I
A
E
S
I
n
t
e
r
n
at
io
n
al
Jou
r
n
al
of
A
r
t
if
ic
ia
l
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
V
ol
.
14
, N
o.
5
,
O
c
to
be
r
2025
, pp.
3724
~
3733
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
14
.i
5
.pp
3724
-
3733
3724
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
O
p
t
i
m
i
z
e
d
c
on
vol
u
t
i
on
n
e
u
r
al
n
e
t
w
or
k
w
i
t
h
an
t
c
ol
on
y
al
gor
i
t
h
m
f
or
ac
c
u
r
at
e
p
l
an
t
d
i
se
ase
d
e
t
e
c
t
i
on
S
h
w
e
t
a V
. B
on
d
r
e
1
, U
m
a Y
ad
av
1
,
V
ip
in
D
. B
on
d
r
e
2
, P
oor
va A
gr
aw
al
3
1
S
c
hool
of
C
om
put
e
r
S
c
i
e
nc
e
a
nd
E
ngi
ne
e
r
i
ng
, S
hr
i
R
a
m
de
oba
ba
C
ol
l
e
ge
of
E
ngi
ne
e
r
i
ng a
nd M
a
na
ge
m
e
nt
,
R
a
m
de
oba
ba
U
ni
ve
r
s
i
t
y,
N
a
gpur
,
I
ndi
a
2
D
e
pa
r
t
m
e
nt
of
E
l
e
c
t
r
oni
c
s
a
nd
T
e
l
e
c
om
m
uni
c
a
t
i
on
, Y
e
s
hw
a
nt
r
a
o C
ha
va
n
C
ol
l
e
ge
of
E
ngi
ne
e
r
i
ng
,
N
a
gpur
, I
ndi
a
3
S
ym
bi
os
i
s
I
ns
t
i
t
ut
e
of
T
e
c
hnol
ogy, N
a
gpur
C
a
m
pus
S
ym
bi
os
i
s
I
nt
e
r
na
t
i
ona
l
(
D
e
e
m
e
d U
ni
ve
r
s
i
t
y)
, P
une
, I
ndi
a
A
r
t
ic
le
I
n
f
o
A
B
S
T
R
A
C
T
A
r
ti
c
le
h
is
to
r
y
:
R
e
c
e
iv
e
d
J
ul
26, 2024
R
e
vi
s
e
d
J
ul
10, 2025
A
c
c
e
pt
e
d
A
ug 6, 2025
In
India,
agriculture
is
the
primary
source
of
income
for
half
the
people.
Even
in
situations
of
fast
population
growth,
agriculture
s
upplies
nourishment
for
all
people.
To
provide
food
for
the
entire
populatio
n,
it
is
advised
to
detect
plant
diseases
at
an
early
stage.
Plant
leaf
diseas
es
are
recognized
using
images
of
the
affected
leaves.
Deep
learning
(DL)
re
search
seems
to
offer
several
opportunities
for
increased
accuracy.
Ant
colony
optimization
with
convolution
-
neural
-
network
(ACO
-
CNN),
a
ne
w
deep
learning
technique
for
identifying
and
categor
izing
diseases,
is
prese
nted
in
this
article.
Ant
colony
optimization
(ACO)
was
used
to
exami
ne
the
efficacy
of
disease
diagnost
ics
in
plant
leaves.
The
convolut
ion
neural
network
(
CNN
)
classifi
er
is
used
to
remove
texture,
color,
and
leaf
arrangement
geometry
from
the
input
images.
The
ACO
-
CNN
model
outperformed
the
support
vector
machine
(
SVM
)
and
CNN
models
in
terms
of
precision,
recall,
and
accuracy.
CNN'
s
rate
is
81.6%
as
compa
red
to
SVM'
s
80%
accuracy
level.
In
the
“ACO
-
CNN”
approach,
the
F1
-
score,
recall,
and
precision
have
higher
rates
as
comp
ared
to
other
models,
a
nd
the
“F1
-
score”
has
the
highest
rate
compared
with
other
models
since
the
ACO
-
CNN model
has an
accur
acy ra
te of 91
.00%.
K
e
y
w
o
r
d
s
:
A
nt
c
ol
ony a
lg
or
it
hm
C
onvolut
io
n ne
ur
a
l
ne
twor
k
D
e
e
p l
e
a
r
ni
ng
M
a
c
hi
ne
l
e
a
r
ni
ng
P
la
nt
di
s
e
a
s
e
P
la
nt
di
s
e
a
s
e
di
a
gno
s
is
S
m
a
r
t
f
a
r
m
in
g
This is an
open
acce
ss artic
le unde
r the
CC BY
-
SA
license.
C
or
r
e
s
pon
di
n
g A
u
th
or
:
S
hw
e
ta
V
. B
ondr
e
S
c
hool
of
C
om
put
e
r
S
c
ie
nc
e
a
nd E
ngi
ne
e
r
in
g, S
hr
i
R
a
m
de
oba
ba
C
ol
le
ge
of
E
ngi
ne
e
r
in
g a
nd
M
a
na
ge
m
e
nt
R
a
m
de
oba
ba
U
ni
ve
r
s
it
y
N
a
gpur
, I
ndi
a
E
m
a
il
:
s
hw
e
ta
kha
r
a
t1
510@
gm
a
il
.c
om
or
bondr
e
s
v@
r
kne
c
.e
du
1.
I
N
T
R
O
D
U
C
T
I
O
N
I
n
I
ndi
a
,
th
e
a
gr
ic
ul
tu
r
e
in
dus
tr
y
e
m
pl
oye
d
50%
of
th
e
w
o
r
kf
or
c
e
a
nd
m
a
de
up
19.9%
of
th
e
c
ount
r
y'
s
gr
os
s
dome
s
ti
c
pr
oduc
t
(
G
D
P
)
in
2020
–
20
21. T
he
m
o
s
t
r
e
c
e
nt
t
e
c
hni
c
a
l
de
v
e
lo
pm
e
nt
s
m
us
t
be
u
s
e
d
to
pr
om
ot
e
th
e
e
f
f
ic
ie
nt
c
r
op
c
ul
ti
va
ti
on
.
D
ue
to
pl
a
nt
di
s
e
a
s
e
s
a
nd
pe
s
t
da
m
a
ge
, c
r
ops
e
xpe
r
ie
nc
e
s
ig
ni
f
ic
a
nt
lo
s
s
e
s
.
B
e
c
a
us
e
of
pe
s
t
da
m
a
ge
a
nd pla
nt
di
s
e
a
s
e
s
, c
r
ops
e
xpe
r
i
e
nc
e
s
ig
ni
f
ic
a
nt
l
os
s
e
s
. B
y 2050, ther
e
w
il
l
be
9.2
bi
ll
io
n
popula
ti
on
on
th
e
pl
a
ne
t,
a
nd
to
f
ul
f
il
l
th
e
ir
f
ood
r
e
qui
r
e
m
e
nt
s
,
f
ood
pr
oduc
ti
on
w
il
l
ne
e
d
to
r
is
e
by
a
lm
os
t
70%
ut
il
iz
e
d
in
[
1]
.
F
ood
c
r
ops
e
xpe
r
ie
nc
e
s
ig
ni
f
ic
a
nt
lo
s
s
e
s
a
s
a
r
e
s
ul
t
of
unf
a
vor
a
bl
e
w
e
a
th
e
r
,
s
tr
ong
w
in
ds
,
dr
ought,
f
ungi
,
vi
r
us
e
s
,
a
nd
ba
c
te
r
ia
.
70
-
80%
of
a
gr
ic
ul
tu
r
a
l
lo
s
s
e
s
w
or
ld
w
id
e
a
r
e
due
to
pl
a
nt
di
s
e
a
s
e
s
.
T
he
de
v
e
lo
pm
e
nt
of
te
c
hnol
ogi
e
s
ha
s
m
a
de
it
po
s
s
i
bl
e
to
pr
oduc
e
e
nough
f
ood
to
m
e
e
t
s
oc
i
e
ta
l
ne
e
ds
.
H
ow
e
ve
r
,
th
e
s
e
c
ur
it
y
a
nd
s
a
f
e
ty
of
th
e
f
ood
a
nd
th
e
c
r
op
w
e
r
e
ne
ve
r
a
tt
a
in
e
d.
F
a
r
m
e
r
s
e
xpe
r
ie
nc
e
c
ha
ll
e
nge
s
be
c
a
us
e
of
th
in
gs
li
ke
c
li
m
a
te
c
h
a
nge
,
a
de
c
li
ne
in
pol
li
na
to
r
s
,
pl
a
nt
di
s
e
a
s
e
s
,
a
nd
ot
he
r
pr
obl
e
m
s
[
2]
.
T
o
gr
ow
he
a
lt
hy
f
ood,
i
t
is
e
s
s
e
nt
ia
l
to
pr
ot
e
c
t
t
he
pl
a
nt
s
a
ga
in
s
t
di
s
e
a
s
e
s
.
P
ol
li
na
to
r
de
c
li
ne
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
e
d c
onv
ol
ut
io
n ne
ur
al
n
e
tw
or
k
w
it
h ant c
ol
ony
al
gor
it
h
m
f
or
ac
c
ur
at
e
pl
ant
…
(
Shw
e
ta
B
ondr
e
)
3725
c
li
m
a
te
c
ha
nge
,
pl
a
nt
di
s
e
a
s
e
s
,
a
nd
w
a
te
r
qua
li
ty
s
ig
ni
f
ic
a
nt
ly
im
pa
c
t
f
ood
s
e
c
ur
it
y.
A
c
c
or
di
ng
to
th
e
F
ood
a
nd
A
gr
ic
ul
tu
r
e
O
r
ga
ni
z
a
ti
on
(
F
A
O
)
,
pe
s
ts
,
a
nd
di
s
e
a
s
e
s
d
e
s
tr
oy
up
to
40%
of
f
ood
c
r
ops
a
nnua
ll
y,
hi
ghl
ig
ht
in
g
th
e
im
por
ta
nc
e
of
m
a
na
gi
ng
pl
a
nt
di
s
e
a
s
e
s
to
i
m
pr
ove
yi
e
ld
qua
li
ty
.
D
is
e
a
s
e
d
le
a
ve
s
s
how
s
ym
pt
om
s
li
ke
de
f
or
m
a
ti
on,
di
s
c
ol
or
a
ti
on,
c
ur
li
ng,
a
nd
de
c
a
y. Q
ui
c
k,
a
c
c
ur
a
te
de
te
c
ti
on
a
nd
id
e
nt
if
ic
a
ti
on
of
th
e
s
e
di
s
e
a
s
e
s
a
r
e
e
s
s
e
nt
ia
l.
T
r
a
di
ti
ona
l
m
a
nua
l
m
e
th
od
s
a
r
e
c
o
s
tl
y,
ti
m
e
-
c
ons
um
in
g,
e
r
r
or
-
pr
one
,
a
nd
r
e
qui
r
e
e
xpe
r
ti
s
e
.
D
e
e
p
le
a
r
ni
ng
of
f
e
r
s
hi
ghe
r
a
c
c
ur
a
c
y
in
di
s
e
a
s
e
de
t
e
c
ti
on,
a
ddr
e
s
s
in
g
li
m
it
a
ti
ons
of
c
onve
nt
io
na
l
m
e
th
ods
.
P
la
nt
di
s
e
a
s
e
s
c
a
n
r
e
duc
e
li
f
e
s
p
a
n,
a
f
f
e
c
t
r
e
pr
oduc
ti
on,
de
gr
a
de
s
oi
l
qua
li
ty
,
a
nd
r
ui
n
c
r
ops
by
pe
r
s
is
ti
ng
in
th
e
s
oi
l
f
or
ye
a
r
s
.
M
ode
r
n
A
I
te
c
hnol
ogi
e
s
,
in
c
lu
d
in
g
s
uppor
t
ve
c
to
r
m
a
c
hi
ne
(
S
V
M
)
,
k
-
ne
a
r
e
s
t
ne
ig
hbor
s
(
KNN
)
,
a
nd
c
onvolut
io
n
ne
ur
a
l
ne
twor
k
(
C
N
N
)
,
c
a
n
he
lp
pr
e
ve
nt
c
r
op
lo
s
s
.
T
hi
s
r
e
s
e
a
r
c
h
us
e
s
da
ta
s
e
ts
to
tr
a
in
th
e
s
e
a
lg
or
it
hm
s
,
s
how
in
g
th
a
t
m
a
ny
pl
a
nt
di
s
e
a
s
e
s
a
r
e
c
a
us
e
d
by
te
m
pe
r
a
tu
r
e
c
ha
nge
s
a
nd
ba
c
te
r
ia
l
in
f
e
c
ti
ons
.
2.
L
I
T
E
R
A
T
U
R
E
R
E
V
I
E
W
P
la
nt
di
s
e
a
s
e
s
a
r
e
th
e
bi
gge
s
t
ob
s
ta
c
le
s
to
pr
oduc
in
g
a
gr
ic
ul
tu
r
a
l
pr
oduc
ts
.
T
hi
s
di
s
e
a
s
e
id
e
nt
if
ic
a
ti
on
te
c
hni
que
ha
s
de
c
e
nt
pot
e
nt
ia
l
b
e
c
a
us
e
it
c
a
n
i
m
m
e
di
a
te
ly
s
pot
pl
a
nt
le
a
f
is
s
ue
s
.
S
pe
e
d
a
nd
a
c
c
ur
a
c
y
a
r
e
th
e
two
k
e
y
a
s
pe
c
ts
of
c
r
op
di
s
e
a
s
e
di
a
gnos
i
s
in
“
m
a
c
hi
ne
le
a
r
ni
ng
s
ys
te
m
s
”
.
A
c
c
or
di
ng
to
[
3]
,
pl
a
nt
di
s
e
a
s
e
s
c
a
n
be
f
ound
us
in
g
th
e
hi
s
to
gr
a
m
m
a
tc
hi
ng
m
e
t
hod.
T
he
de
f
in
it
io
n
of
th
is
hi
s
to
gr
a
m
is
ba
s
e
d
on t
he
de
te
c
ti
on of
e
dge
s
a
nd t
he
f
r
e
que
nc
y of
oc
c
ur
r
e
nc
e
of
e
a
c
h c
ol
or
. L
a
ye
r
s
a
r
e
r
e
pr
e
s
e
nt
e
d by r
e
d, gr
e
e
n,
a
nd
bl
ue
pi
xe
ls
in
th
e
s
a
m
pl
e
phot
ogr
a
phs
u
s
in
g
th
e
la
ye
r
s
e
gr
e
ga
ti
on
a
ppr
oa
c
h.
A
ddi
ti
ona
ll
y,
th
e
pr
oc
e
dur
e
w
il
l
in
vol
ve
c
ont
our
id
e
nt
i
f
ic
a
ti
on.
A
c
c
or
di
ng
to
[
4
]
s
of
twa
r
e
t
ool
s
c
a
n
a
ut
om
a
ti
c
a
ll
y
r
e
c
ogni
z
e
a
nd
c
la
s
s
if
y
pl
a
nt
le
a
f
di
s
e
a
s
e
s
.
H
e
r
e
,
a
vi
s
ua
li
z
a
ti
on
te
c
hni
que
is
us
e
d
to
qui
c
kl
y
id
e
nt
if
y
th
e
di
s
or
de
r
s
.
“
K
-
m
e
a
ns
c
lu
s
te
r
in
g,
gr
a
y
-
le
ve
l
c
o
-
oc
c
ur
r
e
nc
e
m
a
tr
ix
(
G
L
C
M
)
is
a
s
ta
ti
s
ti
c
a
l
te
xt
ur
e
a
na
ly
s
is
te
c
hni
que
us
e
d
in
im
a
ge
pr
oc
e
s
s
in
g t
o e
xt
r
a
c
t
f
e
a
tu
r
e
s
f
r
om
i
m
a
ge
s
)
, a
nd
ba
c
kpr
opa
g
a
ti
on ne
ur
a
l
ne
twor
k
(
B
P
N
N
)
”
a
r
e
t
hr
e
e
e
f
f
e
c
ti
ve
a
ppr
oa
c
he
s
a
nd
k
e
y
te
c
hni
que
s
th
a
t
c
a
n
be
ut
il
iz
e
d
to
id
e
nt
if
y
c
r
op
di
s
e
a
s
e
s
w
it
h
gr
e
a
te
r
a
c
c
ur
a
c
y a
nd
in
le
s
s
ti
m
e
.
S
of
twa
r
e
to
ol
s
c
a
n
a
ut
om
a
ti
c
a
ll
y
r
e
c
ogni
z
e
a
nd
c
la
s
s
i
f
y
pl
a
nt
le
a
f
di
s
e
a
s
e
s
.
H
e
r
e
,
a
vi
s
ua
li
z
a
ti
on
te
c
hni
que
is
us
e
d
to
qui
c
kl
y
id
e
nt
if
y
th
e
di
s
or
de
r
s
.
“
K
-
m
e
a
ns
c
lu
s
te
r
in
g,
G
L
C
M
,
a
nd
B
P
N
N
”
a
r
e
th
r
e
e
e
f
f
e
c
ti
ve
a
ppr
oa
c
he
s
a
nd
ke
y
t
e
c
hni
que
s
th
a
t
c
a
n
be
ut
il
iz
e
d
t
o
id
e
nt
if
y
c
r
op
di
s
e
a
s
e
s
w
it
h
gr
e
a
te
r
a
c
c
ur
a
c
y
a
nd
in
le
s
s
ti
m
e
.
M
or
e
th
a
n
30,000
phot
os
f
r
om
th
e
[
5]
m
ode
l
w
e
r
e
di
vi
de
d
in
to
m
a
ny
c
la
s
s
if
ic
a
ti
ons
,
in
c
lu
di
ng
“
to
m
a
to
,
gr
a
pe
,
m
a
iz
e
, a
ppl
e
,
a
nd
s
ug
a
r
c
a
ne
di
s
e
a
s
e
s
”
[
6]
.
T
he
“
C
N
N
m
ode
l”
w
a
s
de
ve
lo
pe
d
us
in
g
a
n
“
a
ppl
ic
a
ti
on
pr
ogr
a
m
m
in
g
in
te
r
f
a
c
e
(
A
P
I
)
”
th
a
t
w
a
s
c
om
pa
ti
bl
e
w
it
h
“
P
yt
hon'
s
ne
ur
a
l
ne
two
r
k
a
ppl
ic
a
ti
ons
”
.
T
o
ge
t
ove
r
th
e
va
ni
s
hi
ng
gr
a
di
e
nt
'
s
dr
a
w
ba
c
ks
,
[
7]
in
tr
oduc
e
d
th
e
“
C
N
N
a
nd
A
le
xN
e
t
a
r
c
hi
te
c
tu
r
e
s
”
f
or
bui
ld
in
g
a
c
la
s
s
if
ie
r
.
T
he
A
le
xN
e
t
de
s
ig
n
ha
s
a
hi
ghe
r
a
c
c
ur
a
c
y
r
a
te
of
98.33%
.
M
or
e
c
or
r
e
c
t
out
c
om
e
s
c
a
n
be
a
c
hi
e
ve
d
by
ut
il
iz
in
g
th
e
“
A
le
xN
e
t
a
r
c
hi
te
c
tu
r
e
”
to
id
e
nt
if
y
s
ic
k
le
a
ve
s
[
8]
.
T
he
a
c
c
ur
a
c
y
of
th
e
“
C
N
N
”
m
a
ny
pr
e
-
tr
a
in
e
d
a
r
c
hi
te
c
tu
r
e
s
,
i
nc
lu
di
ng
“
vi
s
ua
l
ge
om
e
tr
y
gr
oup
ne
twor
k
(
V
G
G
N
e
t
)
,
r
e
s
id
ua
l
ne
twor
k
(
R
e
s
N
e
t
)
,
a
nd
G
oogL
e
N
e
t
”
,
a
r
e
c
om
pa
r
e
d.
A
ne
twor
k
w
a
s
pr
opos
e
d
by
[
9]
s
o
th
a
t
A
le
xN
e
t
m
a
y
be
c
ont
r
a
s
te
d
w
it
h
th
e
c
onve
nt
io
na
l
“
S
V
M
”
.
B
ot
h
“
S
V
M
a
nd
A
le
xN
e
t”
pe
r
f
or
m
e
d
w
e
ll
,
w
it
h
S
V
M
ha
vi
ng
a
n a
c
c
ur
a
c
y
of
91%
.
A
c
c
or
di
ng
to
[
10]
,
th
e
“
ne
ur
a
l
ne
twor
k
e
ns
e
m
bl
e
(
N
N
E
)
”
,
w
hi
c
h
w
a
s
e
m
pl
oye
d
in
th
e
m
ode
l
to
id
e
nt
if
y
he
a
lt
hy
le
a
ve
s
w
it
h
a
n
a
c
c
ur
a
c
y
of
87.5%
,
c
a
n
id
e
nt
if
y
m
a
ngo
le
a
f
i
ll
ne
s
s
e
s
.
U
s
in
g
in
c
e
pt
io
n
-
vi
s
ua
l
g
e
om
e
tr
y
gr
oup
ne
twor
k
(
I
N
C
-
V
G
G
N
)
,
th
e
id
e
nt
if
ic
a
ti
on
of
il
ln
e
s
s
e
s
in
r
ic
e
pl
a
nt
s
.
V
G
G
N
e
t
is
a
bui
lt
-
in
c
om
pone
nt
of
I
m
a
ge
N
e
t
a
nd
h
a
s
a
s
iz
a
bl
e
c
ol
le
c
ti
on
of
c
a
te
gor
iz
e
d
d
a
ta
s
e
t
s
r
a
th
e
r
th
a
n
ha
vi
ng
to
be
c
r
e
a
te
d
f
r
om
s
c
r
a
tc
h
a
nd
gi
ve
n
va
lu
e
s
[
11]
.
I
t
w
a
s
th
e
m
os
t
e
f
f
ic
ie
nt
m
e
th
od
of
obt
a
in
in
g
c
or
r
e
c
t
f
in
di
ngs
f
or
th
e
f
in
d
in
g
of
c
r
op
di
s
e
a
s
e
s
,
w
it
h
a
pr
e
c
is
io
n
of
91.83%
,
w
hi
c
h
w
a
s
s
ig
ni
f
ic
a
nt
ly
hi
ghe
r
th
a
n
th
a
t
of
ot
he
r
a
ppr
oa
c
he
s
,
e
ve
n
i
n
th
e
f
a
c
e
of
s
ig
ni
f
ic
a
nt
obs
tr
uc
ti
ons
.
T
h
e
pe
r
f
or
m
a
nc
e
of
num
e
r
ous
pr
e
-
tr
a
in
e
d
ne
ur
a
l
ne
twor
ks
w
a
s
s
um
m
a
r
iz
e
d
by
[
12]
,
a
nd
th
e
tr
a
ns
f
e
r
le
a
r
ni
ng
m
ode
l'
s
pr
e
-
tr
a
in
e
d
w
e
ig
ht
s
,
w
hi
c
h
c
om
pr
is
e
s
of
V
G
G
16,
M
obi
le
-
N
e
t,
I
nc
e
pt
io
nV
3,
R
e
s
N
e
t5
0,
I
nc
e
pt
io
n
-
R
e
s
N
e
t
-
V
2,
a
nd
pr
e
-
tr
a
in
e
d
ne
twor
ks
,
w
e
r
e
pr
ovi
de
d
by
bu
il
t
-
in
K
e
r
a
s
a
pps
.
T
he
a
ut
hor
s
of
[
13]
,
[
14]
de
s
c
r
ib
e
us
in
g
C
N
N
to
id
e
nt
if
y
to
m
a
to
pl
a
nt
le
a
ve
s
.
T
he
y
a
c
c
om
pl
is
he
d
th
is
us
in
g
a
n
im
por
te
d
R
e
s
N
e
t
-
50
m
ode
l
a
nd
th
e
t
r
a
ns
f
e
r
le
a
r
ni
ng
th
e
or
y.
T
he
y
di
vi
de
d
a
da
ta
s
e
t
of
2
,
006
pi
c
tu
r
e
s
in
ha
lf
“
80%
a
nd
20%
”
f
o
r
va
li
da
ti
on a
nd t
r
a
in
in
g of
th
e
m
ode
l.
T
he
m
ode
l
c
a
n i
de
nt
if
y t
h
e
a
il
m
e
nt
i
n t
he
s
hor
te
s
t
a
m
ount
of
t
i
m
e
due
t
o
th
e
ir
hi
gh
le
ve
l
o
f
a
c
c
ur
a
c
y
(
97%
)
.
[
15
]
A
s
oppos
e
d
to
w
he
r
e
a
da
ta
s
e
t
of
ove
r
7
,
000
i
m
a
ge
s
w
a
s
us
e
d
to
c
r
e
a
te
th
e
“
C
N
N
-
ba
s
e
d
A
le
x
-
N
e
t
m
ode
l”
a
nd
c
om
pa
r
e
it
a
g
a
in
s
t
th
e
“
V
G
G
16
a
nd
L
e
ne
t5
m
ode
ls
”
.
T
h
e
y
a
ppl
ie
d
c
e
r
ta
in
f
unda
m
e
nt
a
l
m
a
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hm
s
,
s
u
c
h
a
s
S
V
M
a
nd
K
N
N
,
a
nd
w
e
r
e
a
bl
e
to
a
tt
a
in
a
n
a
c
c
ur
a
c
y of
96.7%
how
e
ve
r
, t
he
ir
pe
r
f
or
m
a
nc
e
l
a
gge
d be
hi
nd t
ha
t
of
V
G
G
-
16 a
nd L
e
N
et
-
5.
L
oc
a
l
-
bi
na
r
y
-
pa
tt
e
r
ns
(
L
B
P
)
a
nd
hi
s
to
gr
a
m
of
or
ie
nt
e
d
gr
a
di
e
nt
s
(
HOG
)
a
r
e
ut
il
is
e
d
to
di
s
ti
ngui
s
h
di
f
f
e
r
e
nt
a
s
pe
c
ts
f
r
om
O
ts
u'
s
t
e
c
hni
que
i
n
[
16]
to
s
e
gm
e
nt
s
om
e
of
t
he
di
s
e
a
s
e
s
.
T
he
da
ta
w
a
s
c
l
a
s
s
if
ie
d us
in
g
a
n
S
V
M
te
c
hni
que
,
a
nd
a
pol
ynomi
a
l
ke
r
ne
l
w
a
s
ut
il
iz
e
d
to
ge
t
a
n
a
c
c
ur
a
c
y
of
94.6%
.
T
he
r
e
f
or
e
,
e
a
r
ly
id
e
nt
if
ic
a
ti
on
of
c
r
op
di
s
e
a
s
e
w
il
l
s
to
p
th
e
c
r
op'
s
out
put
f
r
om
d
e
c
li
ni
ng.
T
he
“
K
-
m
e
a
ns
c
lu
s
t
e
r
in
g”
te
c
hni
que
is
us
e
d
by
[
17]
to
id
e
nt
if
y
in
f
e
c
te
d
le
a
ve
s
w
it
h
a
c
c
ur
a
c
y;
it
f
in
ds
th
e
le
a
f
'
s
de
a
d
or
di
s
e
a
s
e
d
a
r
e
a
s
[
18]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
20
25
:
3724
-
3733
3726
A
ddi
ti
ona
ll
y,
th
e
pr
e
c
is
io
n
of
th
e
S
V
M
a
nd
K
N
N
a
lg
or
it
hm
s
w
e
r
e
c
om
pa
r
e
d.
T
he
a
c
c
ur
a
c
y
of
th
e
S
V
M
a
lg
or
it
hm
w
a
s
95%
,
w
hi
le
th
e
a
c
c
ur
a
c
y
of
th
e
K
N
N
a
ppr
oa
c
h
w
a
s
85%
.
T
a
bl
e
1
s
how
s
th
e
pe
r
f
or
m
a
nc
e
of
va
r
io
us
c
r
ops
.
T
he
doc
um
e
nt
h
a
s
be
e
n
or
ga
ni
z
e
d
in
th
e
m
a
nne
r
a
s
f
ol
lo
w
s
.
T
he
te
c
hni
que
s
f
or
pr
e
-
pr
oc
e
s
s
in
g
th
e
im
a
ge
da
ta
s
e
t,
s
uc
h
a
s
im
a
g
e
a
c
qui
s
it
io
n,
im
a
ge
a
ugm
e
nt
a
ti
on,
a
nd
im
a
ge
da
ta
s
e
t
c
ons
tr
uc
ti
on,
a
r
e
in
tr
oduc
e
d
in
s
e
c
ti
on
3.
T
he
A
nt
c
ol
ony
opt
im
iz
a
ti
on
w
i
th
c
onvolut
io
n
-
ne
ur
a
l
-
ne
twor
k
(
A
C
O
-
C
N
N
)
m
e
th
odol
ogy
ut
il
iz
e
d
f
or
th
is
in
ve
s
ti
ga
ti
on
is
di
s
c
us
s
e
d
in
s
e
c
ti
on
4.
T
he
s
e
gm
e
nt
a
ti
on
out
c
om
e
s
of
th
e
s
ugge
s
te
d m
ode
l
a
r
e
e
xa
m
in
e
d i
n
s
e
c
ti
on 5. F
in
a
ll
y, our
c
onc
lu
s
io
n i
s
pr
e
s
e
nt
e
d.
T
a
bl
e
1. P
e
r
f
or
m
a
nc
e
of
va
r
io
us
c
r
ops
C
r
op T
ype
D
a
t
a
s
e
t
P
e
r
f
or
m
a
nc
e
F
r
a
m
e
w
or
k
R
e
f
e
r
e
nc
e
s
Y
e
a
r
A
ppl
e
26,377
A
c
c
ur
a
c
y
i
nc
r
e
a
s
e
d
D
B
N
e
t
[
19]
2023
T
e
a
4
,
000
B
e
t
t
e
r
a
c
c
ur
a
c
y
Y
O
L
O
v7
[
20]
2023
A
ppl
e
, C
or
n, P
ot
a
t
o,
T
om
a
t
o, a
nd R
i
c
e
37,315
G
r
e
a
t
e
r
pr
e
c
i
s
i
on a
nd
a
de
qua
t
e
de
t
e
c
t
i
on s
pe
e
d
A
l
e
xN
e
t
,
M
obi
l
e
N
e
t
[
21]
2022
38 C
l
a
s
s
e
s
54,305
B
e
t
t
e
r
a
c
c
ur
a
c
y
D
e
ns
e
N
e
t
-
121,
R
e
s
N
e
t
-
50
[
22]
2022
A
ppl
e
1
,
200
B
e
t
t
e
r
a
c
c
ur
a
c
y
M
a
s
k R
C
N
N
[
23]
2021
P
a
pa
ya
2
,
000
I
nt
r
oduc
e
t
he
us
a
ge
of
Y
O
L
O
va
r
i
a
nt
s
t
ha
t
a
r
e
l
i
ght
e
r
, m
or
e
e
f
f
e
c
t
i
ve
, a
nd
ha
ve
a
r
a
pi
d de
t
e
c
t
i
on r
a
t
e
.
Y
O
L
O
[
24]
2021
C
i
t
r
us
392
H
i
ghe
r
pr
e
c
i
s
i
on
Y
O
L
O
v4
[
25]
2020
G
r
a
pe
4
,
449
G
r
e
a
t
e
r
pr
e
c
i
s
i
on a
nd s
uf
f
i
c
i
e
nt
r
e
s
pons
e
t
i
m
e
F
a
s
t
e
r
D
R
-
I
A
C
N
N
[
26]
2020
C
a
s
s
a
va
2
,
415
E
va
l
ua
t
e
t
he
m
ode
l
’
s
pe
r
f
or
m
a
nc
e
on
m
obi
l
e
-
c
a
pt
ur
e
d i
m
a
ge
s
a
nd vi
de
os
pos
t
-
de
pl
oym
e
nt
i
n a
n a
pp.
SSD
[
27]
2019
3.
M
A
T
E
R
I
A
L
S
A
N
D
M
E
T
H
O
D
S
F
in
di
ng
a
r
e
a
s
th
a
t
a
r
e
in
f
e
c
te
d
w
it
h
pl
a
nt
di
s
e
a
s
e
s
a
nd
id
e
nt
if
yi
ng
w
he
r
e
th
e
y
a
r
e
in
di
f
f
ic
ul
t
na
tu
r
a
l
c
ondi
ti
ons
is
e
s
s
e
nt
ia
l
f
or
pr
ope
r
c
a
te
gor
iz
a
ti
on
a
nd
id
e
nt
if
ic
a
t
io
n
of
c
r
op
il
ln
e
s
s
e
s
a
s
w
e
ll
a
s
th
e
e
va
lu
a
ti
on
of
c
r
op
di
s
e
a
s
e
s
e
ve
r
it
y
.
C
om
put
e
r
vi
s
io
n
te
c
hnol
ogy
is
us
e
d
in
pl
a
nt
di
s
e
a
s
e
de
te
c
ti
on
to
a
c
hi
e
ve
th
is
.
E
a
r
ly
-
s
ta
ge
phyt
opa
th
ol
ogy
a
na
ly
ti
c
s
e
m
pl
oye
d
a
s
li
di
ng
w
i
ndow
a
ppr
oa
c
h
to
c
hoos
e
c
a
ndi
d
a
te
r
e
gi
ons
,
e
xt
r
a
c
t
c
a
ndi
da
te
r
e
gi
on a
tt
r
ib
ut
e
s
, a
nd t
he
n c
la
s
s
if
y t
he
m
us
in
g
a
c
la
s
s
if
ie
r
t
o de
te
r
m
in
e
a
r
e
a
o
f
i
nt
e
r
e
s
t
. T
hi
s
te
c
hni
que
it
e
r
a
ti
ve
ly
m
ove
s
a
c
r
os
s
th
e
im
a
g
e
w
hi
le
ut
il
iz
in
g
va
r
io
us
s
iz
e
s
a
nd
w
id
th
s
.
E
ve
n
th
ough
th
is
te
c
hni
que
doe
s
n'
t
m
i
s
s
a
ny
in
f
e
c
te
d
z
on
e
ta
r
ge
ts
,
th
e
dupl
ic
a
t
e
c
a
ndi
da
te
r
e
gi
on
s
th
a
t
a
pp
e
a
r
ne
e
d
a
lo
t
of
pr
oc
e
s
s
in
g
w
or
k
a
nd
ta
ke
a
w
hi
le
to
tr
a
ve
r
s
e
th
e
di
s
e
a
s
e
p
ic
tu
r
e
a
ga
in
,
w
hi
c
h
le
a
ds
to
poor
r
e
a
l
-
ti
m
e
de
te
c
ti
on.
V
a
r
io
us
m
e
th
od
s
f
or
c
om
put
a
ti
ona
l
im
a
gi
ng
a
nd
th
e
us
e
of
th
e
pi
c
tu
r
e
c
a
t
e
gor
iz
a
ti
on
s
ys
t
e
m
a
r
e
ba
s
e
d
on
a
r
ti
f
ic
ia
l
vi
s
io
n
due
to
th
e
qui
c
k
gr
ow
th
of
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
te
c
hnol
ogy.
T
he
f
iv
e
s
te
p
s
of
th
e
m
e
th
odol
ogy
s
ugge
s
te
d
by
R
e
f
e
r
e
nc
e
to
id
e
nt
if
y
th
e
c
ogol
le
r
o
w
or
m
da
m
a
ge
in
m
a
iz
e
f
ie
ld
s
a
r
e
im
a
ge
c
ol
le
c
ti
on, pr
e
pr
oc
e
s
s
in
g,
s
e
gm
e
nt
a
ti
on, f
e
a
tu
r
e
e
xt
r
a
c
ti
on, a
nd
c
la
s
s
if
ic
a
ti
on.
3.1
.
S
t
e
p
s
f
or
p
la
n
t
l
e
af
d
is
e
as
e
i
d
e
n
t
if
ic
at
io
n
3.1.1.
I
m
age
a
c
q
u
is
it
io
n
T
he
da
ta
s
e
t
s
f
or
r
ic
e
,
pe
ppe
r
,
a
nd
pot
a
to
e
s
a
r
e
us
e
d
in
th
is
s
tu
dy.
F
or
bo
th
th
e
pot
a
to
,
pe
ppe
r
,
a
nd
r
ic
e
da
ta
s
e
ts
,
th
e
da
ta
a
r
e
s
pl
it
in
to
80:
20
r
a
ti
o
,
w
he
r
e
80%
o
f
th
e
pi
c
tu
r
e
s
a
r
e
u
s
e
d
f
or
le
a
r
ni
ng
a
nd
20%
o
f
im
a
ge
s
a
r
e
us
e
d
f
or
e
va
lu
a
ti
on
.
T
he
to
ta
l
of
5
,
932
im
a
ge
s
in
th
e
R
ic
e
da
ta
s
e
t
de
pi
c
ts
f
our
di
s
ti
nc
t
c
a
t
e
gor
ie
s
of
r
ic
e
c
r
op
in
f
e
c
ti
on
,
c
om
pr
is
in
g
“
bl
a
s
t,
ba
c
te
r
ia
l
bl
ig
ht
,
tu
ng
r
o,
a
nd
br
ow
n
s
pot
”
.
3
,
785
pho
to
s
w
e
r
e
us
e
d
f
or
t
r
a
in
in
g
w
it
h
a
n
“
80:
20”
te
s
t
tr
a
in
s
pl
it
,
a
nd
947
va
r
io
us
i
m
a
ge
s
w
e
r
e
us
e
d
f
or
te
s
ti
ng.
F
ig
u
r
e
1
di
s
pl
a
ys
r
e
pr
e
s
e
nt
a
ti
ve
phot
os
f
r
om
th
e
r
ic
e
da
ta
s
e
t.
T
he
s
tu
dy
m
a
ke
s
us
e
of
a
da
ta
s
e
t
of
1,500
phot
os
of
pot
a
to
le
a
ve
s
.
300
phot
os
w
e
r
e
ut
il
iz
e
d
f
or
te
s
ti
ng,
w
hi
le
1
,
200
pi
c
tu
r
e
s
w
e
r
e
us
e
d
f
or
tr
a
in
in
g
a
nd
e
va
lu
a
ti
on
.
H
e
a
lt
hy
pot
a
to
le
a
ve
s
,
E
a
r
ly
bl
ig
ht
,
a
nd
la
te
bl
ig
ht
a
r
e
r
e
pr
e
s
e
nt
e
d
in
th
e
c
ol
le
c
ti
on
in
di
f
f
e
r
e
nt
w
a
ys
.
I
n
F
ig
ur
e
2,
a
s
a
m
pl
e
of
pot
a
to
le
a
ve
s
i
s
di
s
pl
a
ye
d. T
he
to
ta
l
num
be
r
of
tr
a
in
in
g
a
nd
te
s
t
pi
c
tu
r
e
s
in
th
e
da
ta
s
e
t
s
is
s
how
n i
n T
a
b
le
1.
3.1.2.
P
r
e
-
p
r
oc
e
s
s
in
g
P
r
e
-
pr
oc
e
s
s
in
g
ha
ppe
n
s
onc
e
th
e
pi
c
tu
r
e
ha
s
be
e
n
c
hos
e
n
a
s
th
e
f
ounda
ti
on
f
or
th
e
di
a
gno
s
is
of
le
a
f
di
s
e
a
s
e
.
T
he
m
e
di
a
n
f
il
te
r
ha
s
be
e
n
u
s
e
d
to
im
pr
ove
th
e
c
r
op
phot
os
by
r
e
duc
in
g
noi
s
e
a
nd
r
e
m
ovi
ng
unde
s
ir
a
bl
e
e
le
m
e
nt
s
.
T
he
“
m
e
di
a
n
f
il
te
r
,
a
non
-
li
ne
a
r
,
w
e
ll
-
s
tr
uc
tu
r
e
d
di
gi
ta
l
f
il
te
r
in
g
m
e
th
od”
,
is
f
r
e
que
nt
ly
us
e
d t
o m
in
im
iz
e
th
e
noi
s
e
i
n pi
c
tu
r
e
s
.
T
he
out
c
om
e
of
t
he
m
e
di
a
n f
il
te
r
f
or
m
ul
a
a
s
in
(
1)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
e
d c
onv
ol
ut
io
n ne
ur
al
n
e
tw
or
k
w
it
h ant c
ol
ony
al
gor
it
h
m
f
or
ac
c
ur
at
e
pl
ant
…
(
Shw
e
ta
B
ondr
e
)
3727
(
,
)
=
(
,
)
∈
{
(
,
)
}
(
1)
3.1.3.
S
e
gm
e
n
t
at
io
n
B
a
s
e
d
on
pr
e
de
f
in
e
d
pa
r
a
m
e
te
r
s
,
s
e
gm
e
nt
a
ti
on
br
e
a
ks
up
pl
a
n
t
le
a
f
im
a
ge
s
f
or
e
a
s
y
m
a
ni
pul
a
ti
on.
T
he
s
ugge
s
te
d
s
e
pa
r
a
ti
on
m
e
th
od
di
vi
de
s
th
e
c
r
op
le
a
f
in
to
it
s
c
om
pone
nt
s
.
T
he
s
im
il
a
r
it
y
a
nd
vol
a
ti
li
ty
of
pi
xe
l
c
onc
e
nt
r
a
ti
on
m
a
k
e
s
e
gm
e
nt
a
ti
on
di
f
f
ic
ul
t.
S
im
il
a
r
it
ie
s
a
r
e
f
ound
u
s
in
g
“
c
ol
or
-
ba
s
e
d
th
r
e
s
hol
di
ng
”
.
E
qua
ti
on de
f
in
e
s
s
e
gm
e
nt
a
ti
on.
|
ℎ
(
,
)
|
=
{
0
,
(
,
)
<
1
,
(
,
)
>
(
2)
3.1.4.
F
e
at
u
r
e
e
xt
r
ac
t
io
n
T
he
pr
oc
e
s
s
of
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
dur
in
g
im
a
ge
id
e
nt
if
ic
a
ti
on
is
e
s
s
e
nt
ia
l.
F
ig
ur
e
1(
a
)
s
how
s
th
e
m
a
iz
e
l
e
a
f
bl
ig
ht
, F
ig
ur
e
1(
b
)
s
how
s
t
he
w
he
a
t
le
a
f
s
pot
, F
ig
ur
e
1(
c
)
s
how
s
t
he
w
he
a
t
ye
ll
ow
l
e
a
f
, F
ig
ur
e
1(
d
)
s
how
s
th
e
s
oybe
a
n
br
ow
n
s
pot
,
F
ig
ur
e
1(
e
)
s
how
s
th
e
pot
a
to
m
il
d
le
a
f
c
ur
l,
a
nd
F
ig
ur
e
1
(
f
)
s
how
s
th
e
pot
a
to
da
r
k
br
ow
n
s
pot
.
T
he
f
ig
ur
e
s
s
how
th
e
s
a
m
pl
e
im
a
ge
s
of
th
e
di
s
e
a
s
e
d
da
ta
s
e
t
th
a
t
a
r
e
us
e
d
in
th
e
r
e
s
e
a
r
c
h
w
or
k.
F
or
f
e
a
tu
r
e
e
xt
r
a
c
ti
on
a
nd
c
la
s
s
if
ic
a
ti
on,
A
C
O
a
nd C
N
N
a
r
e
ut
il
is
e
d,
r
e
s
pe
c
ti
ve
ly
.
I
t
is
us
e
d
to
id
e
nt
if
y
th
e
in
f
e
c
te
d
a
r
e
a
ut
il
is
in
g
uni
que
a
nt
c
om
m
uni
c
a
ti
on
b
e
ha
vi
or
a
nd
c
a
te
gor
iz
e
th
e
di
s
e
a
s
e
d
c
r
op
l
e
a
f
f
or
th
e
goa
l
of
upc
om
in
g
pr
e
ve
nt
io
n.
A
ddi
ti
ona
ll
y,
th
e
A
C
O
m
e
t
hod
ha
s
e
m
e
r
ge
d
a
s
a
nove
l
a
ppr
oa
c
h
to
a
ppr
oxi
m
a
te
opt
im
iz
a
ti
on.
T
o
f
in
d
th
e
f
a
s
te
s
t
pa
th
to
th
e
ir
f
ood
s
our
c
e
,
a
nt
s
'
m
a
jo
r
w
a
y
of
c
om
m
uni
c
a
ti
ng
i
s
th
r
ough
in
di
r
e
c
t
m
e
a
ns
.
T
hi
s
a
nt
ha
s
a
s
pe
c
ia
l
c
ha
r
a
c
te
r
is
ti
c
t
ha
t
is
us
e
d
in
A
C
O
.
T
he
A
C
O
is
ut
il
iz
e
d
to
di
f
f
e
r
e
nt
ia
te
be
twe
e
n
he
a
lt
hy
a
nd
s
ic
k
pl
a
nt
le
a
ve
s
in
th
is
in
s
ta
nc
e
.
T
he
phe
r
om
one
r
a
te
is
a
c
r
it
ic
a
l
da
ta
ba
s
e
a
tt
r
ib
ut
e
th
a
t
m
us
t
be
m
odi
f
ie
d
in
it
ia
ll
y.
W
it
h
G
a
s
th
e
num
be
r
of
uni
que
f
e
a
tu
r
e
ve
c
to
r
s
in
it
s
r
ow
s
a
nd
c
ol
um
ns
,
a
m
a
tr
ix
(
h
)
of
di
m
e
ns
io
ns
G
*
G
c
om
pr
is
e
s
th
e
f
e
a
tu
r
e
a
na
ly
s
is
da
ta
.
F
ol
lo
w
in
g
m
odi
f
ic
a
ti
on
o
f
th
e
A
C
O
th
e
m
a
in
c
om
put
a
ti
on
us
in
g
e
xpe
r
im
e
nt
a
l
te
c
hni
que
F
is
c
a
r
r
ie
d
out
.
S
e
le
c
t
th
e
be
s
t
m
a
te
r
ia
ls
a
nd
s
ubgr
oups
f
or
th
e
upc
om
in
g
it
e
r
a
ti
on.
T
he
in
it
ia
l
a
nd
ve
r
y
im
por
ta
nt
pha
s
e
in
a
ppl
yi
ng
th
e
A
C
O
a
lg
or
it
hm
is
th
e
in
it
ia
li
z
a
ti
on
of
it
s
f
a
c
to
r
s
.
A
ddi
ti
ona
ll
y,
th
e
“
A
C
O
a
lg
or
it
hm
”
in
c
lu
de
s
a
s
e
pa
r
a
t
e
c
a
lc
ul
a
ti
on
pr
oc
e
dur
e
a
nd
e
xc
e
ll
e
nt
r
e
s
il
ie
nc
e
.
W
he
n
de
a
li
ng
w
it
h
c
om
pl
e
x
opt
im
i
z
a
ti
on
pr
obl
e
m
s
,
A
C
O
e
x
c
e
ls
a
nd
is
r
e
a
di
ly
in
te
r
c
ha
nge
a
bl
e
w
it
h ot
he
r
m
e
th
ods
. A
nt
s
e
m
pl
oy ma
th
e
m
a
ti
c
a
l
te
c
hni
que
s
t
o f
in
d obje
c
ts
i
n t
he
s
e
a
r
c
h s
pa
c
e
w
he
r
e
a
s
A
C
O
ut
il
is
e
t
he
upda
te
d phe
r
om
one
. L
oc
a
l
a
nd glob
a
l
s
e
a
r
c
he
s
s
e
r
ve
a
s
t
he
ba
s
is
of
A
C
O
.
(
a
)
(
b)
(
c
)
(
d)
(
e
)
(f)
F
ig
ur
e
1.
S
a
m
pl
e
i
m
a
ge
s
of
t
he
di
s
e
a
s
e
d da
ta
s
e
t
of
(
a
)
m
a
iz
e
l
e
a
f
bl
ig
ht
, (
b)
w
he
a
t
le
a
f
s
pot
,
(
c
)
w
he
a
t
ye
ll
ow
l
e
a
f
, (
d)
S
oybe
a
n br
ow
n s
pot
,
(
e
)
p
ot
a
to
m
il
d l
e
a
f
c
ur
l
, a
nd (
f
)
pot
a
to
da
r
k br
ow
n s
pot
3.2
.
C
la
s
s
if
i
c
at
io
n
T
he
le
a
f
i
s
th
e
n s
or
te
d
in
a
ne
ur
a
l
n
e
twor
k
us
in
g
di
f
f
e
r
e
nt
c
a
te
g
or
iz
a
ti
on
te
c
hni
que
s
.
F
ig
ur
e
2
s
how
s
th
e
C
N
N
a
r
c
hi
te
c
tu
r
e
f
or
im
a
ge
c
la
s
s
if
ic
a
ti
on
.
D
if
f
e
r
e
nt
a
lg
or
it
hm
s
a
r
e
c
om
pa
r
e
d
us
in
g
th
e
ir
pe
r
f
o
r
m
a
nc
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
20
25
:
3724
-
3733
3728
m
e
tr
ic
s
a
f
te
r
be
in
g
lo
a
de
d
in
to
th
e
s
ugge
s
te
d
“
n
e
ur
a
l
ne
twor
k
m
ode
l”
f
or
c
la
s
s
if
yi
ng
phot
os
.
T
o
id
e
nt
if
y
a
nd
c
a
te
gor
iz
e
th
e
gr
a
pe
a
nd
m
a
ngo
in
f
e
c
ti
on
-
a
f
f
e
c
te
d
pl
a
nt
le
a
f
,
vi
s
ua
li
z
a
ti
on
m
e
th
ods
a
nd
m
a
ppi
ng
f
unc
ti
ons
a
r
e
ul
ti
m
a
te
ly
ut
il
iz
e
d f
or
f
il
e
de
li
ve
r
y a
nd na
m
in
g.
F
ig
ur
e
2. C
N
N
a
r
c
hi
te
c
tu
r
e
f
or
i
m
a
ge
c
la
s
s
if
ic
a
ti
on
3.2.1.
C
on
vol
u
t
io
n
al
n
e
u
r
al
n
e
t
w
or
k
E
a
c
h
f
or
m
of
le
a
f
in
f
e
c
ti
on
is
d
is
ti
ngui
s
he
d
us
in
g
C
N
N
c
la
s
s
if
ie
r
s
.
I
t
e
va
lu
a
te
s
gr
a
phi
c
a
l
r
e
pr
e
s
e
nt
a
ti
ons
e
f
f
e
c
ti
ve
ly
a
nd
e
li
m
in
a
te
s
e
xt
r
a
ne
ou
s
e
le
m
e
n
ts
th
a
nks
to
it
s
hi
e
r
a
r
c
hi
c
a
l
a
r
c
hi
te
c
tu
r
e
.
I
ts
m
ul
ti
-
la
ye
r
e
d
a
r
c
hi
te
c
tu
r
e
le
ts
it
e
f
f
e
c
ti
ve
ly
a
s
s
e
s
s
vi
s
ua
l
r
e
pr
e
s
e
nt
a
ti
ons
a
nd
c
ut
out
e
xt
r
a
ne
ous
c
om
pone
nt
s
.
F
our
la
ye
r
s
c
om
pr
is
e
th
e
C
N
N
c
la
s
s
if
ie
r
:
out
put
,
f
ul
ly
c
onne
c
te
d,
m
a
x
pool
in
g
,
a
nd
c
onvolut
io
na
l.
B
e
f
or
e
tr
a
in
in
g
a
C
N
N
,
s
pe
c
tr
a
of
pi
xe
l
in
te
ns
it
ie
s
in
th
e
d
a
ta
s
e
t
of
pl
a
nt
le
a
f
im
a
ge
s
.
T
he
C
N
N
m
ode
l
pe
r
f
or
m
s
qui
te
w
e
ll
a
ll
dur
in
g
tr
a
in
in
g.
T
he
pi
c
tu
r
e
s
of
f
e
r
e
d
f
or
in
put
s
h
oul
d
a
ll
be
th
e
s
a
m
e
s
iz
e
.
E
a
c
h
pi
c
tu
r
e
in
th
e
tr
a
in
in
g s
e
t
unde
r
w
e
nt
t
he
f
ol
lo
w
in
g nor
m
a
li
z
a
ti
on
a
s
(
3)
.
(
,
)
=
(
,
)
−
(
3)
–
C
onvolut
io
n l
a
ye
r
:
u
s
in
g di
f
f
e
r
e
nt
la
ye
r
s
t
o e
va
lu
a
te
t
he
c
om
pl
e
xi
ty
of
e
ve
r
y
i
m
a
ge
, t
he
c
onvolut
io
n
l
a
ye
r
e
va
lu
a
te
s
a
r
e
s
tr
ic
te
d numbe
r
of
i
nput
i
m
a
ge
s
. I
t
di
r
e
c
tl
y c
or
r
e
s
ponds
w
it
h t
he
f
e
a
tu
r
e
s
of
t
he
i
m
a
ge
s
.
=
(
∑
−
1
∗
+
∈
)
(
4
)
N
i
r
e
f
e
r
s
to
a
n
in
put
opt
io
n.
A
n
a
ddi
ti
ve
bi
a
s
b
w
a
s
th
e
r
e
by
ge
n
e
r
a
te
d.
A
f
te
r
a
ppl
yi
ng
th
e
ke
r
ne
l
to
m
a
p
i
,
it
de
te
r
m
in
e
d i
f
m
a
p
j
a
nd ma
p
k
a
dde
d up to m
a
p
i
.
–
M
a
x
pool
in
g
la
ye
r
:
i
n
or
de
r
to
de
c
r
e
a
s
e
f
it
ti
ng
a
nd
th
e
s
iz
e
o
f
th
e
ne
ur
ons
us
e
d
in
th
e
dow
n
-
s
a
m
pl
in
g
la
ye
r
,
th
is
la
ye
r
is
de
pl
oye
d.
W
hi
le
r
e
duc
in
g
th
e
c
om
put
a
ti
o
na
l
r
a
te
,
f
e
a
tu
r
e
m
a
p
di
m
e
ns
io
ns
,
tr
a
in
in
g
le
ngt
h,
a
nd
num
be
r
of
pa
r
a
m
e
te
r
s
,
th
e
pool
in
g
la
y
e
r
m
it
ig
a
te
s
ove
r
f
it
ti
ng.
H
a
lf
of
th
e
tr
a
in
in
g
da
ta
a
nd
a
ll
of
t
he
t
e
s
t
da
ta
m
us
t
be
c
on
s
id
e
r
e
d ove
r
f
it
ti
ng.
–
F
ul
ly
c
onne
c
te
d
la
ye
r
:
i
m
a
ge
s
ha
ve
be
e
n
c
l
a
s
s
if
ie
d
u
s
in
g
th
e
f
ul
ly
-
c
onne
c
te
d
la
ye
r
.
B
e
f
or
e
e
v
e
r
y
c
onvolut
io
na
l
la
ye
r
,
th
e
r
e
e
xi
s
t
th
e
f
ul
ly
li
nke
d
la
ye
r
s
.
T
he
m
a
ppi
ng
be
twe
e
n
th
e
in
put
a
nd
out
put
r
e
pr
e
s
e
nt
a
ti
ons
i
s
m
a
de
e
a
s
ie
r
by
th
e
f
ul
ly
c
onne
c
te
d
la
y
e
r
.
A
t
th
e
ve
r
y
to
p
of
th
e
n
e
twor
k,
you'
ll
f
in
d
f
ul
ly
c
onne
c
te
d l
a
ye
r
s
. T
h
e
f
ul
ly
c
onne
c
te
d l
a
ye
r
t
a
ke
s
i
ts
i
nput
f
r
om
t
he
m
a
x pooli
ng l
a
ye
r
.
–
S
of
tM
a
x
la
ye
r
:
a
nor
m
a
li
z
e
d
pr
oba
bi
li
ty
di
s
tr
ib
ut
io
n
is
c
r
e
a
te
d
f
r
om
th
e
s
c
or
e
s
by
m
e
a
n
s
of
th
e
S
of
t
M
a
x
la
ye
r
.
T
he
out
put
is
f
e
d
in
to
th
e
c
la
s
s
if
ie
r
.
I
n
th
e
S
o
f
t
M
a
x
la
ye
r
,
pl
a
nt
di
s
e
a
s
e
s
a
r
e
c
a
te
gor
iz
e
d
us
in
g
th
e
w
e
ll
-
known s
of
tm
a
x c
la
s
s
if
ie
r
.
(
⃗
)
=
ⅇ
=
1
ⅇ
(
5)
A
C
N
N
m
ode
l
id
e
nt
if
ie
s
pa
tt
e
r
ns
in
im
a
ge
s
us
in
g
f
il
te
r
s
a
nd
c
onvolut
io
n
la
ye
r
s
,
w
it
h
th
e
pr
oc
e
s
s
e
d
da
ta
pa
s
s
e
d
th
r
ough
R
e
L
U
to
e
li
m
in
a
te
ne
ga
ti
ve
va
lu
e
s
.
T
h
e
pool
in
g
la
ye
r
th
e
n
r
e
duc
e
s
th
e
in
put
s
iz
e
a
nd
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
e
d c
onv
ol
ut
io
n ne
ur
al
n
e
tw
or
k
w
it
h ant c
ol
ony
al
gor
it
h
m
f
or
ac
c
ur
at
e
pl
ant
…
(
Shw
e
ta
B
ondr
e
)
3729
a
c
c
e
le
r
a
te
s
pr
oc
e
s
s
in
g
us
in
g
hype
r
pa
r
a
m
e
t
e
r
s
li
ke
f
il
te
r
s
iz
e
,
s
tr
id
e
,
a
nd
pool
in
g
ty
pe
(
m
a
x
or
a
ve
r
a
ge
)
.
A
C
N
N
c
a
n
ha
ve
m
ul
ti
pl
e
pool
in
g
-
la
ye
r
s
a
nd
m
ul
ti
pl
e
c
onvolut
io
n
la
ye
r
s
,
e
ndi
ng
in
f
ul
ly
c
onne
c
te
d
la
ye
r
s
f
or
c
la
s
s
if
ic
a
ti
on.
T
he
dr
opout
la
ye
r
pr
e
ve
nt
s
ove
r
f
it
ti
ng,
w
hi
le
th
e
s
of
tm
a
x
f
unc
ti
on
out
put
s
pr
oba
bi
li
ti
e
s
f
or
c
la
s
s
e
s
.
T
h
e
pr
opos
e
d
16
-
la
ye
r
C
N
N
m
ode
l
in
c
lu
de
s
f
iv
e
c
onvolut
io
n
la
ye
r
s
,
th
r
e
e
ba
tc
h
nor
m
a
li
z
a
ti
on
la
ye
r
s
,
two
m
a
x
-
pool
in
g
la
ye
r
s
,
a
nd
f
iv
e
f
ul
ly
c
onne
c
te
d
la
ye
r
s
,
de
s
ig
ne
d
to
de
te
c
t
pl
a
nt
le
a
f
di
s
e
a
s
e
s
.
A
lg
or
it
hm
1 s
how
s
th
e
A
C
O
-
C
N
N
.
A
lg
or
it
hm
1.
A
C
O
-
C
N
N
1.
L
oa
d
d
a
ta
s
e
t
(
le
a
f
i
m
a
ge
s
)
L
=
{
L
1, L
2, L
3,
…}
2.
P
r
e
pr
oc
e
s
s
in
g I
m
a
ge
s
L
pp=
L
-
k
3.
F
e
a
tu
r
e
e
xt
r
a
c
ti
on
S
e
t
th
e
i
nf
e
c
te
d por
ti
on'
s
be
gi
nni
ng point
t
o z
e
r
o
I
f
(
A
nt
m
ove
s
on t
o t
he
ne
xt
pos
it
io
n.)
G
a
th
e
r
t
he
s
ubs
e
t
U
s
in
g
(
1
)
, l
oc
a
te
t
he
a
r
e
a
of
t
he
l
e
a
f
t
ha
t
is
a
f
f
e
c
te
d.
E
ls
e
U
s
in
g
(
2
)
, i
de
nt
if
y t
he
ne
xt
c
om
pone
nt
C
ont
in
ue
unt
il
t
he
s
to
ppi
ng c
ondi
ti
on i
s
s
a
ti
s
f
ie
d.
E
ndi
f
4.
R
e
tu
r
n
O
ut
put
:
H
e
a
lt
hy a
nd unhe
a
lt
hy l
e
a
f
c
la
s
s
if
ic
a
ti
on
4.
R
E
S
U
L
T
S
A
N
D
D
I
S
C
U
S
S
I
O
N
W
e
im
pl
e
m
e
nt
e
d
th
e
s
ugge
s
te
d
a
ppr
oa
c
h
us
in
g
P
yt
hon
3.7.10
a
nd
a
s
s
e
s
s
it
a
ga
in
s
t
ot
h
e
r
w
e
ll
-
known,
s
ta
te
-
of
-
th
e
-
a
r
t
m
a
c
hi
ne
-
le
a
r
ni
ng
im
a
ge
c
la
s
s
i
f
ie
r
s
f
or
pl
a
nt
di
s
e
a
s
e
pr
e
di
c
ti
on.
D
if
f
e
r
e
nt
da
ta
s
e
ts
f
or
r
ic
e
a
nd
pot
a
to
e
s
w
e
r
e
us
e
d
to
tr
a
in
th
e
s
e
m
ode
ls
,
a
nd
th
e
m
ode
l
pe
r
f
or
m
a
nc
e
pa
r
a
m
e
te
r
s
w
e
r
e
th
e
n c
om
pa
r
e
d. I
n c
om
pa
r
is
on t
o e
xi
s
ti
ng c
la
s
s
if
ie
r
s
, t
he
s
ugge
s
te
d C
N
N
m
ode
l
out
pe
r
f
or
m
e
d t
he
da
ta
s
e
ts
f
or
r
ic
e
a
nd
pot
a
to
e
s
.
F
ig
ur
e
3
di
s
pl
a
y
th
e
C
N
N
m
ode
l
a
r
c
hi
te
c
t
ur
e
s
a
nd
hyp
e
r
-
pa
r
a
m
e
te
r
s
e
tu
ps
is
s
how
n
in
T
a
bl
e
2.
I
n
de
e
p
ne
ur
a
l
ne
twor
k
,
v
a
lu
a
bl
e
e
le
m
e
nt
s
f
r
om
th
e
in
put
im
a
ge
a
r
e
e
xt
r
a
c
te
d
to
pr
oduc
e
pr
e
c
i
s
e
pr
e
di
c
ti
ons
.
T
he
a
ppl
ic
a
ti
on
of
f
il
te
r
s
to
th
e
in
put
im
a
ge
a
t
e
a
c
h
la
ye
r
of
th
e
C
N
N
m
ode
l
pr
oduc
e
s
a
c
ti
va
ti
on
m
a
ps
or
f
e
a
tu
r
e
m
a
ps
.
T
a
bl
e
2
. P
a
r
a
m
e
te
r
s
f
or
C
N
N
m
ode
l
tr
a
in
in
g
P
a
r
a
m
e
t
e
r
s
V
a
l
ue
S
i
z
e
od ba
t
c
h
32
A
c
t
i
va
t
i
on f
unc
t
i
on
S
of
t
M
a
x, R
e
LU
M
e
t
r
i
c
s
A
c
c
ur
a
c
y
L
os
s
S
pa
r
s
e
_c
a
t
e
gor
i
c
a
l
_c
r
os
s
e
nt
r
opy
O
pt
i
m
i
z
e
r
A
da
m
a
nd l
e
a
r
ni
ng r
a
t
e
(
l
r
)
=0.0001
F
e
a
tu
r
e
e
xt
r
a
c
ti
on
r
e
tr
ie
ve
d
f
o
r
a
pa
r
ti
c
ul
a
r
pi
c
tu
r
e
c
a
n
be
ga
in
e
d
by
a
na
ly
z
in
g
th
e
out
put
a
c
ti
va
ti
on
m
a
ps
of
e
a
c
h
la
ye
r
.
T
o
unde
r
s
ta
nd
th
e
m
od
e
l'
s
in
ne
r
w
or
ki
ng
s
f
or
a
gi
ve
n
in
put
a
t
a
c
e
r
ta
in
la
ye
r
,
on
e
c
a
n
lo
ok
a
t
th
e
a
c
ti
va
ti
on
m
a
p
or
th
e
f
il
te
r
.
E
a
c
h
m
ode
l
la
ye
r
ga
th
e
r
s
th
e
c
ha
r
a
c
te
r
is
ti
c
s
m
a
ps
f
or
a
s
pe
c
if
ic
in
put
im
a
ge
of
pot
a
to
e
s
,
a
s
s
how
n
in
F
ig
ur
e
3.
A
s
c
a
n
be
obs
e
r
ve
d
i
n
F
ig
ur
e
3,
th
e
f
ir
s
t
la
ye
r
ke
e
ps
pr
a
c
ti
c
a
ll
y
a
ll
of
th
e
de
ta
il
s
or
in
f
or
m
a
ti
on
f
r
om
th
e
or
ig
in
a
l
in
put
te
d
im
a
ge
,
in
c
lu
di
ng
th
e
c
om
pl
e
te
s
ha
pe
of
th
e
le
a
f
.
F
e
a
tu
r
e
s
li
ke
s
in
gl
e
bor
de
r
s
,
c
or
ne
r
s
,
a
nd
a
ngl
e
s
a
r
e
e
xt
r
a
c
te
d
f
r
om
de
e
pe
r
la
ye
r
s
. T
hi
s
m
e
a
n
s
th
a
t
in
or
de
r
to
c
la
s
s
if
y
im
a
ge
s
,
de
e
pe
r
la
ye
r
s
c
a
n
a
c
c
e
s
s
m
or
e
r
e
le
va
nt
d
a
ta
.
T
e
s
ti
ng
a
nd
tr
a
in
in
g
ha
ve
be
e
n
c
om
pl
e
te
d
on
th
e
s
ugge
s
te
d C
N
N
m
ode
l.
A
ppl
yi
ng
th
e
"
s
pa
r
s
e
_c
a
te
gor
ic
a
l_
c
r
os
s
e
nt
r
opy"
lo
s
s
f
unc
ti
on
a
nd
us
in
g
a
c
c
ur
a
c
y
a
s
t
he
m
e
tr
ic
, t
he
A
da
m
K
e
r
a
s
opt
im
iz
e
r
i
s
ut
il
iz
e
d w
it
h a
l
e
a
r
n
in
g r
a
te
of
0.0001.
T
he
pr
opos
e
d
m
e
th
od
ha
s
be
e
n
e
va
lu
a
te
d
ut
il
iz
in
g
im
a
ge
s
of
th
e
ga
th
e
r
e
d
le
a
f
s
a
m
pl
e
s
.
T
o
di
f
f
e
r
e
nt
ia
te
be
twe
e
n
he
a
lt
hy
a
nd
unhe
a
lt
hy
le
a
ve
s
,
th
e
pr
op
os
e
d
te
c
hni
que
us
e
s
A
C
O
-
C
N
N
.
F
our
w
id
e
ly
us
e
d
e
va
lu
a
ti
on
m
e
a
s
ur
e
s
c
a
t
e
gor
iz
a
ti
on
r
e
c
a
ll
,
pr
e
c
is
io
n,
a
c
c
u
r
a
c
y,
a
nd
F
1
s
c
or
e
a
r
e
e
xa
m
in
e
d
in
th
e
s
tu
dy.
T
he
s
uppl
ie
d
le
a
f
im
a
ge
s
a
r
e
a
c
c
ur
a
te
ly
r
e
pr
oduc
e
d
by
pr
e
c
is
io
n
.
P
r
e
c
is
io
n
is
a
m
e
a
s
ur
e
of
how
w
e
ll
a
c
la
s
s
if
ie
r
pe
r
f
or
m
s
.
P
r
e
c
is
io
n
is
in
c
r
e
a
s
e
d
w
he
n
th
e
r
e
a
r
e
f
e
w
e
r
pos
it
iv
e
s
ig
na
ls
f
r
om
th
e
pl
a
nt
le
a
f
,
w
hi
le
pr
e
c
is
io
n i
s
de
c
r
e
a
s
e
d w
he
n t
he
r
e
a
r
e
m
or
e
pos
it
iv
e
s
ig
na
ls
. T
he
r
e
c
a
ll
r
a
te
s
t
he
pe
r
f
or
m
a
nc
e
of
t
he
c
la
s
s
if
ie
r
.
T
he
i
nc
r
e
a
s
e
of
pos
it
iv
e
s
a
m
pl
e
s
f
ound c
or
r
e
s
ponds
w
it
h t
he
r
e
c
a
ll
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
20
25
:
3724
-
3733
3730
=
+
+
+
+
(
6)
=
+
(
7)
=
+
(
8)
R
e
ga
r
di
ng r
e
c
a
ll
, pr
e
c
is
io
n, a
nd a
c
c
ur
a
c
y, t
he
A
C
O
-
C
N
N
m
ode
l
pe
r
f
or
m
e
d be
tt
e
r
t
ha
n t
he
S
V
M
a
nd
C
N
N
m
ode
ls
.
C
N
N
'
s
r
a
te
is
81.6%
a
s
oppos
e
d
to
S
V
M
'
s
8
0%
a
c
c
ur
a
c
y
r
a
te
.
W
it
h
a
n
a
c
c
ur
a
c
y
r
a
te
of
91.00%
,
th
e
A
C
O
-
C
N
N
m
ode
l
h
a
s
th
e
hi
ghe
s
t
F
1
-
s
c
or
e
va
lu
e
,
a
nd
th
e
m
e
th
odol
ogy
a
s
a
w
hol
e
out
pe
r
f
or
m
s
c
om
pe
ti
ng
m
ode
ls
in
te
r
m
s
of
r
e
c
a
ll
,
pr
e
c
is
io
n,
a
nd
a
c
c
ur
a
c
y.
F
ig
ur
e
4
s
how
s
th
e
m
ode
l
pe
r
f
or
m
a
nc
e
,
F
ig
ur
e
4
(
a
)
s
how
s
th
e
pe
r
f
or
m
a
nc
e
m
e
tr
ic
s
a
nd
F
ig
ur
e
4
(
b
)
s
how
s
th
e
c
onf
us
io
n
m
a
tr
ix
,
va
li
da
t
e
s
th
e
s
e
f
in
di
ngs
.
T
he
be
s
t
a
c
c
ur
a
c
y
of
95%
is
pr
ovi
de
d
by
th
e
pr
opo
s
e
d
a
lg
or
it
hm
f
or
pot
a
to
e
a
r
ly
bl
ig
ht
is
95%
,
w
hi
le
t
he
a
c
c
ur
a
c
y
f
or
r
ic
e
ba
c
te
r
ia
l
bl
ig
ht
, a
nd pe
ppe
r
l
e
a
f
s
pot
i
s
91%
a
nd 89%
, r
e
s
pe
c
ti
ve
ly
.
F
ig
ur
e
3
.
C
N
N
m
ode
l
di
a
gr
a
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
e
d c
onv
ol
ut
io
n ne
ur
al
n
e
tw
or
k
w
it
h ant c
ol
ony
al
gor
it
h
m
f
or
ac
c
ur
at
e
pl
ant
…
(
Shw
e
ta
B
ondr
e
)
3731
(
a
)
(
b)
F
ig
ur
e
4
.
M
ode
l
p
e
r
f
or
m
a
nc
e
of
(a
)
pe
r
f
or
m
a
nc
e
m
e
tr
ic
s
a
nd
(
b)
c
onf
us
io
n m
a
tr
ix
5.
C
O
N
C
L
U
S
I
O
N
T
hi
s
pa
pe
r
de
te
c
te
d
di
s
e
a
s
e
s
in
pl
a
nt
le
a
ve
s
us
in
g
a
n
A
C
O
-
C
N
N
m
ode
l.
D
a
ta
s
e
ts
f
or
pot
a
to
e
s
,
pe
ppe
r
s
,
a
nd
r
ic
e
w
e
r
e
us
e
d
in
th
is
in
ve
s
ti
ga
ti
on.
F
or
pot
a
to
le
a
f
di
s
e
a
s
e
,
our
m
ode
l
r
e
a
c
he
d
a
de
te
c
ti
on
a
c
c
ur
a
c
y
of
95.00%
.
A
to
ta
l
of
5,932
r
ic
e
phot
ogr
a
phs
a
nd
1
,500
im
a
ge
s
of
di
s
e
a
s
e
d
a
nd
he
a
lt
hy
pe
ppe
r
,
pot
a
to
,
a
nd
r
ic
e
le
a
ve
s
w
e
r
e
u
s
e
d
in
th
is
in
v
e
s
ti
ga
ti
on.
W
he
n
c
om
pa
r
e
d
to
e
xi
s
ti
ng
to
p
-
ti
e
r
m
a
c
hi
ne
le
a
r
ni
ng
pi
c
tu
r
e
c
la
s
s
if
ie
r
s
li
ke
S
V
M
a
nd C
N
N
,
th
e
pr
opos
e
d
m
ode
l
de
m
ons
tr
a
te
d
s
ig
ni
f
ic
a
nt
ly
hi
ghe
r
a
c
c
ur
a
c
y.
C
N
N
c
la
s
s
if
ie
r
w
a
s
ut
il
iz
e
d f
or
t
he
or
ga
ni
z
a
ti
on, whe
r
e
a
s
A
C
O
w
a
s
e
m
pl
oye
d f
or
f
e
a
tu
r
e
e
xt
r
a
c
ti
on. T
he
s
ugge
s
te
d
m
e
th
od
is
us
e
d
to
di
s
ti
ngui
s
h
be
twe
e
n
in
f
e
c
te
d
a
nd
he
a
lt
hy
le
a
ve
s
u
s
in
g
opt
im
iz
a
ti
on
a
nd
s
e
a
r
c
h
a
lg
or
it
hm
s
in
s
pi
r
e
d
by
na
tu
r
e
,
hype
r
pa
r
a
m
e
te
r
s
o
f
th
e
p
r
opos
e
d
C
N
N
m
ode
l,
in
c
lu
di
ng
ba
tc
h
s
iz
e
,
num
be
r
o
f
e
poc
hs
,
qua
nt
it
y
of
c
onvolut
io
na
l
la
ye
r
s
,
a
c
ti
va
ti
on
f
unc
ti
ons
f
or
c
on
vol
ut
io
na
l
la
ye
r
s
,
num
be
r
of
f
ul
ly
c
onne
c
te
d
la
ye
r
s
,
num
be
r
of
ne
ur
ons
pe
r
la
ye
r
,
a
nd
f
il
te
r
di
m
e
ns
io
ns
,
m
a
y
be
tu
ne
d
in
f
ut
ur
e
s
tu
di
e
s
.
T
he
pe
r
f
or
m
a
nc
e
of
t
he
m
ode
l
m
a
y be
f
ur
th
e
r
i
m
pr
ove
d by c
a
r
e
f
ul
c
hoi
c
e
of
t
he
s
e
hype
r
-
pa
r
a
m
e
te
r
s
.
F
U
N
D
I
N
G
I
N
F
O
R
M
A
T
I
O
N
A
ut
hor
s
s
ta
te
no f
undi
ng i
nvol
ve
d.
A
U
T
H
O
R
C
O
N
T
R
I
B
U
T
I
O
N
S
S
T
A
T
E
M
E
N
T
T
hi
s
jo
ur
na
l
us
e
s
th
e
C
on
tr
ib
ut
or
R
ol
e
s
T
a
xonomy
(
C
R
e
di
T
)
to
r
e
c
ogni
z
e
in
di
vi
dua
l
a
ut
hor
c
ont
r
ib
ut
io
ns
, r
e
duc
e
a
ut
hor
s
hi
p di
s
put
e
s
,
a
nd f
a
c
il
it
a
te
c
ol
la
bo
r
a
ti
on.
N
am
e
o
f
A
u
t
h
or
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
S
hw
e
ta
V
. B
ondr
e
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
U
m
a
Y
a
da
v
✓
✓
✓
✓
✓
✓
V
ip
in
D
. B
ondr
e
✓
✓
✓
✓
✓
P
oor
va
A
gr
a
w
a
l
✓
✓
✓
✓
✓
C
:
C
onc
e
pt
ua
l
i
z
a
t
i
on
M
:
M
e
t
hodol
ogy
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
da
t
i
on
Fo
:
Fo
r
m
a
l
a
na
l
ys
i
s
I
:
I
nve
s
t
i
ga
t
i
on
R
:
R
e
s
our
c
e
s
D
:
D
a
t
a
C
ur
a
t
i
on
O
:
W
r
i
t
i
ng
-
O
r
i
gi
na
l
D
r
a
f
t
E
:
W
r
i
t
i
ng
-
R
e
vi
e
w
&
E
di
t
i
ng
Vi
:
Vi
s
ua
l
i
z
a
t
i
on
Su
:
Su
pe
r
vi
s
i
on
P
:
P
r
oj
e
c
t
a
dm
i
ni
s
t
r
a
t
i
on
Fu
:
Fu
ndi
ng a
c
qui
s
i
t
i
on
C
O
N
F
L
I
C
T
O
F
I
N
T
E
R
E
S
T
S
T
A
T
E
M
E
N
T
T
he
a
ut
hor
s
de
c
la
r
e
no c
onf
li
c
ts
of
i
nt
e
r
e
s
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
, V
ol
.
14
, N
o.
5
,
O
c
to
be
r
20
25
:
3724
-
3733
3732
D
A
T
A
A
V
A
I
L
A
B
I
L
I
T
Y
T
he
da
ta
th
a
t
s
uppor
t
th
e
f
in
di
ngs
of
th
is
s
tu
dy
a
r
e
ope
n
a
va
il
a
bl
e
in
K
a
ggl
e
a
t
ht
tp
s
:/
/ww
w
.ka
ggl
e
.c
om
/d
a
ta
s
e
ts
/a
bda
ll
a
h
a
li
de
v/
pl
a
nt
vi
ll
a
ge
-
da
ta
s
e
t.
R
E
F
E
R
E
N
C
E
S
[
1]
M
. C
.
H
unt
e
r
, R
.
G
.
S
m
i
t
h, M
.
E
. S
c
hi
pa
ns
ki
,
L
.
W
. A
t
w
ood,
a
nd
D
. A
.
M
or
t
e
ns
e
n, “
A
gr
i
c
ul
t
ur
e
i
n
2050:
r
e
c
a
l
i
br
a
t
i
ng t
a
r
ge
t
s
f
or
s
us
t
a
i
na
bl
e
i
nt
e
n
s
i
f
i
c
a
t
i
on,”
B
i
oSc
i
e
nc
e
, vol
. 67, no. 4, pp. 386
–
391, A
pr
. 2017, doi
:
10.1093/
bi
os
c
i
/
bi
x010.
[
2]
J
. E
. V
.
D
.
W
a
a
l
s
, L
. K
or
s
t
e
n, a
nd B
. S
l
i
ppe
r
s
,
“
G
e
ne
t
i
c
di
ve
r
s
i
t
y a
m
ong A
l
t
e
r
n
a
r
i
a
s
ol
a
ni
i
s
ol
a
t
e
s
f
r
om
pot
a
t
oe
s
i
n S
out
h A
f
r
i
c
a
,”
P
l
ant
D
i
s
e
as
e
, vol
. 88, no. 9, pp. 959
–
964, S
e
p. 2004, doi
:
10.1094/
P
D
I
S
.2004.88.9.959.
[
3]
V
.
S
um
a
,
R
.
A
.
S
he
t
t
y,
R
.
F
.
T
a
t
e
d,
S
.
R
oha
n,
a
nd
T
.
S
.
P
uj
a
r
,
“
C
N
N
ba
s
e
d
l
e
a
f
di
s
e
a
s
e
i
de
nt
i
f
i
c
a
t
i
on
a
nd
r
e
m
e
dy
r
e
c
om
m
e
nda
t
i
on
s
ys
t
e
m
,”
i
n
P
r
oc
e
e
di
ngs
of
t
he
3r
d
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
E
l
e
c
t
r
oni
c
s
and
C
om
m
uni
c
at
i
on
and
A
e
r
os
pac
e
T
e
c
hnol
ogy
, I
C
E
C
A
2019
, I
E
E
E
,
J
un. 2019, pp. 395
–
399
, doi
:
10.1109/
I
C
E
C
A
.
2019.8821872.
[
4]
M
.
A
.
J
a
s
i
m
a
nd
J
.
M
.
A
.
-
T
uw
a
i
j
a
r
i
,
“
P
l
a
nt
l
e
a
f
d
i
s
e
a
s
e
s
de
t
e
c
t
i
on
a
nd
c
l
a
s
s
i
f
i
c
a
t
i
on
us
i
ng
i
m
a
ge
pr
oc
e
s
s
i
ng
a
nd
de
e
p
l
e
a
r
ni
ng
t
e
c
hni
que
s
,”
i
n
P
r
oc
e
e
di
ng
s
of
t
he
2020
I
nt
e
r
nat
i
onal
C
onf
e
r
e
n
c
e
on
C
om
put
e
r
Sc
i
e
nc
e
and
Sof
t
w
a
r
e
E
ngi
ne
e
r
i
ng,
C
SA
SE
2020
,
2020, pp. 259
–
265
, doi
:
10.1109/
C
S
A
S
E
48920.2020.9142097.
[
5]
S
.
V
.
M
i
l
i
t
a
nt
e
,
B
.
D
.
G
e
r
a
r
do,
a
nd
N
.
V
.
D
I
oni
s
i
o,
“
P
l
a
nt
l
e
a
f
de
t
e
c
t
i
on
a
nd
di
s
e
a
s
e
r
e
c
ogni
t
i
on
us
i
ng
de
e
p
l
e
a
r
ni
ng,”
i
n
2019
I
E
E
E
E
ur
as
i
a
C
onf
e
r
e
nc
e
on
I
O
T
,
C
om
m
uni
c
at
i
on
and
E
ngi
ne
e
r
i
ng,
E
C
I
C
E
2019
,
I
E
E
E
,
O
c
t
.
2019,
pp.
579
–
582
,
doi
:
10.1109/
E
C
I
C
E
47484.2019.8942686.
[
6]
M
.
A
ga
r
w
a
l
,
A
.
S
i
ngh,
S
.
A
r
j
a
r
i
a
,
A
.
S
i
nha
,
a
nd
S
.
G
upt
a
,
“
T
oL
e
D
:
t
om
a
t
o
l
e
a
f
di
s
e
a
s
e
de
t
e
c
t
i
on
us
i
ng
c
onvol
ut
i
on
ne
ur
a
l
ne
t
w
or
k,”
i
n
P
r
oc
e
di
a C
om
put
e
r
Sc
i
e
nc
e
, 2020, pp. 293
–
301
, doi
:
10.1016/
j
.pr
oc
s
.2020.03.225.
[
7]
S
.
A
r
ya
a
nd
R
.
S
i
ngh,
“
A
c
om
pa
r
a
t
i
ve
s
t
udy
of
C
N
N
a
nd
A
l
e
xN
e
t
f
or
de
t
e
c
t
i
on
of
di
s
e
a
s
e
i
n
pot
a
t
o
a
nd
m
a
ngo
l
e
a
f
,”
i
n
I
E
E
E
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
I
s
s
u
e
s
and
C
hal
l
e
nge
s
i
n
I
nt
e
l
l
i
ge
nt
C
om
put
i
ng
T
e
c
hni
que
s
,
I
C
I
C
T
2019
,
I
E
E
E
,
S
e
p.
2019,
pp.
1
–
6
,
doi
:
10.1109/
I
C
I
C
T
46931.2019.8977648.
[
8]
K
.
A
ur
a
ngz
e
b,
F
.
A
km
a
l
,
M
.
A
.
K
h
a
n,
M
.
S
ha
r
i
f
,
a
nd
M
.
Y
.
J
a
ve
d,
“
A
dva
n
c
e
d
m
a
c
hi
ne
l
e
a
r
ni
ng
a
l
gor
i
t
hm
ba
s
e
d
s
ys
t
e
m
f
or
c
r
ops
l
e
a
f
di
s
e
a
s
e
s
r
e
c
ogni
t
i
on,”
i
n
P
r
oc
e
e
di
ngs
-
2020
6t
h
C
onf
e
r
e
n
c
e
on
D
at
a
Sc
i
e
nc
e
and
M
ac
hi
ne
L
e
a
r
ni
ng
A
ppl
i
c
at
i
ons
,
C
D
M
A
2020
, 2020, pp. 146
–
151
, doi
:
10.1109/
C
D
M
A
47397.2020.00031.
[
9]
S
.
A
.
W
a
gl
e
a
nd
R
.
H
a
r
i
kr
i
s
hna
n,
“
C
om
pa
r
i
s
on
of
pl
a
nt
l
e
a
f
c
l
a
s
s
i
f
i
c
a
t
i
on
us
i
ng
m
odi
f
i
e
d
A
l
e
x
N
e
t
a
nd
s
uppor
t
ve
c
t
or
m
a
c
hi
ne
,
”
T
r
ai
t
e
m
e
nt
du Si
gnal
, vol
. 38, no. 1, pp. 79
–
87, F
e
b. 2021, doi
:
10.18280/
T
S
.38
0108.
[
10]
M
.
R
.
M
i
a
,
S
.
R
oy,
S
.
K
.
D
a
s
,
a
nd
M
.
A
.
R
a
hm
a
n,
“
M
a
ngo
l
e
a
f
di
s
e
a
s
e
r
e
c
ogni
t
i
on
us
i
ng
ne
ur
a
l
ne
t
w
or
k
a
nd
s
uppor
t
ve
c
t
or
m
a
c
hi
ne
,”
I
r
an J
our
nal
of
C
o
m
put
e
r
Sc
i
e
n
c
e
, vol
. 3, no. 3, pp. 185
–
193, S
e
p. 2
020, doi
:
10.1007/
s
42044
-
020
-
00057
-
z.
[
11]
S
.
P
.
M
oha
nt
y,
D
.
P
.
H
ughe
s
,
a
nd
M
.
S
a
l
a
t
hé
,
“
U
s
i
ng
de
e
p
l
e
a
r
ni
ng
f
or
i
m
a
ge
-
ba
s
e
d
pl
a
nt
di
s
e
a
s
e
de
t
e
c
t
i
on,”
F
r
ont
i
e
r
s
i
n
P
l
ant
Sc
i
e
nc
e
, vol
. 7, no. S
e
pt
e
m
be
r
, pp. 1
–
10, 2016, doi
:
10.3389/
f
pl
s
.2016.01419.
[
12]
F
.
I
s
l
a
m
,
M
.
N
.
H
oq,
a
nd
C
.
M
.
R
a
hm
a
n,
“
A
ppl
i
c
a
t
i
on
of
t
r
a
ns
f
e
r
l
e
a
r
ni
ng
t
o
de
t
e
c
t
pot
a
t
o
di
s
e
a
s
e
f
r
om
l
e
a
f
i
m
a
ge
,”
i
n
2019
I
E
E
E
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
R
obot
i
c
s
,
A
ut
om
at
i
on,
A
r
t
i
f
i
c
i
al
-
I
nt
e
l
l
i
ge
n
c
e
and
I
nt
e
r
n
e
t
-
of
-
T
hi
ngs
,
R
A
A
I
C
O
N
2019
,
I
E
E
E
,
N
ov. 2019, pp. 127
–
130
, doi
:
10.1109/
R
A
A
I
C
O
N
48939.2019.53.
[
13]
N
.
K
.
E
.,
K
.
M
.,
P
.
P
.,
A
.
R
.,
a
nd
V
.
S
.,
“
T
om
a
t
o
l
e
a
f
di
s
e
a
s
e
de
t
e
c
t
i
o
n
us
i
ng
c
onvol
ut
i
ona
l
ne
ur
a
l
ne
t
w
o
r
k
w
i
t
h
da
t
a
a
ugm
e
nt
a
t
i
on,”
i
n
2020
5t
h
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
C
om
m
uni
c
at
i
on
and
E
l
e
c
t
r
oni
c
s
S
y
s
t
e
m
s
,
I
E
E
E
,
J
un.
2020,
pp. 1125
–
1132
, doi
:
10.1109/
i
c
c
e
s
48766.2020.9138030.
[
14]
A
.
F
ue
nt
e
s
,
S
.
Y
oon,
S
.
C
.
K
i
m
,
a
nd
D
.
S
.
P
a
r
k,
“
A
r
obus
t
d
e
e
p
-
l
e
a
r
ni
ng
-
ba
s
e
d
de
t
e
c
t
or
f
or
r
e
a
l
-
t
i
m
e
t
om
a
t
o
pl
a
nt
di
s
e
a
s
e
s
a
nd
pe
s
t
s
r
e
c
ogni
t
i
on,”
Se
n
s
or
s
, vol
. 17, no. 9, 2017, doi
:
10.3390/
s
17092022.
[
15]
L
.
L
i
,
S
.
Z
ha
ng,
a
nd
B
.
W
a
ng,
“
P
l
a
nt
di
s
e
a
s
e
de
t
e
c
t
i
on
a
nd
c
l
a
s
s
i
f
i
c
a
t
i
on
b
y
de
e
p
l
e
a
r
ni
ng
-
a
r
e
vi
e
w
,”
I
E
E
E
A
c
c
e
s
s
,
vol
.
9,
pp. 56683
–
56698, 2021, doi
:
10.1109/
A
C
C
E
S
S
.2021.3069646.
[
16]
M
.
E
.
P
ot
he
n
a
nd
D
.
M
.
L
.
P
a
i
,
“
D
e
t
e
c
t
i
on
of
r
i
c
e
l
e
a
f
di
s
e
a
s
e
s
us
i
ng
i
m
a
ge
p
r
oc
e
s
s
i
ng,”
i
n
P
r
oc
e
e
di
ng
s
of
t
he
4t
h
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
C
om
put
i
ng
M
e
t
hodol
ogi
e
s
and
C
om
m
uni
c
at
i
on,
I
C
C
M
C
2020
,
I
E
E
E
,
M
a
r
.
2020,
pp.
424
–
430
,
doi
:
10.1109/
I
C
C
M
C
48092.2020.I
C
C
M
C
-
00080.
[
17]
J
.
N
.
R
e
ddy,
K
.
V
i
nod,
a
nd
A
.
S
.
R
.
A
j
a
i
,
“
A
na
l
ys
i
s
of
c
l
a
s
s
i
f
i
c
a
t
i
on
a
l
gor
i
t
h
m
s
f
or
pl
a
nt
l
e
a
f
di
s
e
a
s
e
de
t
e
c
t
i
on,”
i
n
P
r
oc
e
e
di
ngs
of
2019
3r
d
I
E
E
E
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
E
l
e
c
t
r
i
c
al
,
C
om
put
e
r
and
C
o
m
m
uni
c
at
i
on
T
e
c
hnol
ogi
e
s
,
I
C
E
C
C
T
2019
,
I
E
E
E
,
F
e
b. 2019, pp. 1
–
6
, doi
:
10.1109/
I
C
E
C
C
T
.2019.8869090.
[
18]
M
.
I
s
l
a
m
,
A
.
D
i
nh,
K
.
W
a
hi
d,
a
nd
P
.
B
how
m
i
k,
“
D
e
t
e
c
t
i
on
of
pot
a
t
o
di
s
e
a
s
e
s
us
i
ng
i
m
a
ge
s
e
gm
e
nt
a
t
i
on
a
nd
m
ul
t
i
c
l
a
s
s
s
uppor
t
ve
c
t
or
m
a
c
hi
ne
,”
i
n
C
anadi
an C
onf
e
r
e
nc
e
on E
l
e
c
t
r
i
c
al
and
C
om
put
e
r
E
ngi
ne
e
r
i
ng
, 2017
, doi
:
10.1109/
C
C
E
C
E
.2017.7946594.
[
19]
K
.
P
e
r
ve
e
n
e
t
al
.
,
“
M
ul
t
i
di
m
e
ns
i
ona
l
a
t
t
e
nt
i
on
-
ba
s
e
d
C
N
N
m
od
e
l
f
or
i
de
nt
i
f
yi
ng
a
ppl
e
l
e
a
f
di
s
e
a
s
e
,”
J
our
nal
of
F
ood
Q
ual
i
t
y
,
vol
. 2023, pp. 1
–
12, F
e
b. 2023, doi
:
10.1155/
2023/
9504186.
[
20]
M
.
J
.
A
.
S
oe
b
e
t
al
.
,
“
T
e
a
l
e
a
f
di
s
e
a
s
e
de
t
e
c
t
i
on
a
nd
i
de
nt
i
f
i
c
a
t
i
on
ba
s
e
d
on
Y
O
L
O
v7
(
Y
O
L
O
-
T
)
,”
Sc
i
e
nt
i
f
i
c
R
e
por
t
s
,
vol
.
13,
no. 1, A
pr
. 2023, doi
:
10.1038/
s
41598
-
023
-
33270
-
4.
[
21]
A
.
K
.
S
i
ngh,
S
.
V
.
N
.
S
r
e
e
ni
va
s
u,
U
.
S
.
B
.
K
.
M
a
ha
l
a
xm
i
,
H
.
S
ha
r
m
a
,
D
.
D
.
P
a
t
i
l
,
a
nd
E
.
A
s
e
ns
o,
“
H
ybr
i
d
f
e
a
t
ur
e
-
ba
s
e
d
di
s
e
a
s
e
de
t
e
c
t
i
on
i
n
pl
a
nt
l
e
a
f
u
s
i
ng
c
onvol
ut
i
ona
l
n
e
ur
a
l
ne
t
w
or
k,
ba
y
e
s
i
a
n
opt
i
m
i
z
e
d
S
V
M
,
a
nd
r
a
ndom
f
or
e
s
t
c
l
a
s
s
i
f
i
e
r
,”
J
ou
r
nal
of
F
ood Q
ual
i
t
y
, vol
. 2022, pp. 1
–
16, F
e
b. 2022, doi
:
10.1155/
2022/
2845320.
[
22]
J
.
A
ndr
e
w
,
J
.
E
uni
c
e
,
D
.
E
.
P
ope
s
c
u,
M
.
K
.
C
how
da
r
y,
a
nd
J
.
H
e
m
a
nt
h,
“
D
e
e
p
l
e
a
r
ni
ng
-
ba
s
e
d
l
e
a
f
di
s
e
a
s
e
de
t
e
c
t
i
on
i
n
c
r
ops
us
i
ng i
m
a
ge
s
f
or
a
gr
i
c
ul
t
ur
a
l
a
ppl
i
c
a
t
i
on,”
A
gr
onom
y
, vol
. 12, no. 10, O
c
t
. 202
2, doi
:
10.3390/
a
gr
onom
y12102395.
[
23]
Z
.
ur
R
e
hm
a
n
e
t
al
.
,
“
R
e
c
ogni
z
i
ng
a
ppl
e
l
e
a
f
di
s
e
a
s
e
s
us
i
ng
a
nove
l
pa
r
a
l
l
e
l
r
e
a
l
-
t
i
m
e
p
r
oc
e
s
s
i
ng
f
r
a
m
e
w
or
k
ba
s
e
d
on
M
A
S
K
R
C
N
N
a
nd
t
r
a
ns
f
e
r
l
e
a
r
ni
ng:
a
n
a
ppl
i
c
a
t
i
on
f
or
s
m
a
r
t
a
gr
i
c
ul
t
ur
e
,”
I
E
T
I
m
age
P
r
oc
e
s
s
i
ng
,
vol
.
15,
no.
10,
pp.
2157
–
2168
,
A
ug. 2021, doi
:
10.1049/
i
pr
2.12183.
[
24]
P
.
M
a
s
ki
a
nd
A
.
T
hondi
ya
t
h,
“
P
l
a
nt
di
s
e
a
s
e
de
t
e
c
t
i
on
us
i
ng
a
dva
nc
e
d
de
e
p
l
e
a
r
ni
ng
a
l
gor
i
t
hm
s
:
a
c
a
s
e
s
t
udy
of
pa
pa
ya
r
i
ng
s
po
t
di
s
e
a
s
e
,
”
i
n
2021
6t
h
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
I
m
age
,
V
i
s
i
on
and
C
om
put
i
ng,
I
C
I
V
C
2021
,
I
E
E
E
,
J
ul
.
2021,
pp.
49
–
54
,
doi
:
10.1109/
I
C
I
V
C
52351.2021.9526944.
[
25]
Z
.
L
i
u,
X
.
X
i
a
ng,
J
.
Q
i
n,
Y
.
T
a
n,
Q
.
Z
ha
ng,
a
nd
N
.
N
.
X
i
ong,
“
I
m
a
ge
r
e
c
ogni
t
i
on
of
c
i
t
r
us
di
s
e
a
s
e
s
ba
s
e
d
on
de
e
p
l
e
a
r
ni
ng,”
C
om
put
e
r
s
, M
at
e
r
i
al
s
and C
ont
i
nua
, vol
. 66, no. 1, pp. 457
–
466, 2021, doi
:
10.
32604/
c
m
c
.2020.012165.
[
26]
X
.
X
i
e
,
Y
.
M
a
,
B
.
L
i
u,
J
.
H
e
,
S
.
L
i
,
a
nd
H
.
W
a
ng,
“
A
de
e
p
-
l
e
a
r
ni
ng
-
ba
s
e
d
r
e
a
l
-
t
i
m
e
de
t
e
c
t
or
f
or
g
r
a
pe
l
e
a
f
di
s
e
a
s
e
s
us
i
n
g
i
m
pr
ove
d c
onvol
ut
i
ona
l
ne
ur
a
l
ne
t
w
or
ks
,”
F
r
ont
i
e
r
s
i
n P
l
ant
Sc
i
e
nc
e
, vol
. 11, J
un. 2020, doi
:
10.3389/
f
pl
s
.2020.00751.
[
27]
A
.
R
a
m
c
ha
r
a
n
e
t
al
.
,
“
A
m
obi
l
e
-
ba
s
e
d
de
e
p
l
e
a
r
ni
ng
m
ode
l
f
or
c
a
s
s
a
va
di
s
e
a
s
e
di
a
gnos
i
s
,
”
F
r
ont
i
e
r
s
i
n
P
l
ant
Sc
i
e
nc
e
,
vol
.
10,
M
a
r
. 2019, doi
:
10.3389/
f
pl
s
.2019.00272.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
O
pt
imi
z
e
d c
onv
ol
ut
io
n ne
ur
al
n
e
tw
or
k
w
it
h ant c
ol
ony
al
gor
it
h
m
f
or
ac
c
ur
at
e
pl
ant
…
(
Shw
e
ta
B
ondr
e
)
3733
B
I
O
G
R
A
P
H
I
E
S
O
F
A
U
T
H
O
R
S
Dr.
Shweta
V.
Bondre
holds
a
Ph
.
D
.
in
Computer
Science
and
En
gineering
from
GHRU
Amravati,
India,
and
is
currently
an
Assistant
Professor
at
Ramdeobaba
University
(formerly
Shri
Ramdeobaba
College
of
Engineering
and
Management
).
With
over
14
years
of
experience.
She
has
developed
expertise
in
data
science,
machine
le
arning
,
DBMS,
big
data
manageme
nt,
and
data
mining
.
Her
extensive
contributions
to
aca
demia
and
industry
are
reflected
in
her
authorship
of
20
+
research
papers
published
in
esteemed
journals
an
d
conferences.
She ca
n be c
ontact
ed at
e
mail: shwetab
ondre1510@gmail.com.
Dr.
Uma
Yadav
holds
a
Ph
.
D
.
in
Computer
Science
and
Engineering
from
GHRU
Amravati,
India
.
S
he
currently
serves
as
an
Assistant
Pr
ofessor
at
Ramdeobaba
University
(formerly
Shri
Ramdeobaba
College
of
Engineering
and
Management),
where
she
specializes
in
data
science,
machine
learning,
cloud
computing,
big
data
management,
and
data
mining.
With
over
14
yea
rs
of
p
rofes
siona
l
exp
erien
ce,
sh
e
has
mad
e
sig
nific
ant
contrib
ution
s
to
bot
h
ac
adem
ia
a
nd
i
ndust
ry, autho
ring
and
co
-
autho
rin
g
20+
resea
rch paper
s
in
refere
ed jo
urnal
s and
conf
erenc
es. S
he c
an be
cont
acted
at e
mail:
uma
.y
adav12@
g
mail
.com.
Dr.
Vipin
D.
Bondre
received
his
B.
Tech
.
from
North
Mahar
ashtra
University
Jalgaon
and
M.
Tech
degrees
from
Rashtrasant
Tukadoji
Mahara
raj
Nagpur
University,
Nagpur in 2005 and 2
011 respectively.
He has been
awarded Ph
.
D
.
in
wireless communication
in
2020
from
RTMNU,
Nagpur
University.
He
is
an
Assistant
Pr
ofessor
at
Yeshwantrao
Chavan
College
of
Engineering,
Nagpur,
an
autonomous
institute.
He
researches
interests
include
wireless
communication
and
embedded
systems
.
He
has
au
thored/co
-
authored
25+
research
papers
in
refereed
journals
and
conferences.
H
e
can
be
contacted
at
e
mail:
vipin.bondre@
gmail.com.
Dr.
Poorva
Agrawal
obtained
her
Ph
.
D
.
degree
in
Compute
r
Science
and
Engineering
from
Symbiosis
International
(Deemed
University),
Pune
,
India
in
2020.
She
has
been
engaged
in
research
and
teaching
for
more
than
12
years.
At
pr
esent
she
is
working
a
s
Senior
Assistant
Professor
in
CSE
Dep
artment
at
Symbiosis
Institute
of
Technology
Nagpur,
Symbiosis
Inter
nationa
l
(Dee
med
Univer
sity)
Pune,
MH,
India.
She
has
prese
nted
more
than
30
papers
in
internationa
l/national
journals/con
ferenc
es
.
Her
res
earch
interests
include
artificial
intell
igence,
machine
learning,
data
science,
and
c
ompu
ter
vision
.
She
can
be
contacted
at e
mail: poorva.
agrawa
l@
sitnagpur.siu.e
du.in.
Evaluation Warning : The document was created with Spire.PDF for Python.