I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
15
,
No
.
2
,
A
p
r
il
20
25
,
p
p
.
1
8
5
0
~
1
8
6
0
I
SS
N:
2
0
8
8
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
2
.
p
p
1
8
5
0
-
1
8
6
0
1850
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Autom
a
ted
tom
a
t
o
leaf disea
se re
co
g
nition us
ing
dee
p
co
nv
o
lutiona
l net
wo
rks
Am
ir
So
hel,
M
d M
iza
n
ur
R
a
hm
a
n,
M
d Um
a
id H
a
s
a
n,
M
d
K
a
f
iul
I
s
la
m
,
L
a
m
ia
Ru
k
hs
a
ra
,
T
a
pa
s
y
Ra
bey
a
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
S
c
i
e
n
c
e
a
n
d
En
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
a
n
d
I
n
f
o
r
mat
i
o
n
T
e
c
h
n
o
l
o
g
y
,
D
a
f
f
o
d
i
l
I
n
t
e
r
n
a
t
i
o
n
a
l
U
n
i
v
e
r
si
t
y
,
D
h
a
k
a
,
B
a
n
g
l
a
d
e
s
h
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
r
2
6
,
2
0
2
4
R
ev
is
ed
Oct
1
0
,
2
0
2
4
Acc
ep
ted
Oct
2
3
,
2
0
2
4
Ag
ricu
lt
u
re
is
e
ss
e
n
ti
a
l
f
o
r
t
h
e
e
n
ti
re
g
lo
b
a
l
p
o
p
u
lati
o
n
.
An
a
d
v
a
n
c
e
d
,
ro
b
u
st,
a
n
d
e
m
p
iri
c
a
ll
y
so
u
n
d
a
g
ricu
lt
u
re
se
c
to
r
is
e
ss
e
n
ti
a
l
fo
r
n
o
u
rish
in
g
th
e
g
lo
b
a
l
p
o
p
u
latio
n
.
Va
rio
u
s
lea
f
d
ise
a
se
s
c
a
u
se
fin
a
n
c
ial
h
a
r
d
sh
ip
s
f
o
r
fa
rm
e
rs
a
n
d
re
late
d
b
u
sin
e
ss
e
s.
Early
i
d
e
n
ti
f
ica
ti
o
n
o
f
f
o
li
a
r
d
ise
a
se
s
in
c
ro
p
s
wo
u
l
d
g
re
a
tl
y
h
e
lp
fa
rm
e
rs,
lea
d
in
g
t
o
a
su
b
sta
n
ti
a
l
in
c
re
a
se
in
a
g
ricu
lt
u
ra
l
p
ro
d
u
c
ti
v
it
y
.
T
h
e
to
m
a
to
is
a
wid
e
ly
re
c
o
g
n
ize
d
a
n
d
n
o
u
rish
i
n
g
fo
o
d
th
a
t
is
e
a
sily
a
c
c
e
ss
ib
le
a
n
d
h
ig
h
ly
fa
v
o
re
d
b
y
fa
rm
e
rs.
Earl
y
d
iag
n
o
sis
o
f
to
m
a
t
o
lea
f
d
ise
a
se
s
is
c
ru
c
ial
to
m
a
x
imiz
e
to
m
a
t
o
c
ro
p
p
r
o
d
u
c
ti
o
n
.
T
h
is
stu
d
y
a
ims
to
u
t
il
ize
a
d
e
e
p
lea
rn
in
g
a
p
p
r
o
a
c
h
to
a
c
c
u
ra
tely
d
e
tec
t
a
n
d
c
las
sify
d
a
m
a
g
e
d
lea
v
e
s
a
n
d
d
i
se
a
se
p
a
tt
e
rn
s
in
to
m
a
to
lea
f
i
m
a
g
e
s.
By
e
m
p
lo
y
i
n
g
a
su
b
sta
n
ti
a
l
q
u
a
n
ti
ty
o
f
d
e
e
p
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
two
r
k
m
o
d
e
ls,
we
a
c
h
iev
e
d
a
h
i
g
h
lev
e
l
o
f
p
re
c
isi
o
n
in
d
ia
g
n
o
sin
g
th
e
c
o
n
d
it
i
o
n
.
T
h
e
d
a
tas
e
t
u
se
d
in
o
u
r
st
u
d
y
wo
r
k
is
a
se
lf
-
c
o
n
tain
e
d
d
a
tas
e
t
o
b
tai
n
e
d
b
y
d
irec
t
o
b
se
rv
a
ti
o
n
o
f
t
o
m
a
to
fiel
d
s
in
ru
ra
l
a
re
a
s
o
f
Ba
n
g
lad
e
sh
.
It
c
o
n
sis
ts
o
f
fo
u
r
c
las
se
s:
h
e
a
lt
h
y
,
b
lac
k
m
o
ld
,
g
re
y
m
o
l
d
,
a
n
d
p
o
wd
e
ry
m
il
d
e
w.
In
th
is
stu
d
y
wo
rk
,
we
u
t
il
ize
d
v
a
ri
o
u
s
ima
g
e
p
re
-
p
r
o
c
e
ss
in
g
tec
h
n
i
q
u
e
s
a
n
d
a
p
p
li
e
d
VG
G
1
6
,
In
c
e
p
ti
o
n
V
3
,
De
n
se
Ne
t1
2
1
,
a
n
d
Ale
x
Ne
t
m
o
d
e
ls.
O
u
r
re
su
lt
s
sh
o
we
d
th
a
t
t
h
e
De
n
se
Ne
t1
2
1
m
o
d
e
l
a
tt
a
in
e
d
t
h
e
h
ig
h
e
r
a
c
c
u
ra
c
y
o
f
9
7
%
.
Th
is
d
isc
o
v
e
r
y
g
u
a
ra
n
tee
s
a
c
c
u
r
a
te
d
e
tec
ti
o
n
o
f
t
o
m
a
to
d
ise
a
se
s
in
a
ra
p
id
m
a
n
n
e
r,
u
sh
e
ri
n
g
in
a
n
e
w ag
ricu
l
tu
ra
l
re
v
o
lu
t
io
n
.
K
ey
w
o
r
d
s
:
Au
to
m
ated
r
ec
o
g
n
itio
n
Dee
p
lear
n
in
g
Den
s
eNe
t1
2
1
Dis
ea
s
e
id
en
tific
atio
n
I
m
ag
e
p
r
o
ce
s
s
in
g
I
n
ce
p
tio
n
V3
T
o
m
ato
leaf
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
:
Md
Miz
an
u
r
R
ah
m
an
D
e
p
a
r
t
m
e
n
t
o
f
C
o
m
p
u
t
e
r
Sc
i
en
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
ce
a
n
d
I
n
f
o
r
m
a
ti
o
n
T
ec
h
n
o
l
o
g
y
,
D
a
f
f
o
d
i
l
I
n
t
e
r
n
a
t
i
o
n
a
l
U
n
i
v
e
r
s
it
y
Dh
ak
a,
B
an
g
lad
esh
E
m
ail:
m
izan
u
r
r
a
h
m
an
.
cse@
d
iu
.
ed
u
.
b
d
1.
I
NT
RO
D
UCT
I
O
N
As
a
m
ajo
r
co
n
t
r
ib
u
to
r
to
b
o
th
f
o
o
d
s
ec
u
r
ity
an
d
ec
o
n
o
m
ic
d
ev
elo
p
m
en
t,
to
m
ato
p
r
o
d
u
ctio
n
is
ess
en
tial
to
th
e
s
u
s
tain
ab
ilit
y
o
f
ag
r
icu
ltu
r
al
ec
o
n
o
m
ies
wo
r
ld
wid
e.
I
n
B
an
g
lad
esh
,
wh
er
e
ag
r
icu
ltu
r
e
is
th
e
m
ain
ec
o
n
o
m
ic
p
illar
,
t
o
m
ato
es
ar
e
an
im
p
o
r
tan
t
c
ash
cr
o
p
.
Fu
r
t
h
er
m
o
r
e,
I
n
d
i
a
p
r
o
d
u
ce
s
ar
o
u
n
d
5
,
3
0
0
,
0
0
0
to
n
s
o
f
g
o
o
d
s
an
n
u
ally
o
n
an
ar
ea
o
f
a
b
o
u
t
3
,
5
0
,
0
0
0
h
ec
tar
es
[
1
]
.
I
n
m
an
y
p
ar
ts
o
f
th
e
wo
r
l
d
,
to
m
ato
es
co
n
s
titu
te
a
m
ajo
r
cr
o
p
;
an
av
er
a
g
e
p
er
s
o
n
co
n
s
u
m
es
2
0
k
g
o
f
t
o
m
ato
es
ea
ch
y
ea
r
.
R
o
u
g
h
ly
1
5
%
o
f
all
v
eg
etab
les
ar
e
c
o
n
s
u
m
ed
in
th
is
way
[
2
]
.
T
o
d
ay
,
ag
r
icu
l
tu
r
al
lan
d
s
ca
p
es
r
e
q
u
ir
e
c
o
n
ti
n
u
al
cr
o
p
an
d
p
la
n
t
s
u
r
v
eillan
ce
to
p
r
ev
e
n
t
p
lan
t
d
is
ea
s
es
[
3
]
.
Ho
wev
er
,
wid
esp
r
ea
d
d
is
ea
s
es
th
r
ea
ten
to
m
at
o
cr
o
p
s
,
r
ed
u
cin
g
p
r
o
d
u
ctiv
ity
.
B
an
g
lad
esh
i
to
m
ato
f
ar
m
er
s
s
tr
u
g
g
le
with
b
lack
m
o
ld
,
g
r
ay
m
o
ld
,
an
d
p
o
w
d
er
y
m
ild
ew.
T
h
ese
d
is
ea
s
es
lo
wer
to
m
ato
y
ield
a
n
d
q
u
ality
an
d
in
cr
ea
s
e
p
r
o
d
u
ctio
n
co
s
ts
o
win
g
to
f
u
n
g
icid
e
u
s
e.
Un
d
er
s
tan
d
in
g
an
d
tr
ea
tin
g
to
m
ato
leaf
d
is
ea
s
es is
e
s
s
en
tial f
o
r
a
s
tr
o
n
g
ag
r
icu
ltu
r
al
in
d
u
s
tr
y
in
th
e
n
atio
n
.
Dee
p
lear
n
in
g
h
as
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
u
to
ma
ted
to
ma
to
le
a
f d
is
ea
s
e
r
ec
o
g
n
itio
n
u
s
in
g
d
ee
p
co
n
v
o
lu
tio
n
a
l
n
etw
o
r
ks
(
A
mir
S
o
h
el
)
1851
s
ig
n
if
ican
tly
tr
an
s
f
o
r
m
ed
th
e
co
m
p
u
ter
v
is
io
n
in
d
u
s
tr
y
b
y
o
f
f
er
in
g
ad
v
an
ce
d
ca
p
a
b
ilit
ies
f
o
r
a
u
to
m
atic
an
aly
s
is
o
f
im
ag
es a
n
d
r
ec
o
g
n
itio
n
[
4
]
.
T
h
e
ex
is
ten
ce
o
f
s
ev
er
al
m
in
u
s
cu
le
o
b
jects in
an
im
ag
e
is
a
s
ig
n
if
ican
t
ch
allen
g
e
f
o
r
p
r
ec
is
e
id
e
n
tific
atio
n
o
f
item
s
in
th
e
f
ield
s
o
f
c
o
m
p
u
ter
v
is
io
n
an
d
o
b
ject
d
etec
tio
n
r
esear
ch
[
5
]
.
I
n
th
e
ag
r
ic
u
ltu
r
al
s
ec
to
r
,
th
e
ap
p
licatio
n
o
f
m
o
d
els
b
ased
o
n
d
ee
p
lear
n
i
n
g
f
o
r
th
e
d
ia
g
n
o
s
is
an
d
d
etec
tio
n
o
f
p
lan
t
d
is
ea
s
es
h
as
g
ain
ed
p
o
p
u
lar
ity
r
ec
e
n
tly
.
T
h
is
r
esear
ch
s
tr
iv
es
to
g
iv
e
a
co
m
p
r
e
h
en
s
iv
e
s
u
m
m
ar
y
o
f
m
u
ltip
le
im
p
o
r
ta
n
t
p
u
b
licatio
n
s
th
at
h
av
e
ad
v
an
ce
d
th
is
q
u
ick
l
y
ev
o
l
v
in
g
t
o
p
ic,
clar
if
y
i
n
g
th
e
s
tr
ateg
ies
u
s
ed
,
th
e
d
atasets
u
s
ed
,
an
d
th
e
r
elate
d
ac
c
u
r
ac
y
lev
els
attain
ed
b
y
d
if
f
e
r
en
t
d
ee
p
lear
n
i
n
g
ar
ch
itectu
r
es.
Ag
ar
wal
g
r
o
u
p
:
u
s
in
g
th
e
P
lan
t
V
illag
e
d
ataset
,
Ag
ar
wal
an
d
ass
o
c
iates
ass
e
s
s
ed
th
e
p
er
f
o
r
m
an
ce
o
f
d
ee
p
lear
n
in
g
m
o
d
els
s
u
ch
as
VGG1
6
,
I
n
ce
p
tio
n
V3
,
a
n
d
Mo
b
ileNet.
T
h
e
h
ig
h
est
-
lev
el
ac
cu
r
ac
y
o
f
7
7
.
2
%
was
ac
h
iev
ed
b
y
VGG1
6
[
6
]
.
C
h
en
e
t
a
l.
[
7
]
u
tili
zin
g
th
e
Hu
n
a
n
Veg
etab
le
I
n
s
titu
te
d
ataset,
th
ey
co
n
d
u
cted
ex
p
er
im
en
ts
u
tili
zin
g
m
o
d
els
lik
e
Alex
Net,
R
esNet5
0
,
AR
Net,
an
d
B
-
AR
Net.
B
-
AR
Net
s
u
r
p
ass
ed
th
e
o
th
er
s
with
an
ac
cu
r
ac
y
o
f
8
8
.
4
3
%.
J
ian
g
et
a
l.
[
8
]
u
s
ed
r
ec
tifie
d
l
in
ea
r
u
n
its
(
R
eL
U)
ac
tiv
atio
n
s
to
an
aly
ze
m
a
n
y
R
esNet
to
p
o
lo
g
ies,
with
a
f
o
cu
s
o
n
th
e
AI
C
h
allen
g
e
r
d
at
aset.
A
n
u
m
b
er
o
f
s
ettin
g
s
in
th
eir
an
aly
s
is
s
h
o
wed
a
r
em
ar
k
a
b
le
ac
cu
r
ac
y
o
f
u
p
to
9
8
.
3
%.
Sti
ll,
th
e
d
ataset
was
n
o
t
b
alan
ce
d
.
Z
h
o
u
et
a
l.
[
9
]
u
s
ed
th
e
AI
C
h
allen
g
er
d
ataset
to
an
aly
ze
d
ee
p
co
n
v
o
l
u
tio
n
al
n
eu
r
a
l
n
etwo
r
k
(
C
NN)
,
R
esNet5
0
,
Den
s
eNe
t,
an
d
r
estru
ctu
r
ed
r
esid
u
al
d
en
s
e
n
etwo
r
k
(
R
R
DN)
m
o
d
els.
T
h
e
R
R
DN
m
o
d
el
p
r
o
v
ed
to
b
e
th
e
m
o
s
t a
cc
u
r
ate,
with
a
9
5
% a
cc
u
r
ac
y
r
ate.
B
alak
r
is
h
n
a
an
d
R
ao
[
1
0
]
u
s
in
g
p
r
o
b
ab
ilis
tic
n
eu
r
al
n
etwo
r
k
(
PNN)
an
d
k
-
n
ea
r
est
n
eig
h
b
o
r
(
KNN)
m
o
d
els,
th
ey
ac
h
iev
ed
a
n
o
te
wo
r
th
y
ac
cu
r
ac
y
o
f
9
1
.
8
8
%
wh
en
u
s
in
g
PNN
o
n
f
ar
m
lan
d
p
h
o
to
s
.
Go
n
za
lez
-
Hu
itro
n
et
a
l.
[
1
1
]
u
s
in
g
th
e
Plan
tVillag
e
d
ataset,
th
e
r
ese
ar
ch
er
s
test
ed
m
an
y
m
o
d
els,
i
n
clu
d
in
g
Xce
p
tio
n
,
Mo
b
ileNetV2
,
an
d
NasNetM
o
b
ile.
Su
r
p
r
is
in
g
ly
,
Xce
p
tio
n
o
b
tain
ed
a
f
lawless
ac
cu
r
ac
y
s
co
r
e
o
f
1
.
0
0
.
Ab
b
as
et
a
l.
[
1
2
]
test
ed
Den
s
eNe
t
wi
th
C
-
GAN
o
n
Plan
tVil
lag
e
an
d
s
y
n
th
etic
p
h
o
to
s
,
an
d
Den
s
e
Net
p
er
f
o
r
m
ed
b
est
with
an
ac
cu
r
ac
y
o
f
9
7
.
1
1
%.
Z
h
an
g
et
a
l.
[
1
3
]
R
esNet
u
n
d
er
s
to
ch
asti
c
g
r
ad
ien
t
d
escen
t
(
SGD)
s
h
o
wed
th
e
h
ig
h
est
ac
cu
r
ac
y
o
f
9
6
.
5
1
%
am
o
n
g
th
e
p
h
o
to
s
in
Z
h
an
g
an
d
co
lleag
u
es'
o
p
en
-
ac
ce
s
s
im
ag
e
co
llectio
n
d
ev
o
ted
to
p
la
n
t
h
ea
lth
.
Ho
n
g
et
a
l.
[
1
4
]
u
s
in
g
th
e
Plan
t
Villag
e
d
ataset,
th
ey
test
ed
with
m
o
d
els
s
u
ch
as
Den
s
eNe
t_
Xce
p
tio
n
an
d
Xce
p
tio
n
;
Den
s
eNe
t_
Xce
p
tio
n
ac
h
iev
ed
an
ac
c
u
r
ac
y
o
f
9
7
.
1
0
%.
Ku
m
ar
an
d
Van
i
[
1
5
]
u
s
in
g
a
s
u
b
s
et
o
f
th
e
Plan
tVillag
e
d
ataset,
th
ey
ex
am
in
ed
s
ev
er
al
m
o
d
els;
VGG1
6
s
to
o
d
o
u
t
with
an
ac
cu
r
ac
y
o
f
9
9
.
2
5
%.
Pro
ttas
h
a
an
d
R
ez
a
[
1
6
]
eig
h
t
v
ar
i
o
u
s
s
tate
-
of
-
th
e
-
ar
t
c
o
n
v
o
lu
ti
o
n
n
eu
r
al
n
etwo
r
k
m
o
d
els
h
av
e
h
ad
th
eir
p
e
r
f
o
r
m
an
ce
ev
alu
ate
d
with
an
em
p
h
asis
o
n
r
ice
p
lan
t
d
is
ea
s
e
d
iag
n
o
s
is
.
T
h
e
s
u
g
g
ested
ap
p
r
o
ac
h
ac
cu
r
atel
y
d
iag
n
o
s
es
d
is
ea
s
es
in
r
ice
p
lan
ts
an
d
h
as
test
in
g
an
d
v
alid
atio
n
ac
cu
r
ac
y
o
f
9
6
.
5
% a
n
d
9
5
.
3
%,
r
esp
ec
tiv
el
y
.
T
ab
le
1
p
r
o
v
id
es th
e
s
u
m
m
ar
y
o
f
p
r
ev
i
o
u
s
wo
r
k
wh
ich
w
as d
escr
ib
ed
b
ef
o
r
e.
T
ab
le
1
.
C
o
m
p
a
r
ativ
e
an
aly
s
is
with
ex
is
tin
g
wo
r
k
s
S
e
r
i
a
l
n
o
S
t
u
d
i
e
s
D
a
t
a
s
e
t
M
o
d
e
l
s
B
e
st
a
c
c
u
r
a
c
y
1
A
g
a
r
w
a
l
e
t
a
l
.
[
6
]
P
l
a
n
t
V
i
l
l
a
g
e
d
a
t
a
se
t
V
G
G
1
6
,
I
n
c
e
p
t
i
o
n
V
3
,
M
o
b
i
l
e
N
e
t
V
G
G
1
6
--
7
7
.
2
%
2
C
h
e
n
e
t
a
l
.
[
7
]
H
u
n
a
n
V
e
g
e
t
a
b
l
e
I
n
st
i
t
u
t
e
A
l
e
x
N
e
t
,
R
e
sN
e
t
5
0
,
A
R
N
e
t
,
B
-
A
R
N
e
t
B
-
A
R
N
e
t
-
8
8
.
4
3
%
3
Ji
a
n
g
e
t
a
l
.
[
8
]
A
I
C
h
a
l
l
e
n
g
e
r
Res
N
e
t
,
R
e
LU
,
7
×
7
,
L
-
R
e
LU
,
7
×
7
,
L
-
R
e
LU
,
1
1
×
1
1
L
-
R
e
LU
,
1
1
×
1
1
--
9
8
.
3
%,
4
Zh
o
u
e
t
a
l
.
[
9
]
A
I
C
h
a
l
l
e
n
g
e
r
D
e
e
p
C
N
N
,
R
e
s
N
e
t
5
0
,
D
e
n
s
e
N
e
t
,
R
R
D
N
R
R
D
N
--
9
5
%
5
B
a
l
a
k
r
i
sh
n
a
a
n
d
R
a
o
[
1
0
]
I
mag
e
s c
o
l
l
e
c
t
e
d
f
r
o
m
a
f
a
r
ml
a
n
d
P
N
N
,
K
N
N
P
N
N
--
9
1
.
8
8
%
6
G
o
n
z
a
l
e
z
-
H
u
i
t
r
o
n
e
t
a
l
.
[
1
1
]
P
l
a
n
t
V
i
l
l
a
g
e
d
a
t
a
se
t
M
o
b
i
l
e
N
e
t
V
2
,
N
a
sN
e
t
M
o
b
i
l
e
,
X
c
e
p
t
i
o
n
,
M
o
b
i
l
e
N
e
t
V
3
,
A
l
e
x
N
e
t
,
G
o
o
g
L
e
N
e
t
,
R
e
sN
e
t
1
8
X
c
e
p
t
i
o
n
-
1
.
0
0
7
A
b
b
a
s
e
t
a
l
.
[
1
2
]
P
l
a
n
t
V
i
l
l
a
g
e
,
S
y
n
t
h
e
t
i
c
i
ma
g
e
s
C
N
N
n
e
t
w
o
r
k
,
A
l
e
x
N
e
t
,
D
e
n
seNe
t
,
M
o
b
i
l
e
N
e
t
D
e
n
seN
e
t
,
C
-
G
A
N
9
7
.
1
1
%
8
Zh
a
n
g
e
t
a
l
.
[
1
3
]
O
p
e
n
a
c
c
e
ss
d
a
t
a
r
e
p
o
.
o
f
i
ma
g
e
s
t
h
a
t
f
o
c
u
s
o
n
p
l
a
n
t
h
e
a
l
t
h
A
l
e
x
N
e
t
(
S
G
D
)
,
A
l
e
x
N
e
t
(
A
d
a
m)
,
G
o
o
g
L
e
N
e
t
(
S
G
D
)
,
R
e
sN
e
t
(
S
G
D
)
,
R
e
sN
e
t
(
A
d
a
m)
R
e
sN
e
t
(SGD)
9
6
.
5
1
%
9
H
o
n
g
e
t
a
l
.
[
1
4
]
P
l
a
n
t
V
i
l
l
a
g
e
d
a
t
a
se
t
D
e
n
se
_
N
e
t
_
X
c
e
p
t
i
o
n
,
X
c
e
p
t
i
o
n
,
R
e
s
n
e
_
5
0
,
M
o
b
i
l
e
N
e
t
,
S
h
u
f
f
l
e
N
e
t
D
e
n
seN
e
t
_
X
c
e
p
t
i
o
n
-
9
7
.
1
0
%
10
K
u
mar
a
n
d
V
a
n
i
[
1
5
]
A
p
o
r
t
i
o
n
o
f
t
h
e
P
l
a
n
t
V
i
l
l
a
g
e
d
a
t
a
c
o
l
l
e
c
t
i
o
n
Le
N
e
t
,
V
G
G
1
6
,
R
e
sN
e
t
5
0
,
X
c
e
p
t
i
o
n
V
G
G
1
6
-
9
9
.
1
1
%
Desp
ite
th
e
p
r
o
g
r
ess
s
h
o
wn
in
th
ese
s
tu
d
ies,
th
er
e
is
s
ti
ll
a
l
ac
k
o
f
g
en
er
aliza
tio
n
o
f
th
ese
m
o
d
els
to
d
iv
er
s
e
d
atasets
an
d
en
v
ir
o
n
m
en
tal
co
n
d
itio
n
s
.
Mo
s
t
r
ese
ar
ch
p
r
im
ar
ily
f
o
cu
s
es
o
n
s
p
e
cif
ic
d
atasets
,
wh
ich
ca
n
lim
it
th
e
a
p
p
licab
ilit
y
o
f
th
e
m
o
d
els
to
r
ea
l
-
wo
r
ld
ag
r
icu
ltu
r
al
s
ce
n
ar
io
s
.
I
n
ad
d
iti
o
n
,
th
e
r
e
h
as
b
ee
n
lim
ited
r
esear
ch
co
n
d
u
cted
o
n
th
e
r
esil
ien
ce
o
f
th
ese
m
o
d
els
to
v
ar
iatio
n
s
in
en
v
ir
o
n
m
en
t
al
f
ac
to
r
s
lik
e
s
o
il
co
m
p
o
s
itio
n
a
n
d
lig
h
tin
g
co
n
d
itio
n
s
.
R
eso
lv
in
g
th
ese
is
s
u
es
will
im
p
r
o
v
e
d
ee
p
lear
n
in
g
m
o
d
els'
ab
ilit
y
to
id
en
tify
p
lan
t
d
is
ea
s
es in
ag
r
icu
ltu
r
e,
in
cr
ea
s
in
g
th
eir
d
e
p
en
d
ab
ilit
y
an
d
s
co
p
e
o
f
u
s
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
8
5
0
-
1
8
6
0
1852
I
n
r
ec
en
t
tim
e,
t
h
e
ap
p
licati
o
n
o
f
d
ee
p
lear
n
in
g
tec
h
n
iq
u
es
in
ag
r
icu
ltu
r
e
h
as
g
ain
e
d
tr
ac
tio
n
,
o
f
f
er
in
g
p
r
o
m
is
in
g
s
o
lu
tio
n
s
f
o
r
th
e
ea
r
ly
an
d
ac
cu
r
ate
d
e
tectio
n
an
d
m
an
a
g
em
en
t
o
f
c
r
o
p
d
is
ea
s
es.
Ou
r
r
esear
ch
en
d
ea
v
o
r
aim
s
to
en
h
an
ce
th
e
ex
is
tin
g
k
n
o
wled
g
e
in
th
e
f
ield
o
f
d
ee
p
lear
n
in
g
an
d
ag
r
icu
ltu
r
e
b
y
f
o
cu
s
in
g
o
n
d
etec
tin
g
t
o
m
ato
leaf
d
is
ea
s
es
in
B
an
g
lad
esh
.
T
h
e
f
o
llo
win
g
is
an
o
v
er
v
ie
w
o
f
th
e
p
r
im
a
r
y
co
n
tr
ib
u
tio
n
s
g
iv
en
to
th
is
r
es
ea
r
ch
:
i)
C
r
ea
ted
a
n
ew
d
ataset
co
n
s
is
tin
g
o
f
f
o
u
r
d
is
tin
ct
c
lass
es
(
b
lack
m
o
ld
,
g
r
ay
m
o
ld
,
p
o
wd
er
y
m
ild
ew,
h
ea
lth
y
to
m
ato
leav
es)
o
b
ta
in
ed
d
ir
ec
tly
f
r
o
m
ac
t
u
al
ag
r
icu
ltu
r
al
f
ield
s
o
f
B
an
g
lad
esh
;
ii)
Ap
p
lied
a
r
an
g
e
o
f
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
e
s
to
en
h
an
ce
an
d
o
p
tim
ize
th
e
d
ataset,
also
v
er
if
y
th
e
q
u
ality
o
f
im
a
g
e
af
ter
p
r
ep
r
o
ce
s
s
in
g
;
iii)
W
e
h
av
e
co
n
d
u
cted
a
c
o
m
p
a
r
ativ
e
i
n
v
esti
g
atio
n
o
f
th
e
p
er
f
o
r
m
an
ce
o
f
m
o
d
els
u
s
in
g
b
o
th
p
r
ep
r
o
ce
s
s
ed
an
d
r
aw
d
atasets
.
Pre
p
r
o
ce
s
s
in
g
ap
p
r
o
a
ch
es
p
lay
a
cr
u
cial
an
d
im
p
ac
tf
u
l
r
o
le
in
d
eter
m
i
n
in
g
th
e
p
er
f
o
r
m
a
n
ce
o
f
d
ee
p
lear
n
in
g
m
o
d
els
;
an
d
iv
)
T
h
i
s
s
y
s
tem
o
p
tim
ize
s
ag
r
icu
ltu
r
al
m
o
n
ito
r
in
g
an
d
m
an
ag
em
en
t
p
r
o
ce
d
u
r
es
u
s
in
g
ad
v
an
ce
d
d
ee
p
lear
n
i
n
g
m
o
d
els,
s
p
ec
if
ically
VGG1
6
,
I
n
ce
p
tio
n
V3
,
Den
s
e
Net1
2
1
,
an
d
Alex
Net.
Am
o
n
g
th
em
Den
s
eNe
t1
2
1
g
o
t
th
e
h
ig
h
est
ac
cu
r
ac
y
o
f
9
7
%.
T
h
r
o
u
g
h
th
is
r
esear
ch
,
we
asp
ir
e
to
n
o
t
o
n
ly
en
h
an
c
e
th
e
ef
f
icien
c
y
o
f
d
is
ea
s
e
d
e
tectio
n
b
u
t
also
to
em
p
o
wer
f
ar
m
e
r
s
with
a
v
alu
ab
le
to
o
l
f
o
r
tim
ely
in
ter
v
en
tio
n
,
u
ltima
tely
m
itig
atin
g
th
e
im
p
ac
t
o
f
to
m
ato
leaf
d
is
ea
s
es o
n
cr
o
p
y
ield
an
d
a
g
r
icu
ltu
r
al
s
u
s
tain
ab
ilit
y
in
B
an
g
lad
esh
.
2.
M
E
T
H
O
D
B
a
n
g
l
a
d
es
h
i
s
r
e
n
o
w
n
e
d
f
o
r
i
ts
a
g
r
i
c
u
l
t
u
r
al
s
e
ct
o
r
,
w
h
i
c
h
p
r
o
v
i
d
e
s
e
m
p
l
o
y
m
e
n
t
a
n
d
s
u
s
t
en
a
n
c
e
f
o
r
a
s
i
g
n
i
f
i
c
a
n
t
p
o
r
t
i
o
n
o
f
it
s
p
o
p
u
l
a
t
i
o
n
.
F
o
o
d
c
r
o
p
s
e
n
c
o
m
p
as
s
p
a
d
d
y
,
p
o
t
at
o
e
s
,
v
e
g
e
t
a
b
l
es
,
a
n
d
v
a
r
i
o
u
s
o
t
h
e
r
a
g
r
i
c
u
l
t
u
r
a
l
p
r
o
d
u
c
ts
.
R
e
g
a
r
d
in
g
v
e
g
g
i
e
s
,
t
o
m
at
o
e
s
m
i
g
h
t
b
e
m
e
n
t
i
o
n
e
d
as
o
n
e
o
f
t
h
e
m
.
A
c
c
o
r
d
i
n
g
t
o
d
at
a
p
r
o
v
i
d
e
d
b
y
t
h
e
B
a
n
g
l
a
d
e
s
h
B
u
r
e
a
u
o
f
S
t
a
t
is
ti
c
s
,
t
o
m
a
t
o
o
u
t
p
u
t
i
n
f
i
s
c
a
l
y
e
a
r
2
0
2
1
–
2
0
2
2
a
m
o
u
n
t
e
d
t
o
0
.
4
4
2
m
i
l
l
i
o
n
m
e
t
r
i
c
t
o
n
s
.
T
h
e
d
ec
l
in
e
i
n
t
o
m
a
t
o
p
r
o
d
u
c
t
i
o
n
c
a
n
b
e
a
t
t
r
i
b
u
t
e
d
t
o
d
is
e
as
e
,
wi
t
h
l
ea
f
d
i
s
e
as
e
b
e
i
n
g
t
h
e
m
o
s
t
s
e
r
i
o
u
s
a
m
o
n
g
t
h
e
m
.
S
o
m
e
o
f
t
h
e
l
ea
f
d
is
e
a
s
e
s
i
n
c
l
u
d
e
e
a
r
l
y
b
li
g
h
t
,
l
at
e
b
l
i
g
h
t
,
S
e
p
to
r
i
a
l
e
a
f
s
p
o
t
,
b
l
ac
k
m
o
l
d
,
l
e
a
f
m
o
l
d
,
t
o
m
a
t
o
y
e
l
l
o
w
l
e
a
f
c
u
r
l
v
i
r
u
s
(
T
Y
L
C
V
)
,
b
ac
t
e
r
i
a
l
s
p
o
t
,
b
a
c
t
e
r
i
a
l
m
o
s
a
i
c
v
ir
u
s
,
p
o
w
d
e
r
y
m
o
s
a
i
c
v
i
r
u
s
,
a
n
d
g
r
a
y
l
e
a
f
s
p
o
t
.
T
h
is
r
e
s
e
a
r
c
h
a
i
m
s
t
o
u
t
il
i
z
e
d
e
e
p
l
e
a
r
n
i
n
g
M
o
d
e
l
w
it
h
p
r
e
p
r
o
c
e
s
s
i
n
g
t
e
c
h
n
i
q
u
e
s
t
o
d
e
t
e
c
t
v
a
r
i
o
u
s
t
o
m
a
t
o
l
ea
f
d
i
s
ea
s
es
.
F
i
g
u
r
e
1
p
r
e
s
e
n
t
s
t
h
e
o
v
e
r
a
l
l
w
o
r
k
i
n
g
f
l
o
w
c
h
a
r
t
o
f
o
u
r
w
o
r
k
.
A
f
ew
s
u
b
s
ec
tio
n
s
ex
p
lain
t
h
e
m
eth
o
d
o
lo
g
ical
f
lo
wch
ar
t
s
ee
Fig
u
r
e
1
in
b
r
ief
.
T
h
e
au
to
m
ated
tech
n
iq
u
e
is
b
ein
g
im
p
lem
en
t
ed
b
y
m
ea
n
s
o
f
an
id
le
s
tep
.
W
e
f
ir
s
t
g
ath
er
r
aw
d
ata
f
r
o
m
ac
tu
al
f
ar
m
in
g
ar
ea
s
,
an
d
th
en
we
u
s
e
d
ata
p
r
ep
r
o
ce
s
s
in
g
an
d
n
o
r
m
aliza
tio
n
te
ch
n
iq
u
es.
T
h
r
o
u
g
h
d
ata
lab
el
in
g
,
we
ar
e
ab
le
to
ap
p
ly
n
u
m
er
o
u
s
h
i
g
h
-
lev
el
m
o
d
els
an
d
s
u
cc
ess
f
u
lly
id
en
tif
y
to
m
ato
leaf
d
is
ea
s
es.
An
d
t
h
e
g
o
al
o
f
th
e
en
tire
p
r
o
ce
d
u
r
e
is
to
d
e
v
elo
p
a
n
au
t
o
m
ated
s
y
s
tem
.
Fig
u
r
e
1
.
Me
th
o
d
o
lo
g
y
f
lo
w
d
iag
r
am
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
u
to
ma
ted
to
ma
to
le
a
f d
is
ea
s
e
r
ec
o
g
n
itio
n
u
s
in
g
d
ee
p
co
n
v
o
lu
tio
n
a
l
n
etw
o
r
ks
(
A
mir
S
o
h
el
)
1853
2
.
1
.
Da
t
a
s
et
d
escript
io
n
W
e
ar
e
wo
r
k
in
g
o
n
th
e
to
m
at
o
leaf
d
is
ea
s
e
d
ataset,
wh
ich
was
co
llected
f
r
o
m
a
g
r
icu
ltu
r
al
f
ield
s
o
f
B
an
g
lad
esh
b
y
o
u
r
r
esear
ch
t
ea
m
.
T
h
e
d
ataset
co
n
s
is
ts
o
f
3
7
0
p
iece
s
o
f
im
ag
es,
wh
ich
h
o
ld
th
e
co
m
m
o
n
d
is
ea
s
e
im
ag
es
o
f
b
lack
m
o
ld
,
g
r
ay
m
o
ld
,
p
o
wd
er
y
m
ild
ew
,
an
d
h
ea
lth
y
as
well.
Her
e
w
e
h
av
e
1
3
3
im
ag
es
f
o
r
th
e
h
ea
lth
y
class
.
B
lack
m
o
ld
,
g
r
a
y
m
o
l
d
,
an
d
p
o
wd
e
r
y
m
ild
ew
co
n
tain
6
0
,
5
6
,
an
d
1
3
1
,
r
esp
ec
tiv
ely
.
T
h
is
d
ataset
was
v
alid
ated
an
d
class
if
ied
b
y
th
e
Su
b
Ass
is
tan
t
Ag
r
icu
ltu
r
al
o
f
f
ice
r
,
Dep
ar
tm
en
t
o
f
ag
r
icu
ltu
r
al
ex
p
an
s
io
n
,
p
eo
p
le
r
ep
u
b
lic
o
f
B
an
g
lad
esh
.
Fig
u
r
e
2
u
s
u
ally
r
ep
r
esen
ts
th
e
s
am
p
le
im
ag
e
o
f
all
class
es.
Fig
u
r
e
2
.
Sam
p
le
d
ataset
o
f
ea
ch
class
2
.
2
.
Da
t
a
p
re
pro
ce
s
s
ing
T
o
en
h
an
ce
th
e
n
u
m
b
er
o
f
p
h
o
to
s
f
o
r
ea
ch
co
n
d
itio
n
,
we
p
r
o
ce
s
s
ed
th
e
d
ata
s
et
u
s
in
g
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es su
ch
r
escalin
g
,
s
h
if
tin
g
,
r
o
tatin
g
,
a
n
d
h
o
r
izo
n
tal
f
lip
p
in
g
.
a.
R
esizin
g
:
S
in
ce
we
wer
e
co
ll
ec
tin
g
d
ata
f
r
o
m
ag
r
icu
ltu
r
al
f
ield
s
,
th
e
s
ize
o
f
th
e
im
ag
es
i
s
n
o
t
eq
u
al.
W
e
r
esize
all
th
e
im
ag
es in
to
2
2
4
×
224
×
3
p
ix
els
.
b.
His
to
g
r
am
e
q
u
aliza
tio
n
:
An
i
m
ag
e'
s
co
n
tr
ast
an
d
b
r
ig
h
tn
e
s
s
ca
n
b
e
im
p
r
o
v
ed
u
s
in
g
th
e
d
ig
ital
im
ag
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
e
k
n
o
wn
as
h
is
to
g
r
am
eq
u
aliza
tio
n
.
I
t
wo
r
k
s
b
y
d
is
tr
ib
u
tin
g
p
ix
el
in
ten
s
ity
lev
els
th
r
o
u
g
h
o
u
t
th
e
en
tire
r
an
g
e
o
f
th
e
im
ag
e
[
1
7
]
.
As
a
r
esu
lt,
l
o
w
-
co
n
tr
ast
ar
ea
s
b
ec
o
m
e
ea
s
i
er
to
d
etec
t,
an
d
th
e
o
v
er
all
ap
p
ea
r
an
ce
o
f
th
e
im
ag
e
im
p
r
o
v
es
v
is
u
ally
[
1
7
]
.
I
t
ca
n
alter
th
e
ap
p
ea
r
an
ce
o
f
th
e
p
h
o
to
s
in
o
u
r
d
ataset
b
y
e
m
p
h
asizin
g
th
e
d
ar
k
an
d
lig
h
t
a
r
ea
s
,
wh
ich
is
ad
v
an
tag
e
o
u
s
f
o
r
en
h
an
cin
g
th
e
v
is
ib
ilit
y
o
f
s
p
ec
if
ic
d
etails
o
r
ch
ar
ac
ter
is
tics
in
lo
w
-
co
n
tr
ast
im
ag
es.
Fig
u
r
e
3
s
h
o
ws
th
e
af
ter
an
d
b
ef
o
r
e
ef
f
ec
t
o
f
h
is
to
g
r
am
eq
u
aliza
tio
n
.
c.
Gam
m
a
c
o
r
r
ec
tio
n
:
Gam
m
a
c
o
r
r
ec
tio
n
is
a
d
ig
ital
im
a
g
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
e
th
at
c
h
an
g
es
th
e
in
ten
s
ity
v
alu
es to
ch
an
g
e
an
im
ag
e'
s
b
r
ig
h
tn
ess
an
d
co
n
tr
ast.
I
n
o
r
d
er
to
ac
co
u
n
t
f
o
r
v
ar
iatio
n
s
in
th
e
way
m
o
n
ito
r
s
d
is
p
lay
lig
h
t,
it
p
er
f
o
r
m
s
a
n
o
n
lin
ea
r
o
p
e
r
atio
n
o
n
th
e
p
ix
el
v
alu
es.
I
t
ca
n
im
p
r
o
v
e
th
e
ae
s
th
etic
ap
p
ea
l
o
f
th
e
p
ictu
r
es
i
n
o
u
r
c
o
llectio
n
b
y
ad
j
u
s
tin
g
th
eir
b
r
i
g
h
tn
ess
an
d
co
n
tr
ast,
an
d
it
is
also
f
r
e
q
u
en
tly
u
s
ed
to
ad
ju
s
t
p
ictu
r
es
th
at
s
ee
m
e
x
ce
s
s
iv
ely
b
r
ig
h
t
o
r
d
a
r
k
[
1
8
]
.
Fi
g
u
r
e
4
s
h
o
ws
th
e
af
ter
a
n
d
b
ef
o
r
e
co
n
d
itio
n
o
f
g
am
m
a
co
r
r
ec
tio
n
.
d.
C
o
n
tr
ast
s
tr
etch
in
g
:
C
o
n
tr
ast
s
tr
etch
in
g
ex
p
a
n
d
s
a
n
im
ag
e'
s
in
ten
s
ity
r
an
g
e
to
im
p
r
o
v
e
d
etail
v
is
ib
ilit
y
.
L
in
ea
r
ly
ex
p
a
n
d
in
g
i
n
ten
s
ity
v
alu
es
to
s
p
an
th
e
en
tire
r
an
g
e
is
ty
p
ical
in
8
-
b
it
g
r
ay
s
ca
le
im
ag
es.
C
o
n
tr
ast
in
cr
ea
s
es
as
d
ar
k
ar
ea
s
d
a
r
k
e
n
an
d
b
r
ig
h
t
ar
ea
s
b
r
ig
h
ten
.
C
o
n
tr
ast
s
tr
etch
in
g
ca
n
im
p
r
o
v
e
Ou
r
d
ataset
p
h
o
to
g
r
ap
h
s
,
esp
ec
ially
th
o
s
e
with
lo
w
co
n
tr
ast
d
u
e
to
in
ad
eq
u
ate
lig
h
tin
g
[
1
8
]
.
Fig
u
r
e
5
s
h
o
ws
th
e
af
ter
an
d
b
ef
o
r
e
c
o
n
d
itio
n
o
f
c
o
n
tr
a
s
t stre
tch
in
g
.
e.
A
u
g
m
e
n
t
a
t
i
o
n
:
I
n
t
h
i
s
w
o
r
k
w
e
h
a
v
e
u
s
e
d
r
o
t
a
t
i
o
n
,
w
id
t
h
s
h
i
f
t
i
n
g
,
h
e
i
g
h
t
s
h
i
f
t
i
n
g
,
s
h
e
a
r
i
n
g
,
z
o
o
m
,
h
o
r
i
z
o
n
t
a
l
f
l
i
p
t
ec
h
n
i
q
u
e
s
f
o
r
im
a
g
e
a
u
g
m
e
n
t
a
t
i
o
n
.
T
a
b
l
e
2
s
h
o
w
s
t
h
e
n
u
m
b
e
r
o
f
i
m
a
g
e
s
a
f
t
er
a
u
g
m
e
n
t
a
t
i
o
n
.
f.
Sp
litt
in
g
:
T
o
tal
im
ag
es
ar
e
s
p
litt
ed
in
to
s
u
ch
a
d
is
tr
ib
u
tio
n
,
8
0
%
f
o
r
tr
ain
p
u
r
p
o
s
e
an
d
1
0
%
u
s
ed
f
o
r
test
in
g
an
d
o
th
e
r
1
0
% f
o
r
d
ata
v
alid
atio
n
.
Fig
u
r
e
3
.
Af
ter
h
is
to
g
r
am
e
q
u
aliza
tio
n
ef
f
ec
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
8
5
0
-
1
8
6
0
1854
Fig
u
r
e
4
.
Af
ter
Gam
m
a
co
r
r
ec
tio
n
ef
f
ec
t
Fig
u
r
e
5
.
Af
ter
co
n
tr
ast s
tr
etch
in
g
ef
f
ec
t
T
ab
le
2
.
Data
s
et
o
v
er
v
iew
ac
co
r
d
in
g
to
ea
ch
class
P
l
a
n
t
S
p
e
c
i
e
s
C
l
a
s
s Nam
e
To
t
a
l
I
mag
e
s
A
u
g
m
e
n
t
e
d
I
mag
e
s
To
ma
t
o
H
e
a
l
t
h
y
1
3
3
1
,
0
8
7
B
l
a
c
k
m
o
l
d
50
4
1
0
G
r
a
y
m
o
l
d
56
4
8
8
P
o
w
d
e
r
y
m
i
l
d
e
w
1
3
1
1
,
1
2
1
2.
3
.
I
ma
g
e
v
er
if
ica
t
io
n t
ec
h
niq
ue
s
I
m
ag
e
v
er
if
icatio
n
m
eth
o
d
s
v
er
if
y
a
n
im
a
g
e'
s
au
th
en
ticity
,
in
teg
r
ity
,
o
r
q
u
alities
.
I
n
m
a
n
y
f
ield
s
,
th
ese
m
eth
o
d
s
ar
e
ess
en
tial
f
o
r
en
s
u
r
in
g
th
at
p
h
o
to
g
r
ap
h
s
h
av
e
n
o
t
b
ee
n
m
a
n
ip
u
lated
o
r
m
is
r
ep
r
esen
ted
.
I
m
ag
e
p
r
o
ce
s
s
in
g
u
s
es o
b
jectiv
e
q
u
ality
ass
ess
m
en
t e
x
ten
s
iv
ely
.
Statis
tic
s
in
clu
d
in
g
s
tr
u
ctu
r
e
s
im
ilar
ity
in
d
ex
m
etr
ic
(
SS
I
M)
,
p
ea
k
s
ig
n
al
-
to
-
n
o
is
e
r
atio
(
PS
NR
)
,
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
MSE
)
,
r
o
o
t
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
R
MSE
)
,
an
d
p
ar
am
eter
s
o
f
f
r
eq
u
e
n
cy
lik
e
s
p
ec
tr
al
p
h
ase
an
d
m
ag
n
itu
d
e
d
is
to
r
tio
n
s
ar
e
in
clu
d
ed
[
1
9
]
.
T
ab
le
3
s
h
o
ws
all
1
0
r
a
n
d
o
m
l
y
p
ick
ed
p
h
o
to
g
r
ap
h
s
'
ex
ce
p
tio
n
ally
h
ap
p
y
s
co
r
es.
T
h
is
s
h
o
ws
h
o
w
v
er
if
icatio
n
s
co
r
es in
cr
ea
s
e
r
esear
ch
q
u
alit
y
.
T
ab
le
3
.
I
m
ag
e
v
e
r
if
icatio
n
s
c
o
r
es
I
mag
e
S
S
I
M
P
S
N
R
R
M
S
E
M
S
E
R
a
n
g
e
(
-
1
t
o
1
)
(
0
t
o
3
0
)
(
0
t
o
∞
)
(
0
t
o
∞
)
P
r
e
f
e
r
r
e
d
R
a
n
g
e
C
l
o
se
t
o
1
2
0
t
o
3
0
C
l
o
se
t
o
0
C
l
o
se
t
o
0
I
mag
e
1
.
0
.
7
4
2
8
.
0
9
4
1
.
6
1
1
7
3
1
.
8
1
I
mag
e
2
.
0
.
7
8
2
8
.
1
1
3
4
.
8
9
1
2
1
7
.
3
5
I
mag
e
3
.
0
.
7
7
2
7
.
9
5
3
6
.
1
0
1
3
0
3
.
1
3
I
mag
e
4
.
0
.
8
2
2
7
.
9
8
2
8
.
5
8
8
1
6
.
6
4
I
mag
e
5
.
0
.
8
0
2
7
.
8
1
3
1
.
0
4
9
6
3
.
2
0
I
mag
e
6
.
0
.
8
4
2
8
.
2
1
3
0
.
6
5
9
3
9
.
1
2
I
mag
e
7
.
0
.
8
3
2
7
.
9
1
3
3
.
2
1
1
1
0
2
.
9
4
I
mag
e
8
.
0
.
7
6
2
7
.
9
4
3
6
.
4
2
1
3
2
6
.
2
2
I
mag
e
9
.
0
.
7
5
2
8
.
2
3
3
5
.
8
6
1
2
8
6
.
2
1
I
mag
e
1
0
.
0
.
7
2
2
7
.
9
6
4
1
.
5
4
1
7
2
5
.
8
7
2
.
4
.
M
o
del
i
m
plem
ent
a
t
i
o
n
I
n
o
r
d
er
to
g
et
an
ex
ce
p
tio
n
al
o
u
tco
m
e,
we
attem
p
t to
ap
p
ly
m
u
ltip
le
ad
v
an
ce
d
d
ee
p
lear
n
i
n
g
m
o
d
els
in
th
is
s
tu
d
y
.
Ou
r
to
p
f
o
u
r
im
p
lem
en
ted
m
o
d
els
th
at
ar
e
th
e
s
u
b
ject
o
f
th
is
s
tu
d
y
ar
e
p
r
e
s
en
ted
in
th
is
p
o
s
t,
an
d
we
g
o
o
n
to
ex
p
lain
t
h
e
ev
alu
atio
n
o
f
ea
ch
m
o
d
el
u
s
in
g
a
v
ar
iety
o
f
m
atr
ices
an
d
v
is
u
aliza
tio
n
s
.
A
s
y
n
o
p
s
is
o
f
th
e
m
o
d
el
is
p
r
o
v
i
d
ed
b
elo
w.
a.
VGG1
6
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
o
f
Vis
u
al
Geo
m
etr
y
Gr
o
u
p
1
6
ar
e
u
s
ed
f
o
r
im
a
g
e
class
if
icatio
n
.
I
t
is
am
o
n
g
t
h
e
m
o
s
t
p
o
p
u
lar
tech
n
iq
u
es
f
o
r
ex
tr
ac
tin
g
v
is
u
al
f
ea
tu
r
es
[
1
8
]
.
T
h
e
d
esig
n
atio
n
"
1
6
"
allu
d
es
to
th
e
n
etwo
r
k
'
s
weig
h
t
tier
s
.
VGG1
6
,
a
d
ee
p
an
d
u
n
if
o
r
m
co
n
v
o
lu
tio
n
al
s
tr
u
ctu
r
e,
e
x
ce
ls
at
p
ictu
r
e
class
if
icatio
n
an
d
r
ec
o
g
n
itio
n
b
ec
au
s
e
t
o
its
s
im
p
licity
.
T
h
er
e
ar
e
th
r
ee
f
u
lly
lin
k
ed
lay
er
s
an
d
th
ir
teen
co
n
v
o
l
u
tio
n
al
lay
er
s
in
th
e
d
esig
n
.
Ma
x
-
p
o
o
lin
g
em
p
lo
y
s
2
x
2
f
ilter
s
,
an
d
co
n
v
o
lu
tio
n
al
lay
er
s
u
s
e
tin
y
3
×
3
f
ilter
s
with
a
s
tr
id
e
o
f
o
n
e
.
b.
Alex
Net:
Alex
Net
i
s
a
r
ev
o
lu
tio
n
ar
y
im
ag
e
class
if
icatio
n
C
NN.
I
t
h
as
f
iv
e
co
n
v
o
lu
tio
n
al
an
d
th
r
ee
f
u
lly
lin
k
ed
lay
er
s
.
Alex
Net,
k
n
o
w
n
f
o
r
u
s
in
g
R
eL
U
an
d
o
th
er
m
eth
o
d
s
,
p
io
n
ee
r
ed
d
ee
p
lear
n
in
g
f
o
r
c
o
m
p
u
ter
v
is
io
n
task
s
.
C
NN
-
ab
s
tr
ac
ted
f
ea
tu
r
es
o
f
f
er
s
tr
o
n
g
er
d
if
f
er
e
n
tiatio
n
an
d
m
o
r
e
s
em
an
tic
in
f
o
r
m
atio
n
th
a
n
ar
tific
ial
f
ea
tu
r
es,
ac
co
r
d
in
g
to
r
esear
ch
u
s
in
g
th
e
f
ir
s
t
f
u
ll
c
o
n
n
ec
tio
n
lay
er
o
f
Alex
Net
as
p
ictu
r
e
f
ea
tu
r
es
[
2
0
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
u
to
ma
ted
to
ma
to
le
a
f d
is
ea
s
e
r
ec
o
g
n
itio
n
u
s
in
g
d
ee
p
co
n
v
o
lu
tio
n
a
l
n
etw
o
r
ks
(
A
mir
S
o
h
el
)
1855
c.
I
n
c
e
p
t
i
o
n
V
3
:
G
o
o
g
l
e
c
r
e
a
t
e
d
I
n
c
e
p
t
i
o
n
V
3
f
o
r
i
m
a
g
e
c
l
a
s
s
i
f
i
ca
t
i
o
n
a
n
d
o
b
j
e
c
t
r
e
c
o
g
n
i
t
i
o
n
.
T
r
a
n
s
f
e
r
l
e
a
r
n
i
n
g
i
n
I
n
c
e
p
t
i
o
n
V
3
,
a
n
u
p
g
r
a
d
e
d
G
o
o
g
L
e
N
e
t
a
r
c
h
i
t
e
ct
u
r
e
,
i
m
p
r
o
v
e
s
b
i
o
m
e
d
i
c
a
l
c
a
t
e
g
o
r
i
z
at
i
o
n
[
2
0
]
.
I
n
c
e
p
t
i
o
n
p
r
o
p
o
s
e
s
a
m
o
d
e
l
wi
t
h
s
e
v
e
r
al
c
o
n
v
o
l
u
t
i
o
n
a
l
f
i
lt
e
r
s
o
f
d
i
f
f
er
e
n
t
s
i
z
es
[
2
1
]
.
I
n
c
e
p
t
i
o
n
V
3
ex
c
e
l
s
at
p
i
c
t
u
r
e
c
a
t
e
g
o
r
i
z
a
ti
o
n
a
n
d
o
b
j
e
c
t
d
e
t
e
ct
i
o
n
w
i
t
h
b
a
tc
h
n
o
r
m
a
l
i
z
a
ti
o
n
,
f
a
c
t
o
r
i
z
e
d
c
o
n
v
o
l
u
t
i
o
n
s
,
a
n
d
e
f
f
i
c
i
e
n
c
y
.
d.
Den
s
eNe
t
-
1
2
1
:
A
Den
s
eNe
ts
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
tr
u
ctu
r
e
is
ca
lled
Den
s
eNe
t
-
1
2
1
.
De
n
s
eNe
ts
'
co
n
v
o
l
u
tio
n
al,
p
o
o
lin
g
,
a
n
d
f
u
lly
co
n
n
ec
ted
la
y
er
s
g
et
d
ir
ec
t
in
p
u
t
f
r
o
m
ev
er
y
lay
e
r
th
at
ca
m
e
b
ef
o
r
e
th
em
,
p
r
o
d
u
ci
n
g
h
ig
h
ly
co
n
n
ec
ted
f
ea
tu
r
e
m
a
p
s
[
2
2
]
.
B
ased
o
n
its
1
2
1
lay
er
s
,
Den
s
eN
et
-
1
2
1
im
p
r
o
v
es
tr
ain
in
g
ef
f
icien
cy
an
d
f
ea
tu
r
e
r
eu
s
e.
3.
RE
SU
L
T
AND
DI
SCUS
I
I
O
N
Usi
n
g
o
u
r
co
r
e
d
ataset,
we
wer
e
ab
le
to
ac
h
iev
e
h
ig
h
l
y
s
atis
f
ac
to
r
y
r
esu
lts
in
o
u
r
r
esear
ch
.
I
t
is
q
u
ite
well
r
ef
lecte
d
in
s
ev
er
al
o
u
tco
m
e
d
is
cu
s
s
io
n
d
im
en
s
io
n
s
.
T
h
e
d
if
f
er
en
t m
atr
ix
an
d
a
n
aly
tical
r
esu
lts
p
r
esen
ted
h
er
e,
ea
ch
with
its
o
wn
s
ec
tio
n
,
p
r
o
v
id
e
th
e
p
r
o
o
f
o
f
o
u
r
ac
h
iev
ed
o
u
tc
o
m
e.
3
.
1
.
P
er
f
o
r
m
a
nce
m
e
a
s
urem
ent
m
et
rics
Ou
r
ev
alu
atio
n
co
n
s
id
er
ed
ess
en
tial
m
etr
ics
s
u
ch
as
p
r
ec
is
io
n
,
r
ec
all,
s
p
ec
if
icity
,
an
d
F1
-
s
co
r
e
[
2
3
]
.
T
h
ese
m
etr
ics
p
r
o
v
id
e
a
co
m
p
r
eh
en
s
iv
e
ass
ess
m
en
t
o
f
th
e
m
o
d
els'
p
er
f
o
r
m
an
ce
in
d
iag
n
o
s
in
g
to
m
ato
leaf
d
is
ea
s
es,
o
f
f
er
in
g
v
al
u
ab
le
in
s
ig
h
ts
f
o
r
p
r
ac
tical
ag
r
icu
ltu
r
al
ap
p
licatio
n
s
[
2
4
]
.
a.
Acc
u
r
ac
y
Acc
u
r
ac
y
=
+
+
+
+
(
1
)
b.
T
h
e
J
ac
ca
r
d
s
co
r
e
J
ac
ca
r
d
Sco
r
e
=
(
2
)
c.
Pre
cisi
o
n
Pre
cisi
o
n
=
+
(
3
)
d.
R
ec
all
R
ec
all
=
+
(
4
)
Her
e,
,
,
,
an
d
ar
e
in
f
u
ll
f
o
r
m
,
r
esp
ec
tiv
ely
:
tr
u
e
p
o
s
itiv
e,
tr
u
e
n
eg
ativ
e,
f
alse
p
o
s
itiv
e,
an
d
f
alse n
eg
ativ
e.
Fals
e
p
o
s
itiv
e,
f
alse n
eg
ativ
e
[
2
5
]
,
[
2
6
]
.
e.
T
h
e
F1
-
s
co
r
e
F1
-
s
co
r
e
=
2
∗
∗
+
(
5
)
T
h
e
m
o
d
els
we
u
s
ed
o
n
th
e
t
o
m
ato
leaf
d
is
ea
s
e
d
ataset,
co
n
s
id
er
in
g
all
th
e
m
ea
s
u
r
em
e
n
t
m
etr
ics
f
o
r
ea
ch
m
o
d
el,
ar
e
g
iv
en
n
ex
t sectio
n
s
.
3
.
2
.
Resul
t
dis
cus
s
io
n o
f
prepro
ce
s
s
ed
d
a
t
a
s
et
:
T
h
e
an
aly
s
is
o
f
to
m
ato
leaf
d
i
s
ea
s
e
d
etec
tio
n
alg
o
r
ith
m
s
r
ev
ea
ls
v
ar
y
in
g
p
er
f
o
r
m
a
n
ce
.
I
n
c
ep
tio
n
V3
lead
s
with
th
e
h
ig
h
est
ac
cu
r
ac
y
9
6
.
6
3
%,
p
r
ec
is
io
n
o
f
0
.
8
1
,
R
ec
all
0
.
7
8
,
F1
-
s
co
r
e
0
.
8
0
,
J
ac
ca
r
d
s
co
r
e
7
6
.
3
3
%,
an
d
AUC
0
.
6
8
.
Den
s
eNe
t
-
1
2
1
f
o
llo
ws
clo
s
ely
,
with
9
6
.
6
7
%
ac
cu
r
ac
y
an
d
a
n
o
tab
le
A
UC
s
co
r
e
o
f
7
1
.
7
3
%.
VGG1
6
p
er
f
o
r
m
s
well
with
8
6
.
6
7
%
ac
c
u
r
ac
y
b
u
t
ex
h
ib
its
a
lo
wer
AUC
s
co
r
e
o
f
6
1
.
2
0
.
Su
r
p
r
is
in
g
ly
,
Alex
Net
lag
s
with
8
3
.
3
0
%
a
cc
u
r
ac
y
a
n
d
th
e
l
o
west
p
r
ec
i
s
io
n
,
0
.
6
3
,
an
d
F1
-
s
co
r
e
o
f
0
.
5
7
.
T
h
ese
r
esu
lts
s
u
g
g
est
I
n
ce
p
tio
n
V3
an
d
De
n
s
eNe
t
-
1
2
1
as
r
o
b
u
s
t
ch
o
ices
,
wh
ile
VGG1
6
an
d
Alex
Net
m
ay
b
en
ef
it
f
r
o
m
f
u
r
th
er
o
p
tim
izatio
n
f
o
r
th
is
s
p
ec
if
ic
to
m
ato
leaf
d
is
ea
s
e
d
ataset.
I
n
T
a
b
le
4
we
ca
n
s
ee
th
e
v
is
u
aliza
tio
n
o
f
all
p
er
f
o
r
m
an
ce
m
ea
s
u
r
em
en
t
m
etr
ics.
T
ab
le
4
d
is
p
lay
s
th
e
r
esu
lts
o
f
m
u
ltip
le
m
ea
s
u
r
e
m
en
t
m
atr
ices
an
d
p
r
o
v
id
es
an
ex
ce
llen
t
v
is
u
al
r
e
p
r
esen
tatio
n
o
f
th
e
d
ataset'
s
s
t
ab
ilit
y
wh
en
u
s
in
g
v
ar
io
u
s
m
o
d
els
o
r
alg
o
r
ith
m
s
.
I
s
s
u
es
with
m
o
d
els'
f
it,
s
u
ch
as
b
ein
g
to
o
tig
h
t
o
r
to
o
lo
o
s
e,
ar
e
alm
o
s
t
n
o
n
ex
is
ten
t.
T
h
e
m
o
d
el
ac
cu
r
ac
y
in
th
is
p
r
ep
r
o
ce
s
s
ed
d
ata
is
q
u
ite
s
atis
f
y
in
g
,
an
d
th
e
o
t
h
er
m
ea
s
u
r
em
en
t
is
s
u
es
ar
e
clo
s
e
to
r
ea
lity
.
B
ased
o
n
th
e
co
m
p
ar
is
o
n
with
th
e
p
r
i
o
r
wo
r
k
,
as
s
h
o
wn
in
T
ab
le
1
,
n
o
r
eliab
le
r
ef
er
en
ce
ca
n
m
atch
t
h
e
ac
cu
r
ac
y
o
f
o
u
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
8
5
0
-
1
8
6
0
1856
d
ataset
an
d
m
o
d
el.
W
e
u
s
e
m
an
y
m
o
d
els
an
d
v
alid
ated
d
at
a
f
r
o
m
ex
p
er
t
f
ee
d
b
ac
k
to
p
r
o
v
id
e
an
AUC
v
alu
e,
r
ec
all,
F
1
-
s
co
r
e,
an
d
h
ea
lth
y
p
r
ec
is
io
n
.
T
ab
le
4
.
I
m
ag
e
m
ea
s
u
r
e
m
en
t
m
eth
o
d
s
M
o
d
e
l
s
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
Jac
c
a
r
d
AUC
V
G
G
1
6
8
6
.
6
7
%
0
.
7
5
0
.
7
3
0
.
8
0
0
.
6
4
0
.
6
1
I
n
c
e
p
t
i
o
n
V
3
9
6
.
6
3
%
0
.
8
1
0
.
7
8
0
.
8
0
0
.
7
6
0
.
6
8
D
e
n
seN
e
t
1
2
1
9
6
.
6
7
%
0
.
8
1
0
.
7
8
0
.
8
4
0
.
7
8
0
.
7
2
A
l
e
x
N
e
t
8
3
.
3
0
%
0
.
6
3
0
.
5
7
0
.
5
7
0
.
5
2
0
.
5
0
3
.
3
.
Co
nfusi
o
n
m
a
t
rix
I
t
is
m
o
r
e
ev
id
en
t
f
r
o
m
Fig
u
r
e
6
th
at
th
e
o
u
tco
m
e
d
is
cu
s
s
ed
in
T
ab
le
4
is
q
u
ite
s
a
tis
f
ac
to
r
y
.
Den
s
eNe
t1
2
1
an
d
I
n
ce
p
tio
n
V
3
'
s
co
n
f
u
s
io
n
m
atr
ices
p
r
o
v
id
e
a
clo
s
er
lo
o
k
at
ea
ch
clas
s
's
tr
u
e
p
o
s
itiv
e
an
d
tr
u
e
n
eg
ativ
e
v
al
u
es.
Fig
u
r
e
6
(
a)
-
(
d
)
r
ep
r
esen
ts
all
th
e
ap
p
l
ied
m
o
d
el’
s
co
n
f
u
s
io
n
m
atr
ix
an
d
it
p
r
o
v
i
d
es
th
e
s
tr
o
n
g
r
e
f
er
en
ce
o
f
o
u
r
ac
h
iev
ed
r
esu
lts
.
I
n
ter
m
s
o
f
to
m
ato
leaf
i
d
en
tific
atio
n
,
th
e
u
n
d
e
r
f
itti
n
g
an
d
o
v
er
f
itti
n
g
is
s
u
es a
r
e
ex
tr
em
el
y
s
m
all
an
d
n
o
t v
er
y
s
ig
n
if
ica
n
t.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
6
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
(
a)
VGG1
6
,
(
b
)
I
n
ce
p
tio
n
V
3
,
(
c)
Den
s
eNe
t1
2
1
,
an
d
(
d
)
Alex
Net
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
u
to
ma
ted
to
ma
to
le
a
f d
is
ea
s
e
r
ec
o
g
n
itio
n
u
s
in
g
d
ee
p
co
n
v
o
lu
tio
n
a
l
n
etw
o
r
ks
(
A
mir
S
o
h
el
)
1857
T
h
e
VGG1
6
m
o
d
el'
s
co
n
f
u
s
io
n
m
atr
ix
is
s
h
o
wn
in
Fig
u
r
e
6
(
a)
,
wh
er
e
th
e
t
r
u
e
p
o
s
itiv
e
v
a
lu
e
f
o
r
th
e
f
o
u
r
class
es
is
clea
r
ly
d
is
p
lay
ed
.
Fo
r
th
e
b
lack
m
o
l
d
class
,
th
e
tr
u
e
p
o
s
itiv
e
v
alu
e
is
v
er
y
lo
w,
an
d
f
o
r
th
e
p
o
wd
er
y
m
ild
ew
class
,
b
o
th
t
h
e
tr
u
e
n
e
g
ativ
e
v
alu
e
an
d
th
e
tr
u
e
p
o
s
itiv
e
v
alu
e
ar
e
s
atis
f
ied
,
b
u
t
t
h
e
f
alse
p
o
s
itiv
e
r
ate
is
a
litt
le
p
r
o
b
l
em
atic.
Ho
wev
er
,
t
h
er
e
is
a
m
in
o
r
is
s
u
e
with
th
e
Fig
u
r
e
6
(
b
)
I
n
ce
p
tio
n
V3
co
n
f
u
s
io
n
m
atr
ix
,
wh
ich
s
h
o
ws
a
lo
w
p
er
ce
n
ta
g
e
o
f
f
al
s
e
p
o
s
itiv
e
an
d
n
eg
ativ
e
r
esu
lts
.
Fin
ally
,
it
is
d
is
co
v
er
ed
th
at
Fig
u
r
e
6
(
c)
D
en
s
eNe
t
1
2
1
,
co
n
f
u
s
io
n
m
atr
i
x
,
is
h
ig
h
ly
s
atis
f
y
in
g
o
n
T
P
a
n
d
T
N.
Fig
u
r
e
6
(
d
)
is
th
e
Alex
Net
co
n
f
u
s
io
n
m
atr
ix
,
wh
ich
is
lik
ewise
q
u
ite
co
m
p
ar
ab
le
to
Fig
u
r
es
6
(
b
)
a
n
d
6
(
c)
,
b
u
t
a
litt
le
b
it
f
ar
th
er
awa
y
.
C
o
n
s
id
er
in
g
th
e
B
l
ac
k
_
m
o
ld
d
ata
in
th
e
s
e
co
n
d
class
is
s
o
m
ewh
at
wea
k
er
th
an
th
at
o
f
th
e
o
th
er
s
,
th
er
e
is
a
s
ig
n
if
ican
t
class
v
ar
iatio
n
f
o
r
v
ar
io
u
s
m
o
d
els.
Ho
wev
er
,
co
m
p
a
r
ed
to
r
aw
d
ata,
o
u
r
p
r
ep
r
o
ce
s
s
in
g
tech
n
iq
u
e
h
as v
er
y
h
ig
h
ac
cu
r
ac
y
an
d
lo
wer
s
t
h
e
f
alse
-
p
o
s
itiv
e
r
ate.
3
.
4
.
RO
C
curv
e
Acc
o
r
d
in
g
to
th
e
tr
u
e
p
o
s
itiv
e
r
ate
(
T
PR
)
an
d
f
alse
p
o
s
it
iv
e
r
ate
(
FP
R
)
,
wh
ich
ar
e
d
is
p
lay
ed
o
n
th
e
R
OC
cu
r
v
e
o
f
ev
e
r
y
ap
p
lied
m
o
d
el
in
Fig
u
r
e
7
,
th
is
m
ay
b
e
ac
h
iev
e
d
f
o
r
th
e
m
o
s
t
p
ar
t
o
f
th
e
tim
e,
with
th
e
ex
ce
p
tio
n
o
f
in
s
tan
ce
s
in
w
h
ich
th
e
FP
r
ate
b
ec
o
m
es
ex
ce
s
s
iv
e
b
ec
au
s
e
o
f
d
ata
v
io
latio
n
s
an
d
n
o
is
e
[
2
7
]
.
Ho
wev
er
,
th
e
co
n
f
u
s
io
n
m
at
r
ix
s
ce
n
ar
io
an
d
h
ig
h
lev
el
o
f
s
atis
f
ac
tio
n
in
I
n
ce
p
tio
n
V
3
ar
e
b
o
t
h
p
r
esen
t.
T
h
er
ef
o
r
e,
th
e
R
OC
cu
r
v
e
in
F
ig
u
r
e
7
(
a)
-
(
d
)
p
r
o
v
id
es a
m
o
r
e
s
ig
n
if
ican
t a
n
d
tr
a
n
s
p
ar
en
t
r
esu
lt a
n
aly
s
is
.
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
7
.
R
OC
cu
r
v
e
o
f
(
a)
VGG1
6
,
(
b
)
I
n
ce
p
tio
n
V3
,
(
c)
D
en
s
eNe
t1
2
1
,
an
d
(
d
)
Alex
Net
T
h
e
R
OC
cu
r
v
e
o
f
VGG
1
6
is
s
h
o
wn
in
Fig
u
r
e
7
(
a)
.
Fro
m
th
is
g
r
a
p
h
,
we
ca
n
o
b
s
er
v
e
t
h
at
class
1
f
alls
u
n
d
er
th
e
d
iag
o
n
al
lin
e
f
r
o
m
th
e
s
tar
t,
wh
ile
o
th
er
class
es
im
p
r
o
v
e
af
ter
a
ce
r
tain
a
m
o
u
n
t
o
f
tim
e.
T
h
e
d
iag
o
n
al
lin
e
in
th
is
g
r
ap
h
ac
tu
ally
r
ep
r
esen
ts
n
o
s
k
ill.
I
n
I
n
ce
p
tio
n
V3
'
s
R
O
C
,
Fig
u
r
e
7
(
b
)
,
we
ca
n
o
b
s
er
v
e
th
at
wh
ile
clas
s
0
h
as
a
f
e
w
s
p
o
r
ad
ic
p
r
o
b
lem
s
,
o
th
er
class
es
f
u
n
ctio
n
ad
m
ir
ab
ly
a
n
d
r
ec
eiv
e
p
o
s
itiv
e
ev
alu
atio
n
s
f
r
o
m
th
eir
ass
ess
m
en
ts
.
Ou
r
b
est
ac
cu
r
ac
y
o
n
t
h
e
Den
s
eNe
t
1
2
1
m
o
d
el
is
s
h
o
wn
h
er
e,
w
h
er
e
we
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
2
,
Ap
r
il
20
25
:
1
8
5
0
-
1
8
6
0
1858
ca
n
o
b
s
er
v
e
th
at
o
cc
asio
n
ally
,
p
ar
ticu
lar
ly
b
etwe
en
0
.
2
an
d
0
.
5
,
two
class
es
ar
e
b
elo
w
th
e
d
iag
o
n
al
lin
e;
h
o
wev
er
,
f
r
o
m
th
e
f
ir
s
t
to
th
e
last
ep
o
ch
,
we
co
n
s
is
ten
tly
o
b
tain
an
ex
ce
llen
t
AUC
v
alu
e
f
r
o
m
th
e
c
u
r
v
e.
Fig
u
r
e
7
(
c
)
p
r
o
v
i
d
es
a
s
atis
f
ac
to
r
y
s
u
m
m
a
r
y
an
d
s
h
o
ws
th
e
R
OC
o
f
Den
s
Net1
2
1
at
th
e
s
t
an
d
ar
d
lev
el
f
o
r
all
class
es.
T
h
e
R
O
C
cu
r
v
e
f
o
r
t
h
e
Alex
Net
m
o
d
el
i
n
Fig
u
r
e
7
(
d
)
is
a
litt
le
less
ac
cu
r
ate
b
ec
au
s
e
th
e
n
ex
t
t
h
r
ee
class
es
f
all
with
in
th
e
d
iag
o
n
al
lin
e.
Ho
wev
er
,
th
e
g
r
ea
t
est
AU
C
v
alu
e
o
f
7
1
.
6
3
%
in
Den
s
eNe
t
1
2
1
is
o
b
tain
ed
b
y
m
ea
s
u
r
em
en
t,
in
d
icatin
g
a
v
er
y
h
ea
lth
y
an
d
g
r
atif
y
in
g
ac
c
u
r
ac
y
v
alu
e
f
o
r
t
h
e
class
if
icatio
n
o
f
to
m
ato
leaf
d
is
ea
s
e.
3
.
5
.
Resul
t
dis
cus
s
io
n o
f
ra
w
da
t
a
s
et
Af
ter
p
er
f
o
r
m
i
n
g
th
e
s
am
e
m
o
d
els
o
n
th
e
r
aw
d
ataset
with
o
u
t
an
y
k
in
d
o
f
p
r
ep
r
o
c
ess
in
g
,
ju
s
t
r
esizin
g
th
e
im
ag
es
in
a
2
2
4
*
2
2
4
s
h
ap
e
,
th
e
r
esu
ltan
t
v
alu
es
ar
e
g
iv
en
in
T
ab
le
5
,
wh
er
e
I
n
ce
p
tio
n
V3
p
er
f
o
r
m
ed
o
n
e
o
f
t
h
e
b
est
b
o
th
in
r
aw
an
d
p
r
ep
r
o
ce
s
s
ed
d
ataset
m
o
d
els.
Ho
wev
er
,
VGG1
6
p
er
f
o
r
m
ed
th
e
b
est
o
n
r
aw
d
ataset
with
8
3
.
3
3
%
ac
cu
r
ac
y
,
wh
er
ea
s
Alex
Net
an
d
Den
s
eNe
t
-
1
2
1
ar
e
at
6
3
.
8
9
%
an
d
6
9
.
8
8
%,
r
esp
ec
tiv
ely
.
T
h
e
tr
u
e
r
ep
r
esen
tatio
n
o
f
th
e
r
aw
d
ata'
s
m
ea
s
u
r
em
en
t
m
atr
ix
is
f
o
u
n
d
in
T
ab
le
5
.
T
h
e
ac
cu
r
ac
y
ac
h
iev
ed
af
te
r
r
esh
ap
i
n
g
is
s
lig
h
tly
h
ig
h
er
o
n
th
e
VGG1
6
b
u
t
m
u
c
h
lo
wer
o
n
th
e
o
t
h
er
m
o
d
els.
T
h
e
o
th
e
r
m
atr
ix
,
wh
ich
h
as
v
er
y
lo
w
p
r
ec
is
io
n
an
d
AUC
v
alu
es,
ca
u
s
es
u
n
d
er
f
it
p
r
o
b
lem
s
in
ev
e
r
y
m
o
d
el
f
o
r
ev
er
y
class
.
T
h
u
s
,
we
m
ay
co
n
cl
u
d
e
th
at
o
u
r
p
r
ep
r
o
ce
s
s
in
g
m
eth
o
d
is
q
u
ite
b
en
ef
icial
f
o
r
th
is
im
p
o
r
tan
t iss
u
e.
T
ab
le
5
.
I
m
ag
e
m
ea
s
u
r
e
m
en
t
m
eth
o
d
s
o
n
r
aw
d
ataset
M
o
d
e
l
s
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
Jac
c
a
r
d
AUC
V
G
G
1
6
8
3
.
3
3
%
0
.
4
0
0
.
4
2
0
.
4
0
0
.
1
3
0
.
4
4
I
n
c
e
p
t
i
o
n
V
3
8
0
.
5
6
%
0
.
4
3
0
.
3
9
0
.
3
5
0
.
1
6
0
.
5
7
D
e
n
seN
e
t
1
2
1
6
9
.
8
8
%
0
.
3
5
0
.
3
3
0
.
2
8
0
.
2
0
0
.
6
2
A
l
e
x
N
e
t
6
3
.
8
9
%
0
.
2
8
0
.
3
9
0
.
3
2
0
.
4
1
0
.
4
9
4.
CO
NCLU
SI
O
N
On
th
e
to
m
ato
lea
f
d
is
ea
s
e
d
a
taset,
VGG1
6
,
I
n
ce
p
tio
n
V
3
,
Alex
Net
,
an
d
Den
s
eNe
t
-
1
2
1
wer
e
test
ed
.
Den
s
eNe
t
-
1
2
1
an
d
I
n
ce
p
tio
n
V3
p
er
f
o
r
m
ed
well,
wh
e
r
ea
s
VGG1
6
an
d
Alex
N
et
wer
e
less
ac
cu
r
ate.
Ou
r
d
ataset
ca
n
in
f
lu
en
ce
th
is
f
iel
d
o
f
s
tu
d
y
,
an
d
o
u
r
d
ata
p
r
etr
e
atm
en
t
m
eth
o
d
s
an
d
r
aw
-
tr
ea
t
ed
d
ata
co
m
p
a
r
is
o
n
ca
n
s
o
lid
if
y
it.
Au
to
m
ated
leaf
d
is
ea
s
e
r
ec
o
g
n
itio
n
will
aid
to
m
ato
f
ar
m
e
r
s
co
n
s
id
er
ab
ly
.
W
e
tr
y
to
clo
s
e
th
e
r
esear
ch
g
a
p
u
s
in
g
p
r
e
v
io
u
s
r
e
s
ea
r
ch
er
s
'
an
aly
s
es.
W
e
ca
n
d
etec
t
an
d
v
er
if
y
o
u
r
p
leased
d
ata
co
llectio
n
.
W
ith
s
u
ch
r
esear
ch
an
d
in
n
o
v
atio
n
,
we
ca
n
cr
ea
te
ad
v
an
ce
s
ci
en
tific
ag
r
icu
ltu
r
e.
T
h
is
s
tu
d
y
ca
n
s
er
v
e
as
th
e
f
o
u
n
d
atio
n
f
o
r
th
e
id
en
tific
ati
o
n
o
f
to
m
ato
leaf
d
is
ea
s
es.
I
n
a
lar
g
e
ar
ea
,
m
o
r
e
an
d
m
o
r
e
l
ea
f
d
is
ea
s
es
ca
n
b
e
ad
d
ed
to
d
if
f
e
r
en
t
ar
ea
s
t
h
at
a
r
e
s
m
o
o
th
a
n
d
c
o
v
er
e
d
.
I
t
s
h
o
u
ld
b
e
n
o
ted
th
at,
o
win
g
to
a
lack
o
f
r
eso
u
r
ce
s
,
o
u
r
o
r
ig
in
al
d
ata
was
ex
tr
e
m
ely
n
o
is
y
an
d
u
n
b
alan
ce
d
.
Ho
wev
er
,
o
u
r
n
o
r
m
aliza
tio
n
an
d
p
r
e
-
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
wer
e
tr
a
n
s
f
o
r
m
e
d
in
to
a
h
ea
lth
y
d
ataset
th
at
is
u
s
ef
u
l
f
o
r
th
e
a
g
r
icu
ltu
r
e
s
ec
to
r
.
I
d
en
tific
atio
n
o
f
cr
o
p
lea
f
d
is
ea
s
es
th
r
o
u
g
h
au
t
o
m
atio
n
u
s
in
g
m
ac
h
in
e
lea
r
n
i
n
g
will
b
e
g
r
ea
tly
en
h
an
ce
d
,
a
n
d
it
will
b
e
u
s
ef
u
l
in
th
e
co
m
m
u
n
ity
f
o
r
id
en
tif
y
in
g
m
an
y
o
th
er
c
r
o
p
d
is
ea
s
es
in
ad
d
itio
n
to
t
o
m
ato
leaf
d
is
ea
s
e.
Fo
r
r
o
o
t
-
lev
el
u
s
er
s
wh
o
ar
e
d
ir
ec
tly
in
v
o
lv
ed
in
m
ain
tain
in
g
th
is
s
y
s
tem
,
an
ex
p
an
d
ed
v
er
s
io
n
o
f
th
e
p
r
o
g
r
a
m
o
r
m
o
b
ile
ap
p
licatio
n
m
ay
i
n
clu
d
e
a
h
y
b
r
id
au
to
m
ated
d
etec
tio
n
t
o
o
l
.
W
e
ca
n
p
r
o
m
is
e
th
at
in
th
e
f
u
tu
r
e,
i
d
eo
lo
g
ical
p
r
o
b
lem
s
in
c
o
n
tem
p
o
r
ar
y
r
e
s
ea
r
ch
will
b
e
r
aised
b
y
u
s
in
g
lar
g
e
r
d
atasets
to
d
is
co
v
e
r
v
ar
io
u
s
cr
o
p
lea
f
d
is
ea
s
es in
a
s
in
g
le
f
r
am
e.
RE
F
E
R
E
NC
E
S
[
1
]
P
.
Tm,
A
.
P
r
a
n
a
t
h
i
,
K
.
S
a
i
A
s
h
r
i
t
h
a
,
N
.
B
.
C
h
i
t
t
a
r
a
g
i
,
a
n
d
S
.
G
.
K
o
o
l
a
g
u
d
i
,
“
T
o
ma
t
o
l
e
a
f
d
i
se
a
se
d
e
t
e
c
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
s,
”
i
n
2
0
1
8
E
l
e
v
e
n
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
n
t
e
m
p
o
r
a
ry
C
o
m
p
u
t
i
n
g
(
I
C
3
)
,
A
u
g
.
2
0
1
8
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
3
.
2
0
1
8
.
8
5
3
0
5
3
2
.
[
2
]
M
.
E.
H
.
C
h
o
w
d
h
u
r
y
e
t
a
l
.
,
“
A
u
t
o
m
a
t
i
c
a
n
d
r
e
l
i
a
b
l
e
l
e
a
f
d
i
sea
s
e
d
e
t
e
c
t
i
o
n
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s,
”
Ag
ri
En
g
i
n
e
e
r
i
n
g
,
v
o
l
.
3
,
n
o
.
2
,
p
p
.
2
9
4
–
3
1
2
,
M
a
y
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
a
g
r
i
e
n
g
i
n
e
e
r
i
n
g
3
0
2
0
0
2
0
.
[
3
]
S
.
A
p
a
r
n
a
a
n
d
R
.
A
a
r
t
h
i
,
“
S
e
g
m
e
n
t
a
t
i
o
n
o
f
t
o
mat
o
p
l
a
n
t
l
e
a
f
,
”
i
n
Pr
o
g
r
e
ss
i
n
I
n
t
e
l
l
i
g
e
n
t
C
o
m
p
u
t
i
n
g
T
e
c
h
n
i
q
u
e
s:
T
h
e
o
ry,
Pr
a
c
t
i
c
e
,
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
2
0
1
8
,
p
p
.
1
4
9
–
1
5
6
.
[
4
]
M
.
A
.
I
sl
a
m,
M
.
M
.
R
a
h
m
a
n
,
A
.
A
.
S
h
o
h
a
n
,
R
.
A
.
P
u
l
o
k
,
S
.
A
k
t
e
r
,
a
n
d
M
.
T.
A
h
m
e
d
,
“
D
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
-
b
a
se
d
a
p
p
r
o
a
c
h
t
o
i
d
e
n
t
i
f
y
t
h
e
p
a
d
d
y
l
e
a
f
d
i
sea
s
e
u
s
i
n
g
R
e
sN
e
t
5
0
-
V
2
,
”
i
n
2
0
2
3
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
E
v
o
l
u
t
i
o
n
a
ry
Al
g
o
ri
t
h
m
s
a
n
d
S
o
f
t
C
o
m
p
u
t
i
n
g
T
e
c
h
n
i
q
u
e
s
(
EA
S
C
T
)
,
O
c
t
.
2
0
2
3
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
EA
S
C
T5
9
4
7
5
.
2
0
2
3
.
1
0
3
9
3
5
8
6
.
[
5
]
S
.
S
h
e
d
t
h
i
B
,
M
.
S
i
d
d
a
p
p
a
,
S
.
S
h
e
t
t
y
,
a
n
d
V
.
S
h
e
t
t
y
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
a
r
e
c
a
n
u
t
u
s
i
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s,”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
t
ri
c
a
l
a
n
d
C
o
m
p
u
t
e
r
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
3
,
n
o
.
2
,
p
p
.
1
9
1
4
–
1
9
2
1
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
3
i
2
.
p
p
1
9
1
4
-
1
9
2
1
.
[
6
]
M
.
A
g
a
r
w
a
l
,
A
.
S
i
n
g
h
,
S
.
A
r
j
a
r
i
a
,
A
.
S
i
n
h
a
,
a
n
d
S
.
G
u
p
t
a
,
“
To
Le
D
:
t
o
ma
t
o
l
e
a
f
d
i
se
a
se
d
e
t
e
c
t
i
o
n
u
si
n
g
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
Pr
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
6
7
,
p
p
.
2
9
3
–
3
0
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s.
2
0
2
0
.
0
3
.
2
2
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
u
to
ma
ted
to
ma
to
le
a
f d
is
ea
s
e
r
ec
o
g
n
itio
n
u
s
in
g
d
ee
p
co
n
v
o
lu
tio
n
a
l
n
etw
o
r
ks
(
A
mir
S
o
h
el
)
1859
[
7
]
X
.
C
h
e
n
,
G
.
Z
h
o
u
,
A
.
C
h
e
n
,
J
.
Y
i
,
W
.
Z
h
a
n
g
,
a
n
d
Y
.
H
u
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
t
o
m
a
t
o
l
e
a
f
d
i
s
e
a
s
e
s
b
a
se
d
o
n
c
o
m
b
i
n
a
t
i
o
n
o
f
A
B
C
K
-
B
W
TR
a
n
d
B
-
A
R
N
e
t
,
”
C
o
m
p
u
t
e
rs
a
n
d
E
l
e
c
t
r
o
n
i
c
s
i
n
A
g
r
i
c
u
l
t
u
r
e
,
v
o
l
.
1
7
8
,
N
o
v
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
a
g
.
2
0
2
0
.
1
0
5
7
3
0
.
[
8
]
D
.
Ji
a
n
g
,
F
.
Li
,
Y
.
Y
a
n
g
,
a
n
d
S
.
Y
u
,
“
A
t
o
m
a
t
o
l
e
a
f
d
i
se
a
ses
c
l
a
ss
i
f
i
c
a
t
i
o
n
m
e
t
h
o
d
b
a
se
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
,
”
i
n
2
0
2
0
C
h
i
n
e
se
C
o
n
t
r
o
l
a
n
d
D
e
c
i
s
i
o
n
C
o
n
f
e
r
e
n
c
e
(
C
C
D
C
)
,
A
u
g
.
2
0
2
0
,
p
p
.
1
4
4
6
–
1
4
5
0
,
d
o
i
:
1
0
.
1
1
0
9
/
C
C
D
C
4
9
3
2
9
.
2
0
2
0
.
9
1
6
4
4
5
7
.
[
9
]
C
.
Z
h
o
u
,
S
.
Zh
o
u
,
J.
X
i
n
g
,
a
n
d
J.
S
o
n
g
,
“
T
o
ma
t
o
l
e
a
f
d
i
sea
s
e
i
d
e
n
t
i
f
i
c
a
t
i
o
n
b
y
r
e
st
r
u
c
t
u
r
e
d
d
e
e
p
r
e
s
i
d
u
a
l
d
e
n
se
n
e
t
w
o
r
k
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
2
8
8
2
2
–
2
8
8
3
1
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
0
5
8
9
4
7
.
[
1
0
]
K
.
B
a
l
a
k
r
i
sh
n
a
a
n
d
M
.
R
a
o
,
“
T
o
m
a
t
o
p
l
a
n
t
l
e
a
v
e
s
d
i
s
e
a
se
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
K
N
N
a
n
d
P
N
N
,
”
I
n
t
e
rn
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
s
i
o
n
a
n
d
I
m
a
g
e
Pr
o
c
e
s
si
n
g
,
v
o
l
.
9
,
n
o
.
1
,
p
p
.
5
1
–
6
3
,
J
a
n
.
2
0
1
9
,
d
o
i
:
1
0
.
4
0
1
8
/
I
JC
V
I
P
.
2
0
1
9
0
1
0
1
0
4
.
[
1
1
]
V
.
G
o
n
z
a
l
e
z
-
H
u
i
t
r
o
n
,
J.
A
.
Le
ó
n
-
B
o
r
g
e
s,
A
.
E.
R
o
d
r
i
g
u
e
z
-
M
a
t
a
,
L.
E
.
A
mab
i
l
i
s
-
S
o
s
a
,
B
.
R
a
m
í
r
e
z
-
P
e
r
e
d
a
,
a
n
d
H
.
R
o
d
r
i
g
u
e
z
,
“
D
i
se
a
se
d
e
t
e
c
t
i
o
n
i
n
t
o
ma
t
o
l
e
a
v
e
s
v
i
a
C
N
N
w
i
t
h
l
i
g
h
t
w
e
i
g
h
t
a
r
c
h
i
t
e
c
t
u
r
e
s
i
m
p
l
e
me
n
t
e
d
i
n
R
a
s
p
b
e
r
r
y
P
i
4
,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
r
o
n
i
c
s
i
n
A
g
r
i
c
u
l
t
u
r
e
,
v
o
l
.
1
8
1
,
F
e
b
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
a
g
.
2
0
2
0
.
1
0
5
9
5
1
.
[
1
2
]
A
.
A
b
b
a
s
,
S
.
J
a
i
n
,
M
.
G
o
u
r
,
a
n
d
S
.
V
a
n
k
u
d
o
t
h
u
,
“
T
o
ma
t
o
p
l
a
n
t
d
i
s
e
a
s
e
d
e
t
e
c
t
i
o
n
u
si
n
g
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
w
i
t
h
C
-
G
A
N
s
y
n
t
h
e
t
i
c
i
ma
g
e
s,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
r
o
n
i
c
s i
n
A
g
ri
c
u
l
t
u
r
e
,
v
o
l
.
1
8
7
,
A
u
g
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mp
a
g
.
2
0
2
1
.
1
0
6
2
7
9
.
[
1
3
]
K
.
Z
h
a
n
g
,
Q
.
W
u
,
A
.
L
i
u
,
a
n
d
X
.
M
e
n
g
,
“
C
a
n
d
e
e
p
l
e
a
r
n
i
n
g
i
d
e
n
t
i
f
y
t
o
ma
t
o
l
e
a
f
d
i
s
e
a
se
?
,
”
Ad
v
a
n
c
e
s
i
n
M
u
l
t
i
m
e
d
i
a
,
v
o
l
.
2
0
1
8
,
p
p
.
1
–
1
0
,
S
e
p
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
1
8
/
6
7
1
0
8
6
5
.
[
1
4
]
H
.
H
o
n
g
,
J
.
L
i
n
,
a
n
d
F
.
H
u
a
n
g
,
“
T
o
m
a
t
o
d
i
s
e
a
se
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
b
y
d
e
e
p
l
e
a
r
n
i
n
g
,
”
i
n
2
0
2
0
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
B
i
g
D
a
t
a
,
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
I
n
t
e
r
n
e
t
o
f
T
h
i
n
g
s
E
n
g
i
n
e
e
ri
n
g
(
I
C
BAIE)
,
J
u
n
.
2
0
2
0
,
p
p
.
2
5
–
2
9
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
B
A
I
E4
9
9
9
6
.
2
0
2
0
.
0
0
0
1
2
.
[
1
5
]
A
.
K
u
mar
a
n
d
M
.
V
a
n
i
,
“
I
ma
g
e
b
a
s
e
d
t
o
m
a
t
o
l
e
a
f
d
i
sea
s
e
d
e
t
e
c
t
i
o
n
,
”
i
n
2
0
1
9
1
0
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
C
o
m
p
u
t
i
n
g
,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
N
e
t
w
o
r
k
i
n
g
T
e
c
h
n
o
l
o
g
i
e
s
(
I
C
C
C
N
T
)
,
Ju
l
.
2
0
1
9
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
C
N
T
4
5
6
7
0
.
2
0
1
9
.
8
9
4
4
6
9
2
.
[
1
6
]
S
.
I
.
P
r
o
t
t
a
sh
a
a
n
d
S
.
M
.
S
.
R
e
z
a
,
“
A
c
l
a
ss
i
f
i
c
a
t
i
o
n
m
o
d
e
l
b
a
s
e
d
o
n
d
e
p
t
h
w
i
se
se
p
a
r
a
b
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
t
o
i
d
e
n
t
i
f
y
r
i
c
e
p
l
a
n
t
d
i
se
a
ses,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
2
,
n
o
.
4
,
p
p
.
3
6
4
2
–
3
6
5
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
2
i
4
.
p
p
3
6
4
2
-
3
6
5
4
.
[
1
7
]
Y
.
W
a
n
g
,
Q
.
C
h
e
n
,
a
n
d
B
.
Z
h
a
n
g
,
“
I
mag
e
e
n
h
a
n
c
e
me
n
t
b
a
s
e
d
o
n
e
q
u
a
l
a
r
e
a
d
u
a
l
i
s
t
i
c
s
u
b
-
i
ma
g
e
h
i
s
t
o
g
r
a
m
e
q
u
a
l
i
z
a
t
i
o
n
m
e
t
h
o
d
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
C
o
n
su
m
e
r E
l
e
c
t
ro
n
i
c
s
,
v
o
l
.
4
5
,
n
o
.
1
,
p
p
.
6
8
–
7
5
,
1
9
9
9
,
d
o
i
:
1
0
.
1
1
0
9
/
3
0
.
7
5
4
4
1
9
.
[
1
8
]
A
.
S
o
h
e
l
e
t
a
l
.
,
“
S
u
n
f
l
o
w
e
r
d
i
se
a
se
i
d
e
n
t
i
f
i
c
a
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
:
a
d
a
t
a
-
d
r
i
v
e
n
a
p
p
r
o
a
c
h
,
”
i
n
2
0
2
3
2
6
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
(
I
C
C
I
T
)
,
D
e
c
.
2
0
2
3
,
p
p
.
1
–
6,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
I
T6
0
4
5
9
.
2
0
2
3
.
1
0
4
4
1
2
8
5
.
[
1
9
]
M
.
A
.
I
sl
a
m
e
t
a
l
.
,
“
C
o
mp
r
e
h
e
n
s
i
v
e
a
n
a
l
y
si
s
o
f
C
N
N
a
n
d
Y
O
LO
v
5
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
m
o
d
e
l
t
o
c
l
a
ss
i
f
y
p
h
y
t
o
m
e
d
i
c
i
n
e
t
r
e
e
’
s
l
e
a
f
d
i
s
e
a
se
,
”
Pre
p
ri
n
t
,
O
c
t
.
1
0
,
2
0
2
2
,
d
o
i
:
1
0
.
2
1
2
0
3
/
r
s.
3
.
r
s
-
2
0
9
9
5
3
4
/
v
1
.
[
2
0
]
S
.
S
h
a
k
i
l
,
A
.
A
.
K
.
A
k
a
s
h
,
N
.
N
a
b
i
,
M
.
H
a
s
sa
n
,
a
n
d
A
.
H
a
q
u
e
,
“
P
i
t
h
a
N
e
t
:
a
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
-
b
a
se
d
a
p
p
r
o
a
c
h
f
o
r
t
r
a
d
i
t
i
o
n
a
l
p
i
t
h
a
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
3
,
n
o
.
5
,
p
p
.
5
4
3
1
–
5
4
4
3
,
O
c
t
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
3
i
5
.
p
p
5
4
3
1
-
5
4
4
3
.
[
2
1
]
M
.
A
l
sw
a
i
t
t
i
,
L.
Zi
h
a
o
,
W
.
A
l
o
m
o
u
s
h
,
A
.
A
l
r
o
s
a
n
,
a
n
d
K
.
A
l
i
ssa
,
“
Ef
f
e
c
t
i
v
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
b
i
r
d
s’
s
p
e
c
i
e
s
b
a
se
d
o
n
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
2
,
n
o
.
4
,
p
p
.
4
1
7
2
–
4
1
8
4
,
A
u
g
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
2
i
4
.
p
p
4
1
7
2
-
4
1
8
4
.
[
2
2
]
A
.
M
.
H
a
ss
a
n
,
M
.
B
.
E
l
-
M
a
sh
a
d
e
,
a
n
d
A
.
A
b
o
s
h
o
s
h
a
,
“
D
e
e
p
l
e
a
r
n
i
n
g
f
o
r
c
a
n
c
e
r
t
u
mo
r
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
a
n
d
f
e
a
t
u
r
e
c
o
n
c
a
t
e
n
a
t
i
o
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
E
l
e
c
t
ri
c
a
l
a
n
d
C
o
m
p
u
t
e
r
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
2
,
n
o
.
6
,
p
p
.
6
7
3
6
–
6
7
4
3
,
D
e
c
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
2
i
6
.
p
p
6
7
3
6
-
6
7
4
3
.
[
2
3
]
S
.
N
a
n
d
h
i
n
i
a
n
d
K
.
A
s
h
o
k
k
u
mar,
“
A
n
a
u
t
o
ma
t
i
c
p
l
a
n
t
l
e
a
f
d
i
s
e
a
s
e
i
d
e
n
t
i
f
i
c
a
t
i
o
n
u
si
n
g
D
e
n
seNe
t
-
1
2
1
a
r
c
h
i
t
e
c
t
u
r
e
w
i
t
h
a
mu
t
a
t
i
o
n
-
b
a
se
d
h
e
n
r
y
g
a
s
s
o
l
u
b
i
l
i
t
y
o
p
t
i
mi
z
a
t
i
o
n
a
l
g
o
r
i
t
h
m,
”
N
e
u
ra
l
C
o
m
p
u
t
i
n
g
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
3
4
,
n
o
.
7
,
p
p
.
5
5
1
3
–
5
5
3
4
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s0
0
5
2
1
-
0
2
1
-
0
6
7
1
4
-
z.
[
2
4
]
S
.
B
e
t
r
a
b
e
t
a
n
d
C
.
K
.
B
h
o
g
a
y
t
a
,
“
S
t
r
u
c
t
u
r
a
l
si
m
i
l
a
r
i
t
y
b
a
s
e
d
i
m
a
g
e
q
u
a
l
i
t
y
a
ssessm
e
n
t
u
si
n
g
f
u
l
l
r
e
f
e
r
e
n
c
e
m
e
t
h
o
d
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
S
c
i
e
n
t
i
f
i
c
E
n
g
i
n
e
e
r
i
n
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
4
,
n
o
.
4
,
p
p
.
2
5
2
–
2
5
5
,
2
0
1
5
,
d
o
i
:
1
0
.
1
7
9
5
0
/
i
j
se
t
/
v
4
s
4
/
4
0
7
.
[
2
5
]
J.
M
i
a
,
H
.
I
.
B
i
j
o
y
,
S
.
U
d
d
i
n
,
a
n
d
D
.
M
.
R
a
z
a
,
“
R
e
a
l
-
t
i
m
e
h
e
r
b
l
e
a
v
e
s
l
o
c
a
l
i
z
a
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
Y
O
L
O
,
”
i
n
2
0
2
1
1
2
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
C
o
m
p
u
t
i
n
g
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
N
e
t
w
o
r
k
i
n
g
T
e
c
h
n
o
l
o
g
i
e
s
(
I
C
C
C
N
T
)
,
J
u
l
.
2
0
2
1
,
p
p
.
1
–
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
C
N
T
5
1
5
2
5
.
2
0
2
1
.
9
5
7
9
7
1
8
.
[
2
6
]
H
.
A
l
F
a
h
i
m,
M
.
A
.
H
a
s
a
n
,
M
.
H
.
I
.
B
i
j
o
y
,
A
.
W
.
R
e
z
a
,
a
n
d
M
.
S
.
A
r
e
f
i
n
,
“
S
e
e
d
s
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
:
a
r
e
v
i
e
w
,
”
I
n
t
e
l
l
i
g
e
n
t
C
o
m
p
u
t
i
n
g
a
n
d
O
p
t
i
m
i
z
a
t
i
o
n
L
e
c
t
u
r
e
N
o
t
e
s
i
n
N
e
t
w
o
r
k
s a
n
d
S
y
st
e
m
s
,
p
p
.
1
6
8
–
182
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
5
0
3
3
0
-
6
_
1
7
.
[
2
7
]
M
.
P
.
M
a
h
mu
d
,
M
.
A
.
A
l
i
,
S
.
A
k
t
e
r
,
a
n
d
M
.
H
.
I
.
B
i
j
o
y
,
“
L
y
c
h
e
e
t
r
e
e
d
i
s
e
a
s
e
c
l
a
ssi
f
i
c
a
t
i
o
n
a
n
d
p
r
e
d
i
c
t
i
o
n
u
si
n
g
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
i
n
2
0
2
2
1
3
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
i
n
g
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
N
e
t
w
o
r
k
i
n
g
T
e
c
h
n
o
l
o
g
i
e
s
(
I
C
C
C
N
T
)
,
O
c
t
.
2
0
2
2
,
p
p
.
1
–
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
C
N
T
5
4
8
2
7
.
2
0
2
2
.
9
9
8
4
2
8
6
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Am
ir
S
o
h
e
l
wo
r
k
s
a
s
a
se
n
i
o
r
lec
tu
re
r
a
t
Da
ffo
d
i
l
I
n
tern
a
ti
o
n
a
l
Un
i
v
e
rsity
in
S
a
v
a
r,
Dh
a
k
a
,
Ba
n
g
lad
e
sh
,
i
n
t
h
e
De
p
a
rtme
n
t
o
f
CS
E
.
He
re
c
e
iv
e
d
h
is
B.
S
c
.
i
n
c
o
m
p
u
te
r
sc
ien
c
e
a
n
d
e
n
g
in
e
e
ri
n
g
fr
o
m
Da
ffo
d
il
I
n
tern
a
ti
o
n
a
l
U
n
iv
e
rsit
y
in
S
a
v
a
r,
Dh
a
k
a
,
Ba
n
g
lad
e
sh
,
a
n
d
h
is
M
.
S
c
.
in
c
o
m
p
u
ter
s
c
ien
c
e
fro
m
Ja
h
a
n
g
ir
n
a
g
a
r
Un
iv
e
rsity
i
n
S
a
v
a
r,
Dh
a
k
a
,
Ba
n
g
lad
e
sh
.
He
is
c
u
rre
n
tl
y
fo
c
u
sin
g
o
n
in
f
o
rm
a
ti
o
n
s
y
ste
m
s,
d
e
e
p
lea
rn
in
g
,
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
m
ir.
c
se
@d
iu
.
e
d
u
.
b
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.