t
ab
l
e
a
c
cu
r
ac
y
wh
il
e
h
ig
h
l
ig
h
t
in
g
ar
ea
s
f
o
r
d
a
tab
a
s
e
r
ef
in
em
e
n
t
an
d
th
e
im
p
o
r
tan
c
e
o
f
b
al
an
c
in
g
r
e
s
o
lu
t
io
n
w
i
t
h
co
m
p
u
t
at
io
n
a
l
ef
f
ic
ie
n
cy
f
o
r
i
m
p
r
o
v
ed
m
o
d
e
l
g
en
er
al
iz
at
io
n
.
T
h
i
s
r
ev
i
ew
h
as
k
ey
lim
it
at
io
n
s
,
in
c
lu
d
in
g
r
el
ian
ce
o
n
a
s
p
e
cif
ic
im
ag
e
d
ata
s
et
o
f
ar
o
m
a
ti
c
h
er
b
s
,
t
e
s
t
in
g
i
n
a
co
n
tr
o
ll
ed
en
v
ir
o
n
m
en
t,
an
d
a
f
o
cu
s
m
ain
ly
o
n
a
cc
u
r
ac
y
an
d
co
m
p
u
ta
tio
n
al
ef
f
ic
ien
c
y
,
wi
th
li
m
i
ted
a
tt
en
t
io
n
to
in
ter
p
r
et
ab
i
li
ty
an
d
r
e
al
-
ti
m
e
p
er
f
o
r
m
an
c
e.
G
iv
en
th
e
s
e
l
im
i
ta
t
i
o
n
s
,
f
u
tu
r
e
r
e
s
e
ar
c
h
s
h
o
u
ld
a
im
to
ex
ten
d
th
i
s
in
v
es
tig
at
io
n
ac
r
o
s
s
d
iv
er
s
e
d
a
ta
s
e
t
s
th
at
in
c
lu
d
e
a
b
r
o
ad
er
ar
r
ay
o
f
p
l
an
t
s
p
ec
ie
s
an
d
d
i
f
f
er
en
t
en
v
ir
o
n
m
en
t
al
co
n
t
ex
t
s
to
en
h
an
ce
th
e
m
o
d
el
's
r
o
b
u
s
tn
e
s
s
an
d
ap
p
l
icab
i
li
ty
.
I
t
i
s
a
l
s
o
r
ec
o
m
m
en
d
ed
to
ex
p
lo
r
e
h
i
g
h
er
l
ev
e
l
s
o
f
d
a
ta
d
iv
er
s
it
y
,
in
c
lu
d
in
g
v
ar
y
in
g
im
a
g
e
q
u
al
it
ie
s
,
li
g
h
t
in
g
co
n
d
i
tio
n
s
,
an
d
ev
en
s
e
a
s
o
n
al
v
ar
i
at
io
n
s
in
p
l
an
t
ap
p
ea
r
an
ce
,
to
im
p
r
o
v
e
m
o
d
el
g
en
er
al
iza
t
io
n
an
d
ad
ap
tab
il
ity
.
Fu
r
th
er
m
o
r
e
,
r
e
s
e
ar
ch
s
h
o
u
l
d
e
x
am
in
e
h
o
w
FG
V
C
an
d
o
th
er
s
p
ec
ia
li
ze
d
te
ch
n
iq
u
e
s
ca
n
b
e
in
t
eg
r
a
ted
to
en
h
a
n
ce
th
e
c
l
as
s
if
i
ca
tio
n
ac
c
u
r
a
cy
f
o
r
s
p
e
cie
s
wi
th
s
u
b
t
le
v
i
s
u
al
d
i
s
t
i
n
ct
io
n
s
.
E
x
p
an
d
in
g
m
o
d
el
in
ter
p
r
e
tab
il
it
y
t
o
b
e
t
ter
u
n
d
er
s
tan
d
th
e
d
e
ci
s
io
n
-
m
ak
in
g
p
r
o
ce
s
s
o
f
C
N
Ns
c
o
u
ld
al
s
o
p
r
o
v
id
e
v
alu
ab
l
e
in
s
ig
h
t
s
,
e
s
p
ec
ia
l
ly
f
o
r
p
r
a
ct
ic
al
ap
p
l
ica
ti
o
n
s
i
n
ag
r
i
cu
l
tu
r
e
an
d
b
o
tan
y
.
W
e
p
r
o
p
o
s
e
ex
p
lo
r
in
g
in
t
er
p
r
e
ta
b
i
li
ty
ap
p
r
o
a
ch
e
s
s
u
ch
a
s
g
r
ad
ien
t
-
w
eig
h
t
ed
c
l
as
s
ac
tiv
at
io
n
m
ap
p
in
g
(
Gr
a
d
-
C
AM
)
an
d
lo
ca
l
in
t
er
p
r
e
ta
b
l
e
m
o
d
el
-
ag
n
o
s
ti
c
ex
p
l
an
a
tio
n
s
(
L
I
ME
)
.
T
h
e
s
e
t
ec
h
n
iq
u
e
s
wo
u
ld
al
lo
w
u
s
to
o
b
s
e
r
v
e
th
e
r
eg
io
n
s
o
f
t
h
e
im
ag
e
t
h
at
m
o
s
t
co
n
t
r
ib
u
t
e
to
c
la
s
s
i
f
i
ca
tio
n
,
p
r
o
v
id
in
g
tr
an
s
p
ar
en
cy
to
th
e
m
o
d
el
an
d
en
ab
lin
g
r
es
ea
r
ch
er
s
to
v
al
id
a
te
wh
e
th
er
th
e
m
o
d
el
i
s
u
s
i
n
g
v
al
id
b
o
tan
ic
al
f
e
atu
r
e
s
,
s
u
ch
a
s
t
ex
t
u
r
e
p
at
ter
n
s
,
ed
g
e
s
,
an
d
l
ea
f
v
ein
s
.
E
x
p
an
d
in
g
m
o
d
el
in
t
er
p
r
eta
b
i
li
ty
to
b
e
tt
er
u
n
d
er
s
t
an
d
th
e
d
e
ci
s
io
n
-
m
ak
in
g
p
r
o
ce
s
s
o
f
C
NN
s
co
u
ld
p
r
o
v
id
e
v
alu
ab
le
in
s
i
g
h
t
s
,
esp
ec
ia
lly
f
o
r
p
r
a
ct
ic
a
l
ap
p
l
ica
t
io
n
s
i
n
ag
r
ic
u
l
tu
r
e
an
d
b
o
tan
y
.
T
h
e
im
p
l
ic
at
io
n
s
o
f
th
i
s
r
e
s
ea
r
c
h
ar
e
s
ig
n
if
i
ca
n
t
f
o
r
f
ie
ld
s
th
a
t
d
ep
en
d
o
n
ac
c
u
r
a
te
p
l
an
t
id
e
n
tif
ic
at
io
n
,
s
u
ch
a
s
ag
r
i
cu
l
tu
r
e,
m
ed
ic
in
e
,
an
d
t
h
e
f
o
o
d
in
d
u
s
tr
y
.
Au
to
m
a
ted
h
er
b
c
la
s
s
if
ic
at
io
n
co
u
ld
s
t
r
e
am
l
in
e
p
r
o
ce
s
s
e
s
,
r
ed
u
c
e
h
u
m
an
er
r
o
r
,
an
d
in
cr
ea
s
e
ef
f
ic
ien
c
y
in
ap
p
li
ca
tio
n
s
r
an
g
in
g
f
r
o
m
b
o
t
an
i
ca
l
r
e
s
e
ar
ch
to
c
o
m
m
e
r
c
ia
l
p
r
o
d
u
ct
d
ev
e
lo
p
m
en
t
.
A
s
m
o
d
el
s
co
n
ti
n
u
e
to
im
p
r
o
v
e
in
a
cc
u
r
a
cy
an
d
ad
ap
tab
il
i
ty
,
th
ey
co
u
ld
b
e
in
t
eg
r
a
te
d
in
to
m
o
b
il
e
o
r
I
o
T
d
ev
i
ce
s
,
p
r
o
v
id
in
g
o
n
-
s
it
e
p
lan
t
id
e
n
tif
ic
at
io
n
an
d
f
u
r
th
er
d
e
m
o
cr
a
ti
zin
g
a
cc
e
s
s
to
AI
-
d
r
iv
en
b
o
t
an
i
ca
l c
la
s
s
if
i
ca
t
io
n
to
o
l
.
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
.
5
,
Octo
b
er
2
0
2
5
:
4
1
2
3
-
4
1
3
6
4132
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
s
tu
d
y
was
f
in
an
ce
d
in
p
ar
t
b
y
th
e
C
o
o
r
d
en
aç
ão
d
e
Ap
er
f
eiço
am
en
to
d
e
Pes
s
o
al
d
e
Nív
el
Su
p
er
io
r
–
B
r
asil
(
C
APES)
–
Fin
an
ce
C
o
d
e
0
0
1
.
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
Sam
ir
a
Nascim
en
to
An
tu
n
es
✓
✓
✓
✓
✓
Ma
d
alen
a
De
Oliv
eir
a
B
ar
b
o
s
a
Div
in
o
✓
✓
✓
✓
L
u
an
a
d
o
s
San
to
s
C
o
r
d
eir
o
✓
✓
✓
Fer
n
an
d
a
Per
eir
a
L
eite
Ag
u
iar
✓
✓
✓
✓
✓
✓
✓
Ma
r
ce
lo
T
s
u
g
u
io
Ok
an
o
✓
✓
✓
✓
✓
✓
✓
✓
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.
I
NF
O
RM
E
D
CO
NS
E
N
T
W
e
h
av
e
o
b
tain
ed
in
f
o
r
m
ed
c
o
n
s
en
t f
r
o
m
all
in
d
iv
id
u
als in
c
lu
d
ed
in
t
h
is
s
tu
d
y
.
E
T
H
I
CAL AP
P
RO
V
AL
T
h
is
s
tu
d
y
d
id
n
o
t in
v
o
lv
e
h
u
m
an
p
ar
ticip
an
ts
o
r
an
im
als.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
aila
b
le
o
n
r
eq
u
est
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
,
[
MT
O
]
.
T
h
e
d
ata,
wh
ich
co
n
tain
in
f
o
r
m
atio
n
t
h
at
co
u
ld
co
m
p
r
o
m
is
e
th
e
p
r
iv
ac
y
o
f
r
esear
c
h
p
ar
ticip
an
ts
,
ar
e
n
o
t p
u
b
licly
a
v
ailab
le
d
u
e
to
ce
r
tain
r
estrictio
n
s
.
RE
F
E
R
E
NC
E
S
[
1
]
A
.
P
r
e
e
y
a
V
,
S
.
P
.
R
a
j
a
,
B
.
K
.
D
h
a
n
a
l
a
k
s
h
mi
,
H
.
G
u
r
u
r
a
j
,
V
.
R
a
v
i
,
a
n
d
P
.
R
a
v
i
,
“
P
e
r
c
e
p
t
u
a
l
p
i
g
e
o
n
g
a
l
v
a
n
i
z
e
d
o
p
t
i
m
i
z
a
t
i
o
n
o
f
mu
l
t
i
-
o
b
j
e
c
t
i
v
e
C
N
N
o
n
t
h
e
i
d
e
n
t
i
f
i
c
a
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
ma
n
g
o
l
e
a
v
e
s
d
i
s
e
a
s
e
,
”
T
h
e
O
p
e
n
A
g
ri
c
u
l
t
u
re
J
o
u
rn
a
l
,
v
o
l
.
1
8
,
n
o
.
1
,
N
o
v
.
2
0
2
4
,
d
o
i
:
1
0
.
2
1
7
4
/
0
1
1
8
7
4
3
3
1
5
3
3
1
2
6
4
2
4
0
9
1
7
0
7
3
4
3
1
.
[
2
]
D
.
B
.
V
a
l
d
e
z
,
C
.
J.
G
.
A
l
i
a
c
,
a
n
d
L
.
S
.
F
e
l
i
s
c
u
z
o
,
“
M
e
d
i
c
i
n
a
l
p
l
a
n
t
c
l
a
ssi
f
i
c
a
t
i
o
n
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
a
n
d
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
i
n
4
t
h
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
i
n
E
n
g
i
n
e
e
ri
n
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
I
I
C
AI
ET
2
0
2
2
,
S
e
p
.
2
0
2
2
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
I
C
A
I
ET5
5
1
3
9
.
2
0
2
2
.
9
9
3
6
8
6
8
.
[
3
]
S
.
C
h
a
i
v
i
v
a
t
r
a
k
u
l
,
J.
M
o
o
n
r
i
n
t
a
,
a
n
d
S
.
C
h
a
i
w
i
w
a
t
r
a
k
u
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
s
f
o
r
h
e
r
b
i
d
e
n
t
i
f
i
c
a
t
i
o
n
:
p
l
a
i
n
b
a
c
k
g
r
o
und
a
n
d
n
a
t
u
r
a
l
e
n
v
i
r
o
n
me
n
t
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
n
Ad
v
a
n
c
e
d
S
c
i
e
n
c
e
,
E
n
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
2
,
n
o
.
3
,
p
p
.
1
2
4
4
–
1
2
5
2
,
Ju
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
8
5
1
7
/
i
j
a
se
i
t
.
1
2
.
3
.
1
5
3
4
8
.
[
4
]
R
.
K
.
D
o
m
i
n
g
u
e
s,
J.
A
.
C
o
r
d
i
o
l
i
,
a
n
d
D
.
S
i
l
v
a
,
“
D
e
v
e
l
o
p
me
n
t
o
f
a
m
e
t
h
o
d
o
l
o
g
y
f
o
r
d
i
a
g
n
o
si
n
g
f
a
u
l
t
s
i
n
b
e
a
r
i
n
g
s
o
p
e
r
a
t
i
n
g
u
n
d
e
r
v
a
r
i
a
b
l
e
o
p
e
r
a
t
i
n
g
c
o
n
d
i
t
i
o
n
s
b
a
se
d
o
n
s
e
l
f
-
s
u
p
e
r
v
i
se
d
l
e
a
r
n
i
n
g
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
A
n
n
u
a
l
C
o
n
f
e
re
n
c
e
o
f
t
h
e
Pro
g
n
o
s
t
i
c
s
a
n
d
H
e
a
l
t
h
M
a
n
a
g
e
m
e
n
t
S
o
c
i
e
t
y
,
PH
M
,
v
o
l
.
1
6
,
n
o
.
1
,
2
0
2
4
,
d
o
i
:
1
0
.
3
6
0
0
1
/
p
h
mco
n
f
.
2
0
2
4
.
v
1
6
i
1
.
4
1
8
3
.
[
5
]
O
.
R
u
ssa
k
o
v
s
k
y
e
t
a
l
.
,
“
I
ma
g
e
N
e
t
l
a
r
g
e
sca
l
e
v
i
s
u
a
l
r
e
c
o
g
n
i
t
i
o
n
c
h
a
l
l
e
n
g
e
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
Vi
si
o
n
,
v
o
l
.
1
1
5
,
n
o
.
3
,
p
p
.
2
1
1
–
2
5
2
,
2
0
1
5
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
2
6
3
-
0
1
5
-
0
8
1
6
-
y.
Evaluation Warning : The document was created with Spire.PDF for Python.