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.
15
,
No
.
5
,
Octo
b
er
20
26
,
p
p
.
4
4
0
3
~
4
4
1
1
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
5
.
p
p
4
4
0
3
-
4
4
1
1
4403
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Perf
o
rma
nce
a
na
ly
sis
of
co
nv
o
lutiona
l
neura
l
ne
twor
k
-
ba
sed
ima
g
e
cla
ss
ifiers
f
o
r
a
g
ricultural
fa
rmla
nd
ma
pping
Anj
ela
C.
T
o
lent
ino
,
Ana
ze
l
P.
G
a
m
illa
,
Am
ir
L
edesm
a
D
e
p
a
r
t
me
n
t
of
I
n
f
o
r
m
a
t
i
o
n
Te
c
h
n
o
l
o
g
y
,
C
o
l
l
e
g
e
of
E
n
g
i
n
e
e
r
i
n
g
,
C
e
n
t
r
a
l
L
u
z
o
n
S
t
a
t
e
U
n
i
v
e
r
si
t
y
,
S
c
i
e
n
c
e
C
i
t
y
of
M
u
n
o
z
,
P
h
i
l
i
p
p
i
n
e
s
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
4
,
2
0
2
5
R
ev
is
ed
Au
g
7
,
2
0
2
6
Acc
ep
ted
Au
g
12
,
2
0
2
6
P
re
c
ise
fa
rm
lan
d
m
a
p
p
in
g
is
e
ss
e
n
ti
a
l
fo
r
c
ro
p
m
o
n
it
o
ri
n
g
,
lan
d
-
u
se
in
v
e
n
to
r
y
,
a
n
d
su
b
si
d
y
a
u
d
it
s.
P
a
rti
c
u
larly
,
t
h
e
o
ffice
of
t
h
e
m
u
n
icip
a
li
ty
’
s
a
ss
e
ss
o
r
or
th
e
re
g
istry
of
d
e
e
d
s
n
e
e
d
s
a
d
e
q
u
a
te
ima
g
e
c
las
sifiers
to
c
a
teg
o
rize
fa
rm
lan
d
on
a
lar
g
e
s
c
a
le
a
c
ro
ss
d
iffere
n
t
se
a
so
n
s
a
n
d
re
g
i
o
n
s.
Th
e
stu
d
y
e
v
a
lu
a
ted
t
h
re
e
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(
CNN
)
m
o
d
e
ls,
su
c
h
as
e
fficie
n
t
n
e
tw
o
rk
b
a
se
li
n
e
3
(
Eff
icie
n
tNe
tB3
)
,
re
sid
u
a
l
n
e
tw
o
r
k
5
0
lay
e
rs
v
e
rsio
n
2
(
Re
sN
e
t5
0
V2
)
,
a
n
d
i
n
c
e
p
ti
o
n
v
e
rsio
n
3
(
In
c
e
p
ti
o
n
V
3
)
,
to
i
d
e
n
ti
f
y
th
e
m
o
d
e
l
t
h
a
t
b
e
st
b
a
lan
c
e
s
a
c
c
u
ra
c
y
a
n
d
e
fficie
n
c
y
f
o
r
c
las
sify
in
g
h
ig
h
-
re
so
lu
ti
o
n
fa
rm
lan
d
ima
g
e
ry
.
Eff
i
c
ien
tNe
tB3
p
r
o
v
e
d
to
be
th
e
m
o
s
t
su
it
a
b
le
m
o
d
e
l,
d
e
m
o
n
stra
ti
n
g
h
ig
h
e
r
a
c
c
u
ra
c
y
,
r
o
b
u
stn
e
ss
,
a
n
d
a
stro
n
g
e
r
a
b
il
it
y
to
id
e
n
ti
f
y
rice
fa
rm
lan
d
p
a
tt
e
rn
s.
Wh
il
e
Re
sN
e
t5
0
V2
a
n
d
I
n
c
e
p
ti
o
n
V3
a
re
wid
e
ly
u
se
d
f
o
r
ima
g
e
a
n
a
ly
sis,
t
h
e
ir
p
e
rfo
rm
a
n
c
e
d
e
c
li
n
e
d
w
h
e
n
p
ro
c
e
ss
in
g
fa
rm
lan
d
ima
g
e
s
th
a
t
e
x
h
i
b
it
su
b
sta
n
ti
a
l
d
iffere
n
c
e
s
in
v
isu
a
l
c
h
a
r
a
c
teristics
.
Eff
icie
n
tNe
tB3
d
e
li
v
e
re
d
m
o
re
c
o
n
siste
n
t
re
su
lt
s
f
o
r
fa
rm
la
n
d
ima
g
e
c
las
sifica
ti
o
n
in
larg
e
-
sc
a
le
a
g
ricu
lt
u
ra
l
a
p
p
li
c
a
ti
o
n
s.
T
h
e
se
fin
d
i
n
g
s
h
ig
h
li
g
h
t
its
p
o
ten
ti
a
l
fo
r
su
p
p
o
rti
n
g
d
a
ta
-
d
riv
e
n
a
g
r
icu
lt
u
ra
l
m
a
n
a
g
e
m
e
n
t
a
n
d
p
o
li
c
y
imp
lem
e
n
tatio
n
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
m
o
d
els
Dee
p
lear
n
in
g
E
f
f
icien
tNetB
3
Far
m
lan
d
m
ap
p
in
g
I
m
ag
e
class
if
ier
s
T
h
is
is
an
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r
th
e
CC
BY
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
An
jela
C.
T
o
len
tin
o
Dep
ar
tm
en
t
of
I
n
f
o
r
m
atio
n
T
e
ch
n
o
lo
g
y
,
C
o
lleg
e
of
E
n
g
in
ee
r
in
g
,
C
en
tr
al
L
u
zo
n
State
Un
iv
er
s
ity
Scien
ce
C
ity
of
Mu
n
o
z,
Ph
ilip
p
in
es
E
m
ail:
an
jelato
len
tin
o
@
clsu
.
ed
u
.
p
h
1.
I
NT
RO
D
UCT
I
O
N
Pre
cise
an
d
tim
ely
f
ar
m
lan
d
m
ap
p
in
g
p
la
y
s
an
im
p
o
r
tan
t
r
o
le
in
lan
d
-
u
s
e
in
v
en
to
r
y
,
cr
o
p
m
o
n
ito
r
in
g
,
an
d
a
g
r
icu
ltu
r
al
s
u
b
s
id
y
v
alid
atio
n
,
esp
ec
ially
in
th
e
Ph
ilip
p
in
es,
wh
er
e
ag
r
ic
u
ltu
r
e
co
n
tin
u
es
to
co
n
tr
ib
u
te
s
ig
n
if
ican
tly
to
e
m
p
lo
y
m
e
n
t
an
d
ec
o
n
o
m
ic
ac
tiv
ity
[
1
]
.
Go
v
e
r
n
m
en
t
o
f
f
ices
s
u
ch
as
m
u
n
ici
p
al
ass
es
s
o
r
s
an
d
th
e
r
eg
is
tr
y
of
d
ee
d
s
r
eq
u
ir
e
r
eliab
le
an
d
s
ca
lab
le
ap
p
r
o
ac
h
es
to
class
if
y
ag
r
icu
ltu
r
al
lan
d
s
ac
r
o
s
s
r
eg
io
n
s
.
Far
m
lan
d
m
ap
p
in
g
in
th
e
co
u
n
tr
y
r
em
ai
n
s
ch
allen
g
in
g
due
to
its
ar
ch
ip
elag
ic
s
ettin
g
,
f
r
ag
m
en
te
d
lan
d
p
ar
ce
ls
,
d
if
f
er
en
t
ter
r
ai
n
,
in
te
r
cr
o
p
p
in
g
p
r
ac
tices,
an
d
f
r
eq
u
en
t
cl
o
u
d
co
v
e
r
in
s
atellite
im
ag
er
y
[
2
]
,
[
3
]
.
Ma
n
y
ar
ea
s
s
till
r
ely
on
m
a
n
u
al
s
u
r
v
ey
in
g
an
d
v
is
u
al
p
h
o
to
in
te
r
p
r
etat
io
n
,
wh
ic
h
r
e
q
u
ir
e
s
u
b
s
tan
tial
tim
e,
m
an
p
o
wer
,
an
d
f
in
an
cial
r
eso
u
r
ce
s
.
T
h
ese
lim
itatio
n
s
h
ig
h
lig
h
t
th
e
n
ee
d
f
o
r
more
ef
f
icien
t
an
d
au
to
m
ated
class
if
icatio
n
m
eth
o
d
s
[
4
]
.
R
ec
en
t
ad
v
an
ce
s
in
d
ee
p
l
ea
r
n
in
g
ha
s
p
r
o
v
en
e
f
f
ec
tiv
e
in
im
ag
e
-
b
ased
lan
d
clas
s
if
icatio
n
.
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN)
in
p
ar
ticu
lar
,
ar
e
ca
p
ab
le
of
lear
n
in
g
s
p
atial
f
ea
tu
r
es
s
u
ch
as
ed
g
es,
tex
tu
r
es,
an
d
s
h
ap
es
d
ir
ec
tly
f
r
o
m
im
ag
e
d
ata
with
o
u
t
th
e
n
ee
d
f
o
r
m
an
u
al
f
ea
tu
r
e
e
n
g
in
ee
r
in
g
[
5
]
,
[
6
]
.
B
ec
au
s
e
of
th
is
ca
p
ab
ilit
y
,
C
NNs
h
av
e
b
ee
n
wid
ely
ap
p
li
ed
in
r
em
o
te
s
en
s
in
g
task
s
.
T
h
e
in
teg
r
atio
n
of
u
n
m
an
n
ed
ae
r
ial
v
eh
icle
(
UA
V
)
im
ag
er
y
with
d
ee
p
lear
n
in
g
tech
n
iq
u
es
h
as
f
u
r
th
er
im
p
r
o
v
ed
th
e
ac
c
u
r
ac
y
an
d
r
eso
lu
tio
n
of
f
ar
m
la
n
d
m
ap
p
in
g
[
7
]
,
[
8
]
.
C
o
m
p
a
r
ativ
e
an
aly
s
es
u
s
in
g
p
r
e
-
tr
ain
e
d
C
NN
m
o
d
els
f
u
r
th
er
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.
15
,
No
.
5
,
Octo
b
er
20
26
:
4
4
0
3
-
4
4
1
1
4404
h
ig
h
lig
h
t
th
e
ca
p
a
b
ilit
y
of
d
e
ep
lear
n
in
g
f
o
r
cr
o
p
m
o
n
ito
r
i
n
g
f
r
o
m
r
em
o
te
s
en
s
in
g
d
ata
[
9
]
.
Nev
er
th
eless
,
th
er
e
ar
e
s
till
lim
ited
co
m
p
ar
ativ
e
s
tu
d
ies
f
o
cu
s
ed
s
p
ec
if
ically
on
Ph
ilip
p
in
e
a
g
r
icu
ltu
r
a
l
lan
d
s
ca
p
es
[
1
0
]
.
Alth
o
u
g
h
in
ter
n
ati
o
n
al
s
tu
d
ies
d
em
o
n
s
tr
ate
th
e
ef
f
ec
tiv
en
ess
of
C
NNs
in
cr
o
p
class
if
icat
io
n
an
d
ag
r
icu
ltu
r
al
m
o
n
ito
r
in
g
[
1
1
]
,
[
1
2
]
,
co
n
s
id
e
r
in
g
th
e
s
tr
u
ctu
r
al
co
m
p
le
x
ity
of
lo
ca
l
f
ar
m
s
,
it
is
n
ec
es
s
ar
y
to
ev
alu
ate
wh
ich
C
NN
ar
ch
itectu
r
e
p
er
f
o
r
m
s
b
e
s
t
u
n
d
er
th
ese
co
n
d
itio
n
s
.
T
h
is
s
tu
d
y
co
m
p
ar
es
th
r
ee
C
NN
alg
o
r
ith
m
s
,
s
u
ch
as
r
esid
u
al
n
etwo
r
k
5
0
lay
er
s
v
er
s
io
n
2
(
R
esNet5
0
V2
)
,
in
ce
p
tio
n
v
e
r
s
io
n
3
(
I
n
ce
p
tio
n
V
3
)
,
an
d
as
ef
f
icien
t
n
etwo
r
k
b
aselin
e
3
(
E
f
f
icien
tNetB
3
)
,
to
d
eter
m
in
e
wh
ic
h
m
o
d
el
p
r
o
v
id
es
a
s
u
itab
le
b
ala
n
ce
b
et
wee
n
class
if
icatio
n
ac
cu
r
ac
y
an
d
c
o
m
p
u
tatio
n
al
ef
f
icien
cy
f
o
r
d
r
o
n
e
-
b
ased
f
a
r
m
lan
d
m
ap
p
in
g
[
1
3
]
−
[
1
5
]
.
By
co
n
d
u
ctin
g
a
d
ir
ec
t
p
er
f
o
r
m
an
ce
co
m
p
ar
is
o
n
.
T
he
s
tu
d
y
aim
s
to
p
r
o
v
i
d
e
p
r
a
ctica
l
g
u
id
an
ce
f
o
r
s
elec
tin
g
a
p
p
r
o
p
r
iate
d
ee
p
lea
r
n
in
g
m
o
d
els
f
o
r
ag
r
icu
ltu
r
al
lan
d
class
if
icatio
n
in
th
e
Ph
ilip
p
in
es.
2.
M
E
T
H
O
D
T
h
is
s
ec
tio
n
o
u
tlin
es
th
e
m
eth
o
d
o
lo
g
y
e
m
p
lo
y
ed
to
an
aly
z
e
th
e
p
er
f
o
r
m
a
n
ce
of
th
r
ee
w
id
ely
u
s
ed
C
NN
alg
o
r
ith
m
m
o
d
els
in
th
e
co
n
tex
t
of
im
a
g
e
class
if
icatio
n
.
T
h
e
p
er
f
o
r
m
an
ce
ev
alu
atio
n
is
s
tr
u
ctu
r
ed
i
n
to
th
r
ee
m
ain
co
m
p
o
n
en
ts
:
i
)
d
ataset
p
r
ep
ar
atio
n
,
ii
)
s
elec
tio
n
of
C
NN
alg
o
r
ith
m
s
,
an
d
iii
)
a
p
p
licatio
n
of
C
NN
m
o
d
els.
Fo
llo
win
g
th
e
tr
ain
in
g
p
h
ase,
th
e
m
o
d
els
’
p
er
f
o
r
m
an
ce
is
s
y
s
tem
atica
lly
as
s
es
s
e
d
to
d
eter
m
in
e
th
eir
ef
f
ec
tiv
en
ess
in
class
if
y
in
g
ag
r
icu
ltu
r
al
f
ar
m
la
n
d
im
ag
es.
2
.
1
.
Da
t
a
s
et
prepa
ra
t
io
n
T
h
is
s
tu
d
y
ev
alu
ates
th
e
p
er
f
o
r
m
an
ce
of
th
r
ee
C
NN
alg
o
r
ith
m
m
o
d
els
u
s
in
g
ae
r
ial
im
ag
e
r
y
ca
p
tu
r
e
d
by
d
r
o
n
es.
T
h
e
d
ataset
co
n
s
is
ts
of
f
ar
m
lan
d
im
ag
es
co
llected
th
r
o
u
g
h
d
r
o
n
e
-
b
ased
im
ag
i
n
g
s
y
s
tem
s
.
T
h
ese
im
ag
es
s
er
v
e
as
t
h
e
in
p
u
t
f
o
r
t
r
ain
in
g
an
d
test
in
g
th
e
m
o
d
els
an
d
ar
e
u
s
ed
to
ass
ess
th
e
cla
s
s
if
icatio
n
ac
cu
r
ac
y
of
each
C
NN
m
o
d
el
in
m
ap
p
in
g
ag
r
icu
ltu
r
al
f
ar
m
lan
d
s
.
All
im
ag
es
wer
e
r
e
s
ized
to
th
e
r
eq
u
ir
ed
in
p
u
t
d
im
en
s
io
n
s
of
th
e
s
elec
ted
C
NN
m
o
d
els.
Pix
el
v
alu
es
wer
e
n
o
r
m
alize
d
p
r
i
o
r
to
tr
ain
in
g
.
Data
au
g
m
en
tatio
n
tech
n
iq
u
es,
in
clu
d
in
g
r
o
tatio
n
an
d
h
o
r
izo
n
tal
f
lip
p
in
g
,
wer
e
ap
p
lied
to
i
n
cr
ea
s
e
v
a
r
iab
ilit
y
an
d
r
e
d
u
ce
o
v
er
f
itti
n
g
.
T
h
e
m
o
d
els
wer
e
tr
ain
ed
f
o
r
a
f
ix
ed
n
u
m
b
er
of
ep
o
ch
s
with
ea
r
ly
s
to
p
p
in
g
im
p
lem
e
n
ted
to
p
r
ev
en
t
p
er
f
o
r
m
an
ce
d
eg
r
a
d
atio
n
.
2
.
2
.
Select
io
n
of
co
nv
o
lutio
n
a
l neura
l net
wo
rk
a
lg
o
rit
hm
m
o
dels
T
h
is
s
tu
d
y
u
tili
ze
s
th
r
ee
wi
d
ely
r
ec
o
g
n
ized
C
NN
alg
o
r
ith
m
m
o
d
els
f
o
r
im
ag
e
clas
s
if
icatio
n
:
E
f
f
icien
tNetB
3
,
R
esNet5
0
V2
,
an
d
I
n
ce
p
tio
n
V3
.
T
h
ese
m
o
d
els
wer
e
s
elec
ted
due
to
th
eir
p
r
o
v
e
n
ac
cu
r
ac
y
in
p
r
io
r
s
tu
d
ies
an
d
th
eir
d
is
tin
ct
s
tr
u
ctu
r
al
d
esig
n
s
,
wh
ich
p
r
o
v
id
e
a
v
a
r
ied
p
er
s
p
ec
tiv
e
in
ev
alu
atin
g
m
o
d
el
p
er
f
o
r
m
an
ce
[
1
6
]
.
T
h
eir
r
o
b
u
s
tn
ess
an
d
ab
ilit
y
to
p
r
o
ce
s
s
h
i
g
h
-
r
eso
lu
tio
n
im
ag
e
d
ata
m
ak
e
th
em
well
-
s
u
ited
f
o
r
t
h
e
class
if
icatio
n
of
f
a
r
m
lan
d
im
a
g
er
y
.
T
h
i
s
d
e
m
o
n
s
t
r
a
t
e
s
s
t
r
o
n
g
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
c
a
p
a
b
i
l
i
t
i
es
w
h
i
le
m
a
i
n
t
a
i
n
i
n
g
l
o
w
m
e
m
o
r
y
u
s
a
g
e
a
n
d
p
r
o
c
e
s
s
i
n
g
t
i
m
e
.
In
a
d
d
i
ti
o
n
,
t
h
e
c
o
m
p
a
c
t
d
es
i
g
n
is
s
u
i
ta
b
l
e
f
o
r
d
e
p
l
o
y
m
e
n
t
on
r
e
s
o
u
r
c
e
-
c
o
n
s
t
r
a
i
n
e
d
d
e
v
i
ce
s
s
u
c
h
as
d
r
o
n
e
s
a
n
d
e
d
g
e
d
e
v
ic
e
s
u
s
e
d
in
a
g
r
i
c
u
l
t
u
r
al
m
o
n
i
t
o
r
i
n
g
[
1
7
]
.
R
esNet5
0
V2
b
u
ild
s
upon
th
e
o
r
ig
in
al
R
esNet
by
in
teg
r
atin
g
id
en
tity
s
h
o
r
tcu
t
co
n
n
ec
ti
o
n
s
,
wh
ich
s
u
p
p
o
r
t
s
tab
le
g
r
ad
ien
t
p
r
o
p
a
g
atio
n
d
u
r
in
g
th
e
tr
ain
in
g
of
d
ee
p
n
eu
r
al
n
etwo
r
k
s
.
T
h
e
d
ep
th
an
d
s
tr
u
ctu
r
e
en
ab
le
ef
f
ec
tiv
e
lear
n
in
g
of
h
ier
ar
ch
ical
r
ep
r
esen
tatio
n
s
,
m
ak
in
g
it
h
ig
h
ly
s
u
itab
le
f
o
r
lar
g
e
-
s
ca
le
im
a
g
e
class
if
icatio
n
.
T
h
r
o
u
g
h
its
r
o
b
u
s
tn
ess
,
s
ca
lab
ilit
y
,
an
d
s
tr
o
n
g
tr
an
s
f
er
lear
n
i
n
g
ca
p
ab
il
ities
,
th
is
m
o
d
el
is
wid
ely
ap
p
lied
in
p
r
ac
tical
co
m
p
u
ter
v
is
io
n
task
s
[
1
8
]
.
I
n
ce
p
tio
n
V3
in
tr
o
d
u
ce
s
I
n
ce
p
tio
n
m
o
d
u
les
th
at
u
tili
ze
p
ar
allel
co
n
v
o
lu
tio
n
al
lay
e
r
s
to
ca
p
tu
r
e
f
ea
tu
r
es
at
m
u
ltip
le
s
ca
les.
It
is
d
esig
n
ed
f
o
r
e
f
f
ici
en
cy
,
e
n
ab
lin
g
th
e
p
r
o
ce
s
s
in
g
of
lar
g
e
im
ag
e
d
a
tasets
wi
th
o
u
t
s
ig
n
if
ican
tly
in
cr
ea
s
in
g
co
m
p
u
tatio
n
al
d
em
an
d
.
I
ts
b
alan
ce
of
ac
cu
r
ac
y
an
d
e
f
f
icien
cy
m
ak
es
it
a
p
r
ac
tical
ch
o
ice
f
o
r
f
ar
m
lan
d
m
ap
p
in
g
task
s
wh
er
e
m
u
lti
-
s
ca
le
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
r
eso
u
r
ce
ef
f
icien
cy
ar
e
r
eq
u
ir
ed
[
1
7
]
,
[
1
9
]
.
T
h
ese
th
r
ee
ar
c
h
itectu
r
es
wer
e
im
p
lem
en
ted
an
d
co
m
p
ar
ed
u
n
d
er
th
e
s
am
e
ex
p
er
im
en
tal
co
n
d
itio
n
s
to
d
eter
m
in
e
th
e
m
o
s
t
ef
f
ec
tiv
e
C
NN
m
o
d
el
f
o
r
f
ar
m
lan
d
m
ap
p
in
g
.
2
.
3
.
Appl
ica
t
io
n
of
c
o
nv
o
luti
o
na
l neura
l net
wo
rk
mo
dels
T
h
e
C
NN
m
o
d
els
u
s
ed
in
th
is
s
tu
d
y
wer
e
E
f
f
icien
tNetB
3
,
R
esNet5
0
V2
,
an
d
I
n
ce
p
tio
n
V3
,
wer
e
im
p
lem
en
ted
u
s
in
g
th
e
Py
th
o
n
p
r
o
g
r
am
m
in
g
lan
g
u
ag
e
a
n
d
th
e
T
en
s
o
r
Flo
w
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
.
T
h
e
f
ar
m
lan
d
im
ag
e
d
ataset
was
f
i
r
s
t
p
r
ep
ar
e
d
a
n
d
p
r
ep
r
o
ce
s
s
ed
to
en
s
u
r
e
th
at
th
e
im
ag
es
we
r
e
s
u
itab
le
f
o
r
m
o
d
el
tr
ain
in
g
a
n
d
a
n
aly
s
is
.
T
h
e
ex
p
er
im
en
tal
s
etu
p
f
o
ll
o
wed
th
e
s
eq
u
e
n
ce
s
h
o
wn
b
elo
w
to
ev
alu
ate
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
C
NN
m
o
d
els f
o
r
f
ar
m
lan
d
m
ap
p
in
g
.
Far
m
lan
d
im
ag
e
d
ataset
→
p
r
ep
r
o
ce
s
s
in
g
→
C
NN
m
o
d
els
→
tr
ain
in
g
→
test
in
g
→
p
e
r
f
o
r
m
a
n
ce
ev
alu
atio
n
→
b
est
m
o
d
el
s
elec
tio
n
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
P
erfo
r
ma
n
ce
a
n
a
lysi
s
o
f c
o
n
v
o
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
-
b
a
s
e
d
ima
g
e
cla
s
s
ifier
s
fo
r
…
(
A
n
jela
C
.
To
len
tin
o
)
4405
T
h
e
m
o
d
els
wer
e
tr
ain
ed
an
d
ev
alu
ated
u
s
in
g
a
d
ataset
o
f
d
r
o
n
e
-
ca
p
tu
r
ed
f
ar
m
lan
d
im
ag
es.
E
ac
h
m
o
d
el
was
tr
ain
ed
f
o
r
1
0
ep
o
c
h
s
u
n
d
er
id
en
tical
tr
ain
in
g
co
n
d
itio
n
s
to
en
s
u
r
e
a
f
air
p
er
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
.
A
to
tal
o
f
2
0
f
a
r
m
lan
d
im
a
g
e
s
wer
e
u
s
ed
as
th
e
test
in
g
d
ataset
f
o
r
m
o
d
el
ev
al
u
atio
n
.
Fig
u
r
es
1
to
3
p
r
esen
t
th
e
s
im
u
latio
n
r
esu
lts
o
f
E
f
f
icien
tNetB
3
,
R
esNet5
0
V2
,
an
d
I
n
ce
p
tio
n
V
3
,
r
esp
ec
tiv
ely
,
illu
s
tr
atin
g
t
h
eir
class
if
icatio
n
p
er
f
o
r
m
an
ce
b
ased
o
n
f
ea
tu
r
e
s
im
ilar
ity
.
T
h
e
class
if
icatio
n
o
u
tp
u
ts
o
f
th
e
th
r
ee
C
NN
ar
ch
itectu
r
es
wer
e
co
m
p
ar
ed
b
ased
o
n
th
eir
f
ea
tu
r
e
s
im
ilar
i
ty
an
d
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
to
d
eter
m
in
e
th
e
m
o
s
t e
f
f
ec
tiv
e
m
o
d
el
f
o
r
f
ar
m
l
an
d
m
ap
p
in
g
.
Fig
u
r
e
1.
Simu
latio
n
of
E
f
f
ici
en
tNetB
3
Fig
u
r
e
2.
Simu
latio
n
of
R
esNet5
0
V2
Fig
u
r
e
3.
Simu
latio
n
of
I
n
ce
p
t
io
n
V3
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.
15
,
No
.
5
,
Octo
b
er
20
26
:
4
4
0
3
-
4
4
1
1
4406
3.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
p
r
esen
ts
an
o
v
er
v
iew
an
d
p
er
f
o
r
m
an
ce
an
aly
s
is
of
th
e
th
r
ee
C
NN
alg
o
r
ith
m
m
o
d
els
u
s
ed
in
th
is
s
tu
d
y
:
E
f
f
icien
tNetB
3
,
R
esNe
t5
0
V2
,
an
d
I
n
ce
p
tio
n
V3
.
T
h
ese
m
o
d
els
wer
e
te
s
ted
u
s
in
g
a
d
ataset
of
f
ar
m
lan
d
im
ag
es
ca
p
tu
r
e
d
by
d
r
o
n
es.
T
h
e
ev
al
u
atio
n
f
o
cu
s
ed
on
th
eir
ac
cu
r
ac
y
,
lear
n
in
g
ef
f
icien
c
y
,
an
d
ab
ilit
y
to
class
if
y
d
if
f
er
en
t
a
g
r
icu
ltu
r
al
lan
d
f
ea
t
u
r
es.
3.
1
.
P
er
f
o
r
m
a
nce
a
na
ly
s
is
of
co
nv
o
lutio
na
l neura
l net
wo
rk
a
lg
o
rit
hm
m
o
dels
T
h
e
c
la
s
s
i
f
i
c
a
t
i
o
n
p
e
r
f
o
r
m
an
c
e
of
E
f
f
i
c
i
en
t
N
e
t
B
3
,
R
e
s
N
et
5
0
V
2
,
an
d
I
n
c
e
p
t
io
n
V3
was
e
v
a
lu
a
t
ed
u
s
i
n
g
a
d
a
t
a
s
e
t
of
1
1
5
d
r
o
n
e
-
c
ap
t
u
r
ed
f
a
r
m
l
a
n
d
i
m
ag
es
.
D
u
e
to
t
h
e
l
im
i
t
e
d
av
a
i
la
b
i
l
i
t
y
of
l
a
b
e
l
e
d
a
g
r
i
cu
l
t
u
r
a
l
d
a
t
a,
tr
an
s
f
e
r
l
e
ar
n
i
n
g
was
em
p
lo
y
ed
,
a
l
lo
w
i
n
g
th
e
p
r
e
t
r
a
in
e
d
m
o
d
e
l
s
to
l
ev
e
r
ag
e
p
r
e
v
io
u
s
l
y
l
e
a
r
n
ed
f
e
a
t
u
r
e
r
e
p
r
e
s
e
n
t
a
t
io
n
s
a
n
d
a
d
ap
t
th
e
m
to
t
h
e
t
a
r
g
e
t
t
a
s
k
.
T
h
e
a
cc
u
r
a
c
y
of
t
h
e
m
o
d
e
l
s
is
e
v
a
lu
a
t
e
d
d
u
r
in
g
t
wo
t
e
s
t
i
n
g
c
y
c
l
e
s
w
i
t
h
10
i
te
r
a
t
io
n
s
[
2
0
]
,
[
2
1
]
.
T
h
e
d
a
t
a
s
e
t
was
d
iv
i
d
ed
i
n
to
t
r
a
i
n
i
n
g
a
n
d
t
e
s
t
i
n
g
s
e
t
s
to
a
s
s
e
s
s
c
l
a
s
s
i
f
i
c
a
t
i
o
n
ac
c
u
r
ac
y
on
u
n
s
e
en
i
m
a
g
e
s
[
2
2
]
.
T
h
e
te
s
t
i
n
g
p
h
a
s
e
i
n
c
lu
d
es
b
o
t
h
f
ar
m
l
an
d
an
d
non
-
f
ar
m
l
an
d
im
a
g
e
s
to
co
m
p
a
r
e
th
e
t
h
r
e
e
C
N
N
m
o
d
e
l
s
as
s
h
o
wn
in
F
i
g
u
r
e
4.
T
h
e
f
a
r
m
l
a
n
d
im
a
g
e
s
t
ak
e
n
f
r
o
m
d
r
o
n
e
s
p
r
o
v
id
e
d
as
e
x
a
m
p
l
e
s
a
r
e
i
l
lu
s
t
r
a
t
ed
in
F
ig
u
r
e
4
(
a)
,
wh
i
l
e
t
h
e
n
o
n
-
f
a
r
m
l
a
n
d
im
a
g
e
s
ar
e
i
l
l
u
s
t
r
a
t
ed
in
F
i
g
u
r
e
4
(
b
)
.
(
a
)
(
b
)
Fig
u
r
e
4.
Sam
p
le
d
ataset
f
o
r
t
h
e
p
er
f
o
r
m
an
ce
an
aly
s
is
of
th
e
th
r
ee
C
NN
alg
o
r
ith
m
m
o
d
els
(
a)
f
ar
m
la
n
d
im
ag
es
an
d
(
b
)
n
o
n
-
f
ar
m
lan
d
im
a
g
es
3.
2
.
T
est
ca
s
es
T
h
r
ee
C
NN
m
o
d
els
a
r
e
ap
p
lied
to
s
atellite
im
ag
er
y
f
o
r
f
a
r
m
lan
d
m
a
p
p
in
g
to
i
d
en
tify
an
d
d
elin
ea
te
f
ar
m
lan
d
a
r
ea
s
.
T
h
e
s
tu
d
y
e
v
alu
ates
th
eir
p
er
f
o
r
m
an
ce
b
ased
on
ac
cu
r
ac
y
an
d
r
eliab
ilit
y
in
g
en
er
atin
g
f
ar
m
lan
d
m
ap
s
[
2
3
]
.
R
esu
lts
ar
e
co
m
p
a
r
ed
to
d
eter
m
in
e
w
h
ich
m
o
d
el
p
er
f
o
r
m
s
m
o
s
t
ef
f
ec
tiv
ely
u
n
d
er
t
h
e
g
iv
en
co
n
d
itio
n
s
.
3.
2
.1
.
T
est
ca
s
e
1
T
h
e
f
ir
s
t
test
ca
s
e
in
v
o
lv
es
an
aly
zin
g
th
e
ac
cu
r
ac
y
of
th
e
th
r
ee
alg
o
r
ith
m
s
u
s
in
g
th
e
test
s
et
co
n
s
is
ts
of
18
f
ar
m
la
n
d
im
ag
es
a
n
d
2
non
-
f
a
r
m
lan
d
im
ag
es.
T
h
e
ex
p
ec
ted
r
esu
lt
is
85%
m
atch
f
o
r
s
im
ilar
ity
an
d
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
P
erfo
r
ma
n
ce
a
n
a
lysi
s
o
f c
o
n
v
o
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
-
b
a
s
e
d
ima
g
e
cla
s
s
ifier
s
fo
r
…
(
A
n
jela
C
.
To
len
tin
o
)
4407
m
atch
f
o
r
n
o
n
-
s
im
ilar
ity
[
2
3
]
.
Fig
u
r
e
5
p
r
esen
ts
th
e
p
er
f
o
r
m
an
ce
an
aly
s
is
o
f
th
e
th
r
ee
C
NN
m
o
d
els
b
ased
o
n
f
ea
tu
r
e
s
im
ilar
ity
.
As
s
h
o
wn
in
Fig
u
r
e
5
(
a
)
,
E
f
f
icien
tNet
B
3
ac
h
iev
ed
8
5
%
s
im
ilar
ity
m
atch
es
an
d
1
5
%
non
-
s
im
ilar
ity
m
atc
h
es,
co
r
r
esp
o
n
d
in
g
to
a
p
er
f
ec
t
class
if
icatio
n
ac
cu
r
ac
y
o
f
1
0
0
%.
I
n
Fig
u
r
e
5
(
b
)
,
R
esNet5
0
V2
r
ec
o
r
d
ed
8
0
%
s
im
ilar
ity
m
atch
es
an
d
2
0
%
n
o
n
-
s
im
ilar
ity
m
atc
h
es,
r
esu
ltin
g
in
a
n
o
v
e
r
all
ac
cu
r
ac
y
o
f
8
0
%.
Me
a
n
wh
ile,
Fig
u
r
e
5
(
c)
s
h
o
ws
th
at
I
n
ce
p
t
io
n
V3
attain
ed
7
5
%
s
im
ilar
ity
m
atch
es
an
d
2
5
%
non
-
s
im
ilar
ity
m
atch
es,
with
an
ac
cu
r
ac
y
r
ate
o
f
7
5
%.
T
h
e
s
e
r
esu
lts
h
ig
h
lig
h
t
E
f
f
icien
tN
etB
3
’
s
ex
ce
p
tio
n
al
p
er
f
o
r
m
an
ce
in
im
ag
e
class
if
icatio
n
,
wh
ile
R
esNet5
0
V2
an
d
I
n
ce
p
tio
n
V
3
f
ell
s
h
o
r
t
o
f
th
e
r
eq
u
ir
e
d
ac
cu
r
ac
y
th
r
esh
o
ld
.
(
a)
(
b
)
(
c)
Fig
u
r
e
5.
C
lass
if
icatio
n
ac
cu
r
a
cy
test
ca
s
e
1
f
o
r
th
e
th
r
ee
C
NN
m
o
d
els
:
(
a)
E
f
f
icien
tNetB
3
,
(
b
)
R
esNet5
0
V2
,
an
d
(
c)
I
n
ce
p
tio
n
V3
3
.
2
.
2
.
T
est
ca
s
e
nu
m
ber
2
T
h
e
s
ec
o
n
d
test
ca
s
e
f
o
r
clas
s
if
icatio
n
ac
cu
r
ac
y
o
f
th
e
th
r
ee
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
co
n
s
is
ts
o
f
1
6
f
ar
m
lan
d
im
a
g
es
an
d
4
n
o
n
-
f
ar
m
lan
d
im
a
g
es
s
am
p
le
test
s
et.
T
h
e
ex
p
ec
ted
r
esu
lt
is
8
0
%
m
atch
f
o
r
s
im
ilar
ity
an
d
2
0
%
m
atch
f
o
r
n
o
n
-
s
im
ilar
ity
.
Fig
u
r
e
6
p
r
e
s
en
ts
th
e
p
er
f
o
r
m
an
ce
r
esu
lts
o
f
t
h
e
th
r
ee
C
NN
m
o
d
els
in
class
if
y
in
g
im
a
g
es
with
8
0
%
f
ea
tu
r
e
s
im
ilar
ity
a
n
d
2
0
%
n
o
n
-
s
im
ilar
ity
.
As
s
h
o
wn
in
Fig
u
r
e
6
(
a)
E
f
f
icien
tNetB
3
ac
h
iev
ed
8
0
%
s
im
ilar
ity
m
atch
es
an
d
2
0
%
n
o
n
-
s
im
ilar
ity
m
atch
es,
co
r
r
esp
o
n
d
in
g
to
a
p
e
r
f
ec
t
1
0
0
% c
lass
if
icatio
n
ac
cu
r
ac
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.
15
,
No
.
5
,
Octo
b
er
20
26
:
4
4
0
3
-
4
4
1
1
4408
I
n
Fig
u
r
e
6
(
b
)
,
R
esNet5
0
V2
r
ec
o
r
d
ed
8
5
%
s
im
ilar
ity
m
atch
es
an
d
1
5
%
n
o
n
-
s
im
ilar
it
y
m
atch
es,
y
ield
in
g
a
9
0
%
ac
c
u
r
ac
y
r
at
e,
wh
ile
Fig
u
r
e
6
(
c)
I
n
ce
p
tio
n
V3
o
b
tain
ed
6
0
%
s
im
ilar
ity
m
atch
es
an
d
4
0
%
non
-
s
im
ilar
ity
m
atch
es,
r
esu
l
tin
g
in
8
0
%
ac
cu
r
ac
y
.
T
h
ese
r
esu
lts
h
ig
h
lig
h
t
E
f
f
icien
tN
etB
3
’
s
ex
ce
p
tio
n
al
p
er
f
o
r
m
an
ce
,
as b
o
th
R
esNet5
0
V2
an
d
I
n
ce
p
tio
n
V3
f
ell
s
h
o
r
t
o
f
th
e
r
eq
u
ir
ed
ac
c
u
r
ac
y
lev
e
l.
T
ab
le
1
p
r
esen
ts
th
e
p
er
f
o
r
m
an
ce
e
v
alu
atio
n
of
th
e
th
r
ee
C
NN
alg
o
r
ith
m
m
o
d
els,
b
ased
on
th
e
p
e
r
ce
n
tag
e
of
s
im
ilar
ity
an
d
non
-
s
im
ilar
ity
b
etwe
en
t
h
e
p
r
ed
icted
o
u
tp
u
ts
an
d
th
e
ac
t
u
a
l
o
u
tp
u
ts
f
o
r
each
test
ca
s
e.
T
h
e
r
esu
lts
in
d
icate
th
at
th
e
E
f
f
icien
tNetB
3
m
o
d
el
ac
h
iev
ed
h
ig
h
er
im
ag
e
cla
s
s
if
icatio
n
ac
cu
r
ac
y
co
m
p
ar
e
d
to
th
e
o
th
er
two
m
o
d
els,
d
em
o
n
s
tr
atin
g
its
ef
f
e
ctiv
en
ess
in
th
is
task
.
(
a)
(
b
)
(
c)
Fig
u
r
e
6.
C
lass
if
icatio
n
ac
cu
r
a
cy
test
ca
s
e
2
f
o
r
th
e
th
r
ee
C
NN
m
o
d
els
:
(
a)
E
f
f
icien
tNetB
3
,
(
b
)
R
esNet5
0
V2
,
an
d
(
c)
I
n
ce
p
tio
n
V3
T
ab
le
1.
Su
m
m
a
r
y
of
th
e
p
er
f
o
r
m
an
ce
a
n
aly
s
is
of
th
e
th
r
ee
C
NN
alg
o
r
ith
m
m
o
d
els
C
N
N
a
l
g
o
r
i
t
h
m
mo
d
e
l
Te
st
c
a
s
e
n
o
.
1
Te
st
c
a
s
e
n
o
.
2
P
e
r
c
e
n
t
a
g
e
of
si
mi
l
a
r
i
t
y
(
8
5
%)
P
e
r
c
e
n
t
a
g
e
of
n
o
n
-
si
m
i
l
a
r
i
t
y
(
1
5
%)
P
e
r
c
e
n
t
a
g
e
of
si
mi
l
a
r
i
t
y
(
8
0
%)
P
e
r
c
e
n
t
a
g
e
of
n
o
n
-
si
m
i
l
a
r
i
t
y
(
2
0
%)
Ef
f
i
c
i
e
n
t
N
e
t
B
3
85
15
80
20
R
e
sN
e
t
5
0
v
2
80
20
85
15
I
n
c
e
p
t
i
o
n
V
3
75
25
60
40
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
P
erfo
r
ma
n
ce
a
n
a
lysi
s
o
f c
o
n
v
o
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
-
b
a
s
e
d
ima
g
e
cla
s
s
ifier
s
fo
r
…
(
A
n
jela
C
.
To
len
tin
o
)
4409
T
h
e
r
esu
lts
f
r
o
m
b
o
t
h
s
im
ilar
ity
test
s
s
h
o
w
a
clea
r
p
atter
n
—
E
f
f
icien
tNetB
3
co
n
s
is
ten
tly
d
eliv
er
ed
th
e
m
o
s
t
ac
c
u
r
ate
class
if
icatio
n
s
,
ac
h
iev
i
n
g
a
p
er
f
ec
t
1
0
0
%
ac
cu
r
ac
y
in
b
o
th
ca
s
es.
T
h
is
m
ea
n
s
th
at
r
eg
ar
d
less
of
wh
eth
er
th
e
s
im
ilar
ity
r
atio
was
85%
–
15%
or
80%
–
2
0
%,
th
e
m
o
d
el
was
ab
le
to
co
r
r
ec
tly
id
en
tify
ev
e
r
y
f
a
r
m
lan
d
im
ag
e
.
R
esNet5
0
V2
also
p
er
f
o
r
m
ed
well,
with
ac
cu
r
ac
y
s
co
r
es
of
80%
an
d
9
0
%,
b
u
t
its
r
esu
lts
wer
e
less
co
n
s
is
ten
t
.
T
h
is
v
ar
iatio
n
m
ig
h
t
be
b
ec
a
u
s
e,
wh
ile
its
r
esid
u
al
lear
n
in
g
ap
p
r
o
ac
h
is
g
r
ea
t
at
m
an
ag
in
g
d
ee
p
lay
e
r
s
an
d
av
o
id
in
g
tr
ain
in
g
is
s
u
es,
it
can
s
o
m
etim
es
s
tr
u
g
g
le
w
h
en
th
e
d
i
f
f
er
en
ce
s
b
etwe
en
im
ag
e
class
es
ar
e
s
u
b
tle
—
s
o
m
eth
in
g
q
u
ite
co
m
m
o
n
in
f
ar
m
lan
d
m
ap
p
in
g
[
2
4
]
.
I
n
ce
p
tio
n
V3
h
ad
t
h
e
lo
west
ac
cu
r
ac
y
in
th
is
s
tu
d
y
,
with
75%
an
d
8
0
%.
I
ts
s
tr
en
g
t
h
lies
in
ca
p
tu
r
in
g
p
atter
n
s
at
m
u
ltip
le
s
ca
les,
wh
ich
wo
r
k
s
well
in
m
an
y
im
ag
e
class
if
icatio
n
p
r
o
b
lem
s
,
b
u
t
in
th
is
ca
s
e,
th
e
co
m
p
lex
ity
of
f
ar
m
lan
d
tex
tu
r
es
m
ay
h
av
e
lim
ited
its
ab
ilit
y
to
p
er
f
o
r
m
as
s
tr
o
n
g
ly
as
th
e
o
th
er
two
m
o
d
els.
E
f
f
icien
tNetB
3
o
u
tp
er
f
o
r
m
s
R
esNet5
0
v
2
an
d
I
n
ce
p
tio
n
V3
by
em
p
lo
y
in
g
a
co
m
p
o
u
n
d
s
ca
lin
g
s
tr
ateg
y
th
at
o
p
tim
izes
d
ep
th
,
wid
th
,
a
n
d
i
n
p
u
t
r
eso
lu
tio
n
f
o
r
im
p
r
o
v
ed
f
ea
tu
r
e
ex
tr
ac
tio
n
.
T
h
e
ar
c
h
itectu
r
e
en
h
an
ce
s
th
e
ca
p
tu
r
e
of
s
p
atial
p
atter
n
s
in
f
ar
m
lan
d
im
a
g
er
y
,
cr
u
cial
f
o
r
ac
cu
r
ate
class
if
icatio
n
,
wh
ile
ad
v
an
ce
d
f
ea
tu
r
e
r
ec
alib
r
atio
n
m
ec
h
an
is
m
s
allo
w
f
o
r
b
etter
f
o
c
u
s
on
r
ele
v
an
t
ar
ea
s
an
d
n
o
is
e
r
ed
u
ctio
n
.
As
r
ef
lecte
d
in
th
e
r
esu
lts
,
E
f
f
icien
tNetB
3
d
em
o
n
s
tr
ated
g
r
ea
te
r
co
n
s
is
ten
cy
a
n
d
s
tab
ilit
y
ac
r
o
s
s
d
if
f
e
r
en
t
s
i
m
ilar
ity
co
n
d
itio
n
s
,
in
d
icatin
g
s
tr
o
n
g
er
g
en
er
aliz
atio
n
ca
p
a
b
ilit
y
an
d
r
o
b
u
s
tn
ess
.
T
h
ese
ch
ar
ac
ter
is
tics
m
ak
e
it
p
a
r
ticu
lar
ly
s
u
itab
le
f
o
r
p
r
ec
is
io
n
lan
d
m
a
p
p
in
g
a
n
d
a
g
r
icu
ltu
r
al
m
o
n
ito
r
in
g
ap
p
licatio
n
s
.
T
h
is
r
esear
ch
s
h
o
ws
how
d
ee
p
lear
n
in
g
can
m
ak
e
a
r
ea
l
d
if
f
er
en
ce
in
ag
r
ic
u
ltu
r
e
[
2
5
]
.
By
test
in
g
th
r
ee
well
-
k
n
o
wn
C
NN
m
o
d
e
ls
,
we
f
o
u
n
d
th
at
E
f
f
icien
tNet
B
3
s
to
o
d
out
as
th
e
m
o
s
t
r
eliab
le
f
o
r
class
if
y
in
g
f
ar
m
lan
d
im
a
g
es
[
2
6
]
.
I
ts
ab
ilit
y
to
ca
p
tu
r
e
f
in
e
d
etails
in
d
r
o
n
e
im
ag
er
y
m
ak
es
it
esp
ec
ially
u
s
ef
u
l
f
o
r
s
u
p
p
o
r
tin
g
task
s
lik
e
cr
o
p
m
o
n
ito
r
in
g
a
n
d
lan
d
m
an
ag
e
m
en
t.
In
th
is
way
,
th
e
s
tu
d
y
not
o
n
ly
h
ig
h
lig
h
ts
th
e
s
tr
en
g
th
s
of
E
f
f
icien
tNetB
3
but
also
o
f
f
er
s
p
r
ac
tical
in
s
i
g
h
ts
th
at
can
h
elp
f
ar
m
er
s
a
n
d
r
esear
ch
e
r
s
u
s
e
ar
tific
ial
in
tellig
en
ce
to
o
ls
m
o
r
e
ef
f
ec
tiv
ely
in
p
r
ec
is
io
n
a
g
r
i
cu
ltu
r
e
[
2
4
]
.
E
f
f
icien
tNetB
3
s
h
o
ws
h
ig
h
p
er
f
o
r
m
a
n
ce
in
f
ar
m
la
n
d
im
a
g
e
class
if
icatio
n
,
but
f
u
tu
r
e
r
e
s
ea
r
ch
can
im
p
r
o
v
e
p
r
ec
is
io
n
lan
d
m
ap
p
i
n
g
s
y
s
tem
s
th
r
o
u
g
h
s
ev
er
al
a
v
en
u
es.
I
n
teg
r
atin
g
m
u
ltimo
d
al
d
ata
s
o
u
r
ce
s
lik
e
s
atellite
im
ag
er
y
,
d
r
o
n
e
im
ag
es,
an
d
g
r
o
u
n
d
s
en
s
o
r
d
ata
co
u
ld
en
h
an
ce
class
if
icatio
n
ac
cu
r
ac
y
by
p
r
o
v
id
in
g
co
m
p
r
eh
e
n
s
iv
e
s
p
atial
an
d
e
n
v
ir
o
n
m
en
tal
co
n
tex
t.
Utilizin
g
tr
an
s
f
er
lear
n
in
g
f
r
o
m
ex
ten
s
iv
e
ag
r
icu
ltu
r
a
l
d
atasets
m
ay
f
u
r
th
er
o
p
tim
ize
m
o
d
el
g
en
er
aliza
tio
n
an
d
p
e
r
f
o
r
m
an
ce
s
tab
ilit
y
ac
r
o
s
s
v
a
r
io
u
s
r
eg
io
n
s
an
d
cr
o
p
ty
p
es.
I
n
co
r
p
o
r
atin
g
ex
p
lai
n
ab
ilit
y
m
eth
o
d
s
,
s
u
ch
as
g
r
ad
ien
t
-
weig
h
te
d
class
ac
tiv
atio
n
m
ap
p
in
g
(
Gr
ad
-
C
AM
)
,
is
es
s
en
tial
to
b
u
ild
tr
u
s
t
am
o
n
g
s
tak
eh
o
l
d
er
s
by
r
ev
ea
lin
g
wh
ich
im
ag
e
ar
ea
s
in
f
lu
en
ce
class
if
icatio
n
d
ec
is
io
n
s
.
L
astl
y
,
in
teg
r
atio
n
with
g
eo
g
r
a
p
h
ic
in
f
o
r
m
atio
n
s
y
s
tem
s
(
GI
S)
wi
ll
en
ab
le
ac
tio
n
ab
le
an
aly
s
is
an
d
r
eso
u
r
ce
p
lan
n
in
g
in
a
g
r
icu
ltu
r
e.
Fu
tu
r
e
r
esear
ch
s
h
o
u
l
d
f
o
cu
s
on
th
ese
f
o
u
r
asp
ec
ts
to
en
h
an
ce
b
o
th
tech
n
ical
p
er
f
o
r
m
a
n
ce
a
n
d
p
r
ac
tical
ap
p
licatio
n
in
p
r
e
cisi
o
n
ag
r
icu
ltu
r
e
.
Ad
d
itio
n
al
ev
alu
atio
n
m
et
r
ics,
in
clu
d
in
g
p
r
ec
is
io
n
,
r
ec
all,
a
n
d
F1
-
s
co
r
e,
will
be
in
co
r
p
o
r
ated
in
f
u
t
u
r
e
r
esear
ch
to
f
u
r
th
er
s
tr
en
g
t
h
en
p
er
f
o
r
m
an
ce
ass
ess
m
en
t
an
d
r
ea
l
-
wo
r
ld
ap
p
licab
ilit
y
.
4.
CO
NCLU
SI
O
N
T
h
e
p
er
f
o
r
m
an
ce
a
n
aly
s
is
of
th
e
th
r
ee
C
NN
alg
o
r
ith
m
m
o
d
els
E
f
f
icien
tNetB
3
,
R
esNet5
0
V2
,
an
d
I
n
ce
p
tio
n
V3
r
ev
ea
led
th
at
E
f
f
i
cien
tNetB
3
co
n
s
is
ten
tly
d
eliv
er
ed
th
e
h
i
g
h
est
class
if
icatio
n
ac
cu
r
ac
y
ac
r
o
s
s
all
test
ca
s
e
s
.
I
ts
ab
ilit
y
to
ef
f
ec
tiv
ely
ca
p
tu
r
e
f
in
e
-
g
r
ain
ed
f
ea
tu
r
es
in
h
ig
h
-
r
eso
lu
tio
n
im
ag
es
en
ab
led
it
to
o
u
tp
er
f
o
r
m
th
e
o
th
er
two
ar
ch
itectu
r
es
in
b
o
t
h
s
im
ila
r
ity
an
d
non
-
s
im
ilar
ity
class
if
icatio
n
s
.
W
h
ile
R
esNet5
0
V2
an
d
I
n
ce
p
tio
n
V
3
ac
h
iev
ed
r
esp
ec
tab
le
r
esu
lts
,
th
eir
ac
cu
r
ac
y
lev
els
f
ell
s
h
o
r
t
of
t
h
e
tar
g
et
th
r
esh
o
ld
s
in
ce
r
tai
n
s
ce
n
ar
io
s
.
T
h
ese
f
in
d
in
g
s
s
u
g
g
est
th
a
t
E
f
f
icien
tNetB
3
is
th
e
m
o
s
t
s
u
itab
le
alg
o
r
ith
m
m
o
d
el
am
o
n
g
th
e
th
r
ee
f
o
r
f
ar
m
lan
d
im
ag
e
class
if
icatio
n
task
s
,
o
f
f
e
r
in
g
a
r
eliab
le
b
a
lan
ce
of
p
r
ec
is
io
n
,
ef
f
icien
cy
,
a
n
d
r
o
b
u
s
tn
ess
in
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
.
ACK
NO
WL
E
DG
M
E
N
T
S
T
h
e
au
th
o
r
s
wo
u
ld
lik
e
to
ex
p
r
ess
th
eir
s
in
ce
r
e
g
r
atitu
d
e
to
all
in
d
iv
id
u
als
an
d
o
r
g
a
n
iza
tio
n
s
wh
o
co
n
tr
ib
u
ted
to
th
e
s
u
cc
ess
f
u
l
c
o
m
p
letio
n
of
th
is
r
esear
ch
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
no
f
u
n
d
in
g
in
v
o
lv
ed
.
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
t
r
ib
u
to
r
R
o
les
T
a
x
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
i
d
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
.
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.
15
,
No
.
5
,
Octo
b
er
20
26
:
4
4
0
3
-
4
4
1
1
4410
Na
m
e
of
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
An
jela
C.
T
o
len
tin
o
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
An
az
el
P.
Gam
illa
✓
✓
✓
✓
✓
✓
✓
Am
ir
L
ed
esm
a
✓
✓
✓
✓
✓
✓
✓
✓
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
OF
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
no
co
n
f
lict
of
in
t
er
est.
DATA
AV
AI
L
AB
I
L
I
T
Y
Data
s
u
p
p
o
r
tin
g
th
is
s
tu
d
y
ar
e
av
ailab
le
f
r
o
m
t
h
e
co
r
r
esp
o
n
d
in
g
a
u
th
o
r
,
[
AC
T
]
,
u
p
o
n
r
e
q
u
est
but
ar
e
not
p
u
b
licly
s
h
ar
ed
due
to
p
r
iv
ac
y
r
estrictio
n
s
.
RE
F
E
R
E
NC
E
S
[
1
]
R
e
p
u
b
l
i
c
o
f
t
h
e
P
h
i
l
i
p
p
i
n
e
,
P
h
i
l
i
p
p
i
n
e
S
t
a
t
i
st
i
c
s
A
u
t
h
o
r
i
t
y
,
“
S
e
l
e
c
t
e
d
st
a
t
i
s
t
i
c
s
o
n
a
g
r
i
c
u
l
t
u
r
e
,
”
p
sa
.
g
o
v
.
p
h
,
2
0
2
2
A
c
c
e
sse
d
:
Jan
.
1
4
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s:
/
/
p
sa
.
g
o
v
.
p
h
/
c
o
n
t
e
n
t
/
se
l
e
c
t
e
d
-
s
t
a
t
i
s
t
i
c
s
-
a
g
r
i
c
u
l
t
u
r
e
[
2
]
F
a
r
mo
n
a
u
t
,
“
P
h
i
l
i
p
p
i
n
e
s
l
a
n
d
u
se
:
t
o
p
7
a
g
r
i
c
u
l
t
u
r
e
c
h
a
l
l
e
n
g
e
s
,
”
f
a
r
mo
n
a
u
t
.
c
o
m
.
A
c
c
e
sse
d
:
J
a
n
.
1
5
,
2
0
2
4
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
f
a
r
m
o
n
a
u
t
.
c
o
m
/
a
s
i
a
/
p
h
i
l
i
p
p
i
n
e
s
-
l
a
n
d
-
u
se
-
t
o
p
-
7
-
a
g
r
i
c
u
l
t
u
r
e
-
c
h
a
l
l
e
n
g
e
s
.
[
3
]
W
.
L
i
u
e
t
a
l
.
,
“
F
a
r
m
l
a
n
d
p
a
r
c
e
l
ma
p
p
i
n
g
i
n
m
o
u
n
t
a
i
n
a
r
e
a
s u
s
i
n
g
t
i
me
-
ser
i
e
s
S
A
R
d
a
t
a
a
n
d
V
H
R
o
p
t
i
c
a
l
i
m
a
g
e
s
,”
Re
m
o
t
S
e
n
s
i
n
g
,
v
o
l
.
12
,
n
o
.
22
,
N
o
v
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
r
s1
2
2
2
3
7
3
3
.
[
4
]
M
.
W
a
q
a
s,
A
.
N
a
s
e
e
m
,
U
.
W
.
H
u
mp
h
r
i
e
s,
P
.
T
.
H
l
a
i
n
g
,
P
.
D
e
c
h
p
i
c
h
a
i
,
a
n
d
A
.
W
a
n
g
w
o
n
g
c
h
a
i
,
“
A
p
p
l
i
c
a
t
i
o
n
s o
f
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
i
n
a
g
r
i
c
u
l
t
u
r
e
:
a
c
o
m
p
r
e
h
e
n
si
v
e
r
e
v
i
e
w
,
”
G
r
e
e
n
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
S
u
st
a
i
n
a
b
i
l
i
t
y
,
v
o
l
.
3
,
n
o
.
3
,
J
u
l
.
2
0
2
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
g
r
e
t
s.
2
0
2
5
.
1
0
0
1
9
9
.
[
5
]
Y
.
Le
C
u
n
,
Y
.
B
e
n
g
i
o
,
a
n
d
G
.
H
i
n
t
o
n
,
“
D
e
e
p
l
e
a
r
n
i
n
g
,
”
N
a
t
u
re
,
v
o
l
.
5
2
1
,
n
o
.
7
5
5
3
,
p
p
.
4
3
6
–
4
4
4
,
2
0
1
5
,
d
o
i
:
1
0
.
1
0
3
8
/
n
a
t
u
r
e
1
4
5
3
9
.
[
6
]
J.
S
o
n
g
,
S
.
G
a
o
,
Y
.
Z
h
u
,
a
n
d
C
.
M
a
,
“
A
s
u
r
v
e
y
o
f
r
e
mo
t
e
se
n
si
n
g
i
m
a
g
e
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
se
d
o
n
C
N
N
s,
”
Bi
g
Ea
rt
h
D
a
t
a
,
v
o
l
.
3
,
n
o
.
3
,
p
p
.
2
3
2
–
2
5
4
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
8
0
/
2
0
9
6
4
4
7
1
.
2
0
1
9
.
1
6
5
7
7
2
0
.
[
7
]
A
.
K
a
mi
l
a
r
i
s
a
n
d
F
.
X
.
P
.
-
B
o
l
d
ú
,
“
D
e
e
p
l
e
a
r
n
i
n
g
i
n
a
g
r
i
c
u
l
t
u
r
e
:
a
s
u
r
v
e
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
ri
c
u
l
t
u
re
,
v
o
l
.
1
4
7
,
p
p
.
7
0
–
9
0
,
A
p
r
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
a
g
.
2
0
1
8
.
0
2
.
0
1
6
.
[
8
]
L.
P
.
O
s
c
o
e
t
a
l
.
,
“
A
r
e
v
i
e
w
o
n
d
e
e
p
l
e
a
r
n
i
n
g
i
n
U
A
V
r
e
mo
t
e
s
e
n
s
i
n
g
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
A
p
p
l
i
e
d
Ea
rt
h
O
b
serv
a
t
i
o
n
a
n
d
G
e
o
i
n
f
o
rm
a
t
i
o
n
,
v
o
l
.
1
0
2
,
p
p
.
1
3
5
–
1
5
0
,
M
a
r
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
a
g
.
2
0
2
1
.
1
0
2
4
5
6
.
[
9
]
M
.
P
e
n
g
e
t
a
l
.
,
“
C
r
o
p
m
o
n
i
t
o
r
i
n
g
u
s
i
n
g
r
e
m
o
t
e
se
n
si
n
g
l
a
n
d
u
s
e
a
n
d
l
a
n
d
c
h
a
n
g
e
d
a
t
a
:
c
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
f
d
e
e
p
l
e
a
r
n
i
n
g
met
h
o
d
s
u
s
i
n
g
p
r
e
-
t
r
a
i
n
e
d
C
N
N
m
o
d
e
l
s,”
Bi
g
D
a
t
a
Re
se
a
r
c
h
,
v
o
l
.
3
6
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
d
r
.
2
0
2
4
.
1
0
0
4
4
8
.
[
1
0
]
A
.
P
a
t
n
a
l
a
,
M
.
G
.
S
c
h
u
l
t
z
,
a
n
d
J.
G
a
l
l
,
“
B
E
R
T
B
i
-
m
o
d
a
l
se
l
f
-
su
p
e
r
v
i
s
e
d
l
e
a
r
n
i
n
g
f
o
r
c
r
o
p
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
si
n
g
se
n
t
i
n
e
l
-
2
a
n
d
p
l
a
n
e
t
s
c
o
p
e
,”
Fro
n
t
i
e
rs
i
n
Re
m
o
t
e
S
e
n
si
n
g
,
v
o
l
.
6
,
2
0
2
5
,
d
o
i
:
1
0
.
3
3
8
9
/
f
r
sen
.
2
0
2
5
.
1
5
5
5
8
8
7
.
[
1
1
]
R
.
Es
p
i
n
e
l
,
G
.
H
.
-
F
r
a
n
c
o
,
J.
L.
R
.
G
a
r
c
í
a
,
a
n
d
P
.
E
.
-
P
a
n
c
h
a
n
a
,
“
A
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
i
n
a
g
r
i
c
u
l
t
u
r
a
l
m
a
p
p
i
n
g
:
a
r
e
v
i
e
w
,”
Ag
ri
c
u
l
t
u
r
e
,
v
o
l
.
1
4
,
n
o
.
7
,
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
a
g
r
i
c
u
l
t
u
r
e
1
4
0
7
1
0
7
1
.
[
1
2
]
M
.
E
.
S
a
k
k
a
,
M
.
I
v
a
n
o
v
i
c
i
,
L
.
C
h
a
a
r
i
,
a
n
d
J.
M
o
t
h
e
,
“
A
r
e
v
i
e
w
o
f
C
N
N
a
p
p
l
i
c
a
t
i
o
n
s
i
n
smar
t
a
g
r
i
c
u
l
t
u
r
e
u
si
n
g
mu
l
t
i
m
o
d
a
l
d
a
t
a
,
”
S
e
n
so
rs
,
v
o
l
.
2
5
,
n
o
.
2
,
2
0
2
5
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
5
0
2
0
4
7
2
.
[
1
3
]
K
.
H
e
,
X
.
Z
h
a
n
g
,
S
.
R
e
n
,
a
n
d
J.
S
u
n
,
“
I
d
e
n
t
i
t
y
ma
p
p
i
n
g
s
i
n
d
e
e
p
r
e
s
i
d
u
a
l
n
e
t
w
o
r
k
s
,
”
i
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
–
E
C
C
V
2
0
1
6
,
2
0
1
6
,
p
p
.
6
3
0
–
6
4
5
.
[
1
4
]
C
.
S
z
e
g
e
d
y
,
V
.
V
a
n
h
o
u
c
k
e
,
S
.
I
o
f
f
e
,
J.
S
h
l
e
n
s,
a
n
d
Z.
W
o
j
n
a
,
“
R
e
t
h
i
n
k
i
n
g
t
h
e
i
n
c
e
p
t
i
o
n
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
c
o
mp
u
t
e
r
v
i
s
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
,
2
0
1
6
,
p
p
.
2
8
1
8
–
2
8
2
6
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
3
0
8
.
[
1
5
]
M
.
Ta
n
a
n
d
Q
.
V
.
Le
,
“
Ef
f
i
c
i
e
n
t
N
e
t
:
r
e
t
h
i
n
k
i
n
g
mo
d
e
l
sc
a
l
i
n
g
f
o
r
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
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
3
6
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
R
e
se
a
rc
h
,
2
0
1
9
,
p
p
.
6
1
0
5
–
6
1
1
4
.
[
1
6
]
V
.
M
a
z
z
i
a
,
A
.
K
h
a
l
i
q
,
a
n
d
M
.
C
h
i
a
b
e
r
g
e
,
“
I
mp
r
o
v
e
me
n
t
i
n
l
a
n
d
c
o
v
e
r
a
n
d
c
r
o
p
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
se
d
o
n
t
e
m
p
o
r
a
l
f
e
a
t
u
r
e
s
l
e
a
r
n
i
n
g
f
r
o
m
se
n
t
i
n
e
l
-
2
d
a
t
a
u
s
i
n
g
r
e
c
u
r
r
e
n
t
-
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
(
R
-
C
N
N
)
,
”
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
0
,
n
o
.
1
,
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
0
0
1
0
2
3
8
.
[
1
7
]
F
.
B
a
l
a
n
d
F
.
K
a
y
a
a
l
p
,
“
R
e
v
i
e
w
o
f
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
s
i
n
a
g
r
i
c
u
l
t
u
r
e
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
A
d
v
a
n
c
e
d
Re
se
a
rc
h
e
s
a
n
d
E
n
g
i
n
e
e
ri
n
g
J
o
u
rn
a
l
,
v
o
l
.
5
,
n
o
.
2
,
p
p
.
3
0
9
–
3
2
3
,
2
0
2
1
,
d
o
i
:
1
0
.
3
5
8
6
0
/
i
a
r
e
j
.
8
4
8
4
5
8
.
[
1
8
]
S
.
M
a
sc
a
r
e
n
h
a
s
a
n
d
M
.
A
g
a
r
w
a
l
,
“
A
c
o
mp
a
r
i
s
o
n
b
e
t
w
e
e
n
V
G
G
1
6
,
V
G
G
1
9
a
n
d
R
e
sN
e
t
5
0
a
r
c
h
i
t
e
c
t
u
r
e
f
r
a
m
e
w
o
r
k
s
f
o
r
i
m
a
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
I
E
EE
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
D
i
s
ru
p
t
i
v
e
T
e
c
h
n
o
l
o
g
i
e
s
f
o
r
M
u
l
t
i
-
D
i
s
c
i
p
l
i
n
a
ry
R
e
se
a
rc
h
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
C
EN
T
C
O
N
2
0
2
1
,
2
0
2
1
,
p
p
.
9
6
–
99
,
d
o
i
:
1
0
.
1
1
0
9
/
C
EN
TC
O
N
5
2
3
4
5
.
2
0
2
1
.
9
6
8
7
9
4
4
.
[
1
9
]
Y
.
C
h
e
n
e
t
a
l
.
,
“
C
l
a
ssi
f
i
c
a
t
i
o
n
o
f
l
u
n
g
s
i
n
f
e
c
t
e
d
C
O
V
I
D
-
1
9
i
m
a
g
e
s
b
a
se
d
o
n
i
n
c
e
p
t
i
o
n
-
R
e
sN
e
t
,
”
C
o
m
p
u
t
e
r
M
e
t
h
o
d
s
a
n
d
Pro
g
ra
m
s
i
n
B
i
o
m
e
d
i
c
i
n
e
,
v
o
l
.
2
2
5
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
m
p
b
.
2
0
2
2
.
1
0
7
0
5
3
.
[
2
0
]
H
.
A
mi
n
,
A
.
D
a
r
w
i
s
h
,
A
.
E.
H
a
ssa
n
i
e
n
,
a
n
d
M
.
S
o
l
i
m
a
n
,
“
En
d
-
to
-
e
n
d
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
f
o
r
c
o
r
n
l
e
a
f
d
i
se
a
s
e
c
l
a
ssi
f
i
c
a
t
i
o
n
,”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
3
1
1
0
3
–
3
1
1
1
5
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
1
5
9
6
7
8
.
[
2
1
]
A
.
N
o
w
a
k
o
w
s
k
i
e
t
a
l
.
,
“
C
r
o
p
t
y
p
e
m
a
p
p
i
n
g
b
y
u
s
i
n
g
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
p
p
l
i
e
d
E
a
r
t
h
O
b
serv
a
t
i
o
n
a
n
d
G
e
o
i
n
f
o
rm
a
t
i
o
n
,
v
o
l
.
9
8
,
J
u
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
a
g
.
2
0
2
1
.
1
0
2
3
1
3
.
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
P
erfo
r
ma
n
ce
a
n
a
lysi
s
o
f c
o
n
v
o
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
-
b
a
s
e
d
ima
g
e
cla
s
s
ifier
s
fo
r
…
(
A
n
jela
C
.
To
len
tin
o
)
4411
[
2
2
]
S
.
J.
P
a
n
a
n
d
Q
.
Y
a
n
g
,
“
A
su
r
v
e
y
o
n
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
I
EE
E
T
r
a
n
sa
c
t
i
o
n
s
o
n
K
n
o
w
l
e
d
g
e
a
n
d
D
a
t
a
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
2
2
,
n
o
.
1
0
,
p
p
.
1
3
4
5
–
1
3
5
9
,
O
c
t
.
2
0
1
0
,
d
o
i
:
1
0
.
1
1
0
9
/
T
K
D
E.
2
0
0
9
.
1
9
1
.
[
2
3
]
C
.
S
h
i
,
J.
S
u
n
,
a
n
d
L
.
W
a
n
g
,
“
H
y
p
e
r
sp
e
c
t
r
a
l
i
ma
g
e
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
s
e
d
o
n
sp
e
c
t
r
a
l
m
u
l
t
i
s
c
a
l
e
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
,
”
Re
m
o
t
e
S
e
n
s
i
n
g
,
v
o
l
.
1
4
,
n
o
.
8
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
r
s
1
4
0
8
1
9
5
1
.
[
2
4
]
L.
L
i
,
Z.
Ta
n
,
a
n
d
X
.
H
a
n
,
“
A
n
i
m
p
r
o
v
e
d
Ef
f
i
c
i
e
n
t
N
e
t
m
o
d
e
l
a
n
d
i
t
s
a
p
p
l
i
c
a
t
i
o
n
s
i
n
p
n
e
u
mo
n
i
a
i
ma
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
En
g
i
n
e
e
ri
n
g
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
Re
v
i
e
w
,
v
o
l
.
1
5
,
n
o
.
6
,
p
p
.
4
9
–
5
4
,
2
0
2
2
,
d
o
i
:
1
0
.
2
5
1
0
3
/
j
e
s
t
r
.
1
5
6
.
0
7
.
[
2
5
]
F
.
A
d
n
a
n
,
M
.
J
.
A
w
a
n
,
A
.
M
a
h
m
o
u
d
,
H
.
N
o
b
a
n
e
e
,
A
.
Y
a
s
i
n
,
a
n
d
A
.
M
.
Za
i
n
,
“
Ef
f
i
c
i
e
n
t
N
e
t
B
3
-
a
d
a
p
t
i
v
e
a
u
g
m
e
n
t
e
d
d
e
e
p
l
e
a
r
n
i
n
g
(
A
A
D
L)
f
o
r
mu
l
t
i
-
c
l
a
ss
p
l
a
n
t
d
i
s
e
a
s
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
,”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
8
5
4
2
6
–
8
5
4
4
0
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
0
3
1
3
1
.
[
2
6
]
S
.
D
.
T
o
mm
a
so
,
S
.
W
a
n
g
,
a
n
d
D
.
B
.
Lo
b
e
l
l
,
“
C
o
m
b
i
n
i
n
g
G
ED
I
a
n
d
s
e
n
t
i
n
e
l
-
2
f
o
r
w
a
l
l
-
to
-
w
a
l
l
ma
p
p
i
n
g
o
f
t
a
l
l
a
n
d
s
h
o
r
t
c
r
o
p
s,”
En
v
i
r
o
n
m
e
n
t
a
l
R
e
se
a
rc
h
L
e
t
t
e
rs
,
v
o
l
.
1
6
,
n
o
.
1
2
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
8
-
9
3
2
6
/
a
c
3
5
8
c
.
B
I
O
G
RAP
H
I
E
S
OF
AUTH
O
RS
Dr
.
Anjela
C.
To
lenti
n
o
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
a
n
d
h
e
a
d
o
f
t
h
e
De
p
a
rtme
n
t
o
f
In
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
a
t
t
h
e
Co
ll
e
g
e
o
f
En
g
i
n
e
e
rin
g
,
Ce
n
tr
a
l
Lu
z
o
n
S
tate
Un
i
v
e
rsity
,
S
c
ien
c
e
Cit
y
o
f
M
u
ñ
o
z
,
Nu
e
v
a
Ecija,
P
h
il
ip
p
in
e
s.
S
h
e
e
a
rn
e
d
h
e
r
Do
c
to
ra
te
in
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
(DIT)
fr
o
m
t
h
e
Un
iv
e
rsity
o
f
t
h
e
Co
r
d
il
lera
s,
Ba
g
u
io
Cit
y
,
P
h
i
li
p
p
in
e
s.
He
r
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
n
e
two
r
k
in
g
,
so
ftwa
re
d
e
v
e
lo
p
m
e
n
t
,
a
p
p
li
c
a
ti
o
n
s,
a
n
d
d
a
ta
sc
ien
c
e
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
n
j
e
lato
len
ti
n
o
@c
lsu
.
e
d
u
.
p
h
.
Dr
.
Ana
z
e
l
P.
G
a
m
il
la
h
o
ld
s
a
Do
c
to
ra
te
in
In
f
o
rm
a
ti
o
n
Tec
h
n
o
lo
g
y
(DIT)
fro
m
th
e
Un
i
v
e
rsity
o
f
t
h
e
Co
r
d
il
ler
a
s
(UC
-
BCF
),
P
h
il
i
p
p
i
n
e
s.
A
n
a
ss
istan
t
p
r
o
fe
ss
o
r
o
f
th
e
De
p
a
rtme
n
t
o
f
In
f
o
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
,
C
o
ll
e
g
e
o
f
En
g
in
e
e
ri
n
g
,
f
o
rm
e
r
c
h
ief
o
f
M
a
n
a
g
e
m
e
n
t
In
fo
rm
a
ti
o
n
S
y
ste
m
s
Offic
e
a
t
C
e
n
tral
Lu
z
o
n
S
tate
Un
iv
e
rsity
(C
LS
U)
a
n
d
a
De
p
a
rtme
n
t
o
f
In
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
a
n
d
Co
m
m
u
n
ica
ti
o
n
s
Tec
h
n
o
l
o
g
y
(
DICT
-
ILCDB)
train
e
r.
He
r
c
u
rre
n
t
re
se
a
rc
h
i
n
tere
sts
in
c
l
u
d
e
c
o
m
p
u
ter
n
e
two
rk
s,
S
DN
,
c
y
b
e
r
se
c
u
rit
y
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
p
g
a
m
il
la@
c
lsu
.
e
d
u
.
p
h
.
Am
ir
Le
d
e
sm
a
is
c
u
rre
n
tl
y
p
u
rs
u
in
g
a
m
a
ste
r’s
d
e
g
re
e
in
In
fo
rm
a
ti
o
n
Tec
h
n
o
l
o
g
y
a
t
De
La
S
a
ll
e
Un
iv
e
rsity
(DLS
U),
P
h
il
i
p
p
i
n
e
s,
with
3
3
a
c
a
d
e
m
ic
u
n
it
s
c
o
m
p
lete
d
.
He
is
a
n
i
n
stru
c
t
o
r
a
t
th
e
De
p
a
rtme
n
t
o
f
I
n
fo
rm
a
ti
o
n
Tec
h
n
o
lo
g
y
,
Co
l
leg
e
o
f
En
g
i
n
e
e
rin
g
,
Ce
n
tral
Lu
z
o
n
S
ta
te
Un
iv
e
rsity
(CLS
U)
a
n
d
a
train
e
r
fo
r
v
a
rio
u
s
train
i
n
g
se
m
in
a
rs
fo
r
th
e
De
p
a
rtme
n
t
o
f
In
fo
rm
a
ti
o
n
a
n
d
C
o
m
m
u
n
ica
ti
o
n
s
Tec
h
n
o
lo
g
y
(DICT).
His
a
c
a
d
e
m
ic
a
n
d
p
ro
fe
ss
io
n
a
l
i
n
tere
sts
fo
c
u
s
o
n
a
rti
f
icia
l
in
telli
g
e
n
c
e
,
b
u
sin
e
ss
in
telli
g
e
n
c
e
,
a
n
d
e
n
terp
rise
a
rc
h
it
e
c
tu
re
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
m
irl
e
d
e
sm
a
@c
lsu
.
e
d
u
.
p
h
.
Evaluation Warning : The document was created with Spire.PDF for Python.