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.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
,
p
p
.
1399
~
1
4
0
6
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
1
6
i
3
.
pp
1
3
9
9
-
1
4
0
6
1399
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Tra
nsfo
rmer
-
ba
s
ed hybrid
cla
ss
ifi
ca
tion for pla
n
t
le
a
f
disea
se
detec
tion usin
g
v
i
sio
n t
ra
nsfo
rmer
,
principa
l
co
mpo
nent
a
na
ly
sis
,
a
nd
supp
o
rt
v
ector ma
ch
i
ne
Vij
a
y
a
la
k
s
hm
i S
.
Abbi
g
er
i,
G
ee
t
ha
D
.
Dev
a
na
g
a
v
i
S
c
h
o
o
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
R
EV
A
U
n
i
v
e
r
si
t
y
,
B
e
n
g
a
l
u
r
u
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
9
,
2
0
2
6
R
ev
is
ed
Ma
r
2
0
,
2
0
2
6
Acc
ep
ted
Ap
r
2
7
,
2
0
2
6
P
lan
t
d
ise
a
se
s
re
m
a
in
a
c
rit
ica
l
c
h
a
ll
e
n
g
e
in
a
g
ric
u
lt
u
re
,
c
a
u
sin
g
su
b
sta
n
ti
a
l
y
ield
l
o
ss
e
s
a
n
d
t
h
re
a
ten
in
g
fo
o
d
se
c
u
rit
y
.
In
th
is
w
o
rk
,
we
p
ro
p
o
s
e
a
h
y
b
ri
d
d
e
e
p
fe
a
tu
re
e
n
g
i
n
e
e
rin
g
fra
m
e
wo
rk
th
a
t
in
te
g
ra
tes
d
e
e
p
lea
rn
in
g
-
b
a
se
d
fe
a
tu
re
e
x
trac
ti
o
n
with
c
las
sic
a
l
m
a
c
h
in
e
lea
rn
in
g
fo
r
a
c
c
u
ra
te
p
l
a
n
t
d
ise
a
se
d
e
tec
ti
o
n
.
A
p
re
train
e
d
v
isio
n
t
ra
n
sfo
rm
e
r
(ViT)
m
o
d
e
l
is
e
m
p
lo
y
e
d
to
e
x
trac
t
d
isc
rimin
a
ti
v
e
fe
a
tu
re
s
fro
m
lea
f
ima
g
e
s,
e
ffe
c
ti
v
e
l
y
c
a
p
tu
rin
g
c
o
m
p
lex
sp
a
ti
a
l
re
latio
n
sh
i
p
s.
To
a
d
d
re
ss
t
h
e
c
u
rse
o
f
d
ime
n
sio
n
a
li
ty
,
p
rin
c
i
p
a
l
c
o
m
p
o
n
e
n
t
a
n
a
l
y
sis
(
P
CA)
is
a
p
p
li
e
d
,
re
tain
i
n
g
9
8
%
o
f
t
h
e
v
a
rian
c
e
wh
il
e
re
d
u
c
in
g
fe
a
tu
re
sp
a
c
e
c
o
m
p
lex
it
y
.
Th
e
re
fin
e
d
fe
a
tu
re
s
a
re
th
e
n
c
las
sified
u
si
n
g
a
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
(S
VM)
o
p
t
imiz
e
d
th
r
o
u
g
h
h
y
p
e
r
p
a
ra
m
e
ter
tu
n
in
g
.
Ex
p
e
rime
n
tal
re
su
lt
s
o
n
th
e
b
e
a
n
lea
f
les
io
n
s
d
a
tas
e
t
d
e
m
o
n
stra
te
stro
n
g
p
e
rfo
rm
a
n
c
e
,
a
c
h
iev
in
g
9
2
%
a
c
c
u
ra
c
y
a
n
d
a
we
ig
h
ted
F1
-
sc
o
re
o
f
0
.
9
2
.
Th
e
p
ro
p
o
s
e
d
ViT
–
P
CA
–
S
VM
p
ip
e
li
n
e
e
ffe
c
ti
v
e
ly
b
a
lan
c
e
s
a
c
c
u
ra
c
y
,
c
o
m
p
u
tati
o
n
a
l
e
fficie
n
c
y
,
a
n
d
g
e
n
e
ra
li
z
a
ti
o
n
,
m
a
k
in
g
it
a
p
ro
m
isin
g
so
lu
ti
o
n
fo
r
re
a
l
-
ti
m
e
s
m
a
rt
fa
rm
in
g
a
p
p
l
ica
ti
o
n
s
.
K
ey
w
o
r
d
s
:
Featu
r
e
en
g
in
ee
r
i
n
g
Plan
t le
af
d
is
ea
s
e
d
etec
tio
n
Prin
cip
al
co
m
p
o
n
en
t a
n
al
y
s
is
Su
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
Vis
io
n
tr
an
s
f
o
r
m
er
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
:
Vijay
alak
s
h
m
i S
.
Ab
b
ig
er
i
Sch
o
o
l o
f
C
o
m
p
u
ter
Scien
ce
a
n
d
E
n
g
in
ee
r
in
g
,
R
E
VA
Un
iv
er
s
ity
B
en
g
alu
r
u
,
I
n
d
ia
E
m
ail: v
ijay
alax
m
is
a@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Ag
r
icu
ltu
r
e
is
a
co
r
n
er
s
to
n
e
o
f
g
lo
b
al
f
o
o
d
s
ec
u
r
ity
,
y
et
p
lan
t
d
is
ea
s
es
p
o
s
e
a
m
ajo
r
th
r
ea
t
to
s
u
s
tain
ab
le
cr
o
p
p
r
o
d
u
ctio
n
.
Acc
o
r
d
in
g
to
th
e
Fo
o
d
an
d
Ag
r
icu
ltu
r
e
Or
g
an
izatio
n
(
FAO)
,
n
ea
r
ly
4
0
%
o
f
an
n
u
al
cr
o
p
y
ield
s
ar
e
lo
s
t
to
p
ests
an
d
p
ath
o
g
en
s
,
r
esu
ltin
g
in
s
ig
n
if
ican
t
ec
o
n
o
m
ic
an
d
f
o
o
d
s
u
p
p
ly
ch
allen
g
es
[
1
]
.
E
a
r
ly
an
d
ac
c
u
r
ate
d
etec
tio
n
o
f
p
lan
t
d
is
ea
s
es
is
th
er
ef
o
r
e
ess
en
tial
f
o
r
r
ed
u
cin
g
cr
o
p
lo
s
s
es
an
d
en
a
b
lin
g
tim
ely
i
n
ter
v
en
t
io
n
s
.
T
r
ad
itio
n
al
d
iag
n
o
s
is
m
eth
o
d
s
,
s
u
ch
as
v
is
u
al
in
s
p
ec
t
io
n
b
y
ex
p
e
r
ts
,
ar
e
o
f
ten
tim
e
-
co
n
s
u
m
in
g
,
s
u
b
ject
iv
e,
an
d
u
n
s
u
itab
le
f
o
r
lar
g
e
-
s
ca
le
f
ar
m
in
g
o
p
er
atio
n
s
[
2
]
.
R
ec
en
t
ad
v
an
ce
s
in
ar
tific
ial
i
n
tellig
en
ce
(
AI
)
an
d
co
m
p
u
ter
v
is
io
n
h
av
e
en
ab
led
au
to
m
ate
d
s
y
s
tem
s
to
id
en
tify
p
lan
t
d
is
ea
s
es
d
ir
ec
tly
f
r
o
m
leaf
im
ag
es.
Dee
p
l
ea
r
n
in
g
m
o
d
els,
p
ar
ticu
lar
ly
c
o
n
v
o
lu
ti
o
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NNs),
h
av
e
d
e
m
o
n
s
tr
ated
s
tr
o
n
g
ca
p
ab
ilit
ies
in
ca
p
tu
r
in
g
h
ier
a
r
ch
ical
im
ag
e
f
ea
tu
r
es
with
o
u
t
h
an
d
cr
a
f
ted
d
esig
n
[
3
]
.
Ar
ch
itectu
r
es
s
u
ch
as
A
lex
Net,
V
GG,
an
d
R
esNet
h
av
e
b
ee
n
wid
ely
ad
o
p
ted
in
ag
r
icu
ltu
r
al
a
p
p
licatio
n
s
,
ac
h
iev
in
g
p
r
o
m
is
in
g
r
esu
lts
o
n
b
en
ch
m
ar
k
d
atasets
lik
e
Pla
n
tVillag
e
[
4
]
,
[
5
]
.
Desp
ite
th
eir
s
u
cc
ess
,
th
e
s
e
m
o
d
els ty
p
ically
r
eq
u
ir
e
lar
g
e
tr
ain
in
g
d
atasets
an
d
h
ig
h
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
,
wh
ich
lim
its
th
eir
d
ep
lo
y
m
en
t
in
r
eso
u
r
ce
-
c
o
n
s
tr
ain
ed
e
n
v
ir
o
n
m
en
ts
[
6
]
.
T
h
e
em
er
g
en
ce
o
f
v
is
io
n
tr
an
s
f
o
r
m
er
s
(
ViT
s
)
h
as
i
n
tr
o
d
u
ce
d
a
n
ew
p
a
r
ad
ig
m
f
o
r
im
a
g
e
-
b
a
s
ed
task
s
,
lev
er
ag
in
g
s
elf
-
atten
tio
n
m
ec
h
an
is
m
s
to
ca
p
t
u
r
e
g
lo
b
al
d
ep
en
d
en
cies
m
o
r
e
ef
f
ec
tiv
ely
th
an
co
n
v
o
lu
tio
n
-
b
ased
m
o
d
els
[
7
]
.
ViT
s
h
a
v
e
s
h
o
wn
s
tate
-
of
-
t
h
e
-
ar
t
p
er
f
o
r
m
an
ce
ac
r
o
s
s
v
ar
i
o
u
s
co
m
p
u
t
er
v
is
io
n
d
o
m
ain
s
,
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.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
3
9
9
-
1
4
0
6
1400
in
clu
d
in
g
m
ed
ical
im
a
g
in
g
an
d
r
em
o
te
s
en
s
in
g
[
8
]
.
Ho
wev
er
,
th
eir
h
ig
h
-
d
im
en
s
io
n
al
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
ca
n
lead
to
in
cr
ea
s
ed
co
m
p
u
tatio
n
al
d
em
an
d
s
an
d
o
v
er
f
itti
n
g
r
is
k
s
wh
en
ap
p
lied
to
r
elativ
ely
s
m
all
ag
r
icu
ltu
r
al
d
atasets
.
T
o
o
v
er
co
m
e
th
ese
ch
allen
g
es,
h
y
b
r
i
d
ap
p
r
o
ac
h
es
th
at
co
m
b
in
e
d
ee
p
f
ea
tu
r
e
ex
tr
ac
tio
n
with
class
ical
m
ac
h
in
e
lear
n
in
g
class
if
ier
s
h
av
e
g
ain
ed
atten
tio
n
.
Prin
cip
al
co
m
p
o
n
en
t
an
aly
s
is
(
P
C
A)
h
as
b
ee
n
wid
ely
em
p
lo
y
e
d
to
ad
d
r
ess
th
e
cu
r
s
e
o
f
d
im
en
s
io
n
ality
b
y
r
ed
u
ci
n
g
r
e
d
u
n
d
an
t
f
ea
tu
r
e
s
wh
ile
p
r
eser
v
in
g
m
o
s
t
o
f
th
e
v
ar
ian
ce
[
9
]
.
Me
an
wh
ile,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
es
(
SVMs)
r
em
ain
p
o
wer
f
u
l
class
if
ier
s
f
o
r
h
ig
h
-
d
im
en
s
io
n
al
d
ata
d
u
e
to
th
eir
r
o
b
u
s
tn
ess
an
d
g
en
e
r
aliza
tio
n
a
b
ilit
y
[
1
0
]
.
I
n
teg
r
atin
g
ViT
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
with
PC
A
f
o
r
d
im
en
s
io
n
ality
r
ed
u
ctio
n
an
d
SVM
f
o
r
class
i
f
icatio
n
o
f
f
er
s
a
b
alan
ce
d
s
o
lu
tio
n
th
at
co
m
b
in
es
th
e
s
tr
en
g
th
s
o
f
d
ee
p
lear
n
in
g
an
d
tr
ad
itio
n
al
m
ac
h
in
e
lear
n
i
n
g
.
T
h
is
s
tu
d
y
p
r
o
p
o
s
es
a
h
y
b
r
i
d
tr
an
s
f
o
r
m
er
-
b
ased
class
if
icatio
n
f
r
am
ewo
r
k
f
o
r
p
lan
t
leaf
d
is
ea
s
e
d
etec
tio
n
,
wh
er
e
a
p
r
etr
ain
e
d
ViT
is
u
s
ed
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
,
f
o
llo
wed
b
y
PC
A
f
o
r
d
im
en
s
io
n
ality
r
ed
u
ctio
n
an
d
a
tu
n
ed
SVM
f
o
r
class
if
icatio
n
.
E
x
p
er
im
en
ta
l
r
esu
lts
o
n
th
e
B
ea
n
L
ea
f
L
esio
n
s
d
ataset
s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
ViT
–
PC
A
–
SVM
p
ip
elin
e
ac
h
iev
es
9
2
%
ac
cu
r
ac
y
an
d
a
weig
h
ted
F1
-
s
co
r
e
o
f
0
.
9
2
,
o
u
tp
er
f
o
r
m
in
g
co
n
v
en
tio
n
al
C
NN
-
b
ased
b
aselin
e
m
o
d
els.
T
h
e
co
n
tr
ib
u
tio
n
s
o
f
o
u
r
wo
r
k
a
r
e
th
r
ee
f
o
l
d
:
a.
A
h
y
b
r
i
d
p
lan
t
d
is
ea
s
e
d
etec
tio
n
f
r
am
ew
o
r
k
t
h
at
in
teg
r
a
tes
ViT
f
ea
tu
r
e
ex
tr
ac
tio
n
w
ith
PC
A
-
b
ased
d
im
en
s
io
n
ality
r
ed
u
ctio
n
a
n
d
SVM
class
if
icatio
n
.
,
b.
An
ef
f
icien
t
f
ea
tu
r
e
o
p
tim
izati
o
n
s
tr
ateg
y
u
s
in
g
p
r
i
n
cip
al
co
m
p
o
n
e
n
t
an
aly
s
is
to
r
ed
u
ce
h
i
g
h
-
d
im
e
n
s
io
n
al
tr
an
s
f
o
r
m
er
f
ea
tu
r
es wh
ile
p
r
e
s
er
v
in
g
im
p
o
r
tan
t d
is
cr
im
in
ati
v
e
in
f
o
r
m
atio
n
,
a
n
d
c.
E
x
ten
s
iv
e
ex
p
er
im
en
tal
ev
alu
atio
n
o
n
th
e
b
ea
n
lea
f
lesi
o
n
s
d
ataset
,
d
em
o
n
s
tr
atin
g
im
p
r
o
v
ed
class
if
icatio
n
p
er
f
o
r
m
an
ce
co
m
p
ar
e
d
with
c
o
n
v
en
tio
n
al
C
NN
-
b
ased
ap
p
r
o
ac
h
es.
Desp
ite
s
ig
n
if
ican
t
p
r
o
g
r
ess
in
p
lan
t
d
is
ea
s
e
d
etec
tio
n
u
s
in
g
d
ee
p
lear
n
in
g
,
m
o
s
t
ex
is
tin
g
a
p
p
r
o
ac
h
es
r
ely
h
ea
v
ily
o
n
C
NN
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
class
if
icatio
n
.
Ho
wev
er
,
C
NNs
p
r
im
ar
ily
ca
p
tu
r
e
lo
ca
l
s
p
atial
f
ea
tu
r
es
an
d
m
ay
f
ail
to
m
o
d
e
l
lo
n
g
-
r
an
g
e
d
ep
en
d
en
cies
in
co
m
p
lex
leaf
p
atter
n
s
.
R
ec
en
t
tr
an
s
f
o
r
m
er
-
b
ased
m
o
d
els
ad
d
r
ess
th
is
lim
itatio
n
b
u
t
o
f
ten
in
v
o
lv
e
h
ig
h
co
m
p
u
tatio
n
al
co
s
t
an
d
lack
in
teg
r
atio
n
with
ef
f
icien
t
class
ical
clas
s
if
ier
s
.
Mo
r
eo
v
er
,
lim
ited
r
esear
ch
e
x
p
lo
r
es
th
e
c
o
m
b
in
atio
n
o
f
tr
a
n
s
f
o
r
m
er
-
b
ased
f
ea
tu
r
e
ex
tr
ac
tio
n
with
d
im
en
s
io
n
alit
y
r
ed
u
ctio
n
tech
n
iq
u
es
an
d
tr
a
d
itio
n
al
m
ac
h
in
e
lear
n
i
n
g
alg
o
r
ith
m
s
.
T
o
ad
d
r
ess
th
ese
lim
itatio
n
s
,
th
is
s
tu
d
y
p
r
o
p
o
s
es
a
h
y
b
r
i
d
f
r
am
ewo
r
k
th
at
in
teg
r
ates
ViT
,
PC
A
,
an
d
SVM
to
ac
h
ie
v
e
im
p
r
o
v
e
d
ac
cu
r
ac
y
wh
ile
r
ed
u
cin
g
co
m
p
u
tatio
n
al
co
m
p
lex
it
y
.
2.
RE
L
AT
E
D
WO
RK
Au
to
m
ated
p
lan
t
d
is
ea
s
e
d
etec
tio
n
h
as
b
ee
n
ex
ten
s
iv
ely
s
tu
d
ied
u
s
in
g
b
o
th
class
ical
m
ac
h
in
e
lear
n
in
g
an
d
d
ee
p
lear
n
in
g
tech
n
iq
u
es.
E
ar
ly
ap
p
r
o
ac
h
es
r
elied
o
n
h
an
d
c
r
af
ted
f
ea
tu
r
es
s
u
ch
as
tex
tu
r
e
d
escr
ip
to
r
s
,
co
l
o
r
h
is
to
g
r
am
s
,
an
d
s
h
ap
e
m
ea
s
u
r
es,
c
o
m
b
i
n
e
d
with
class
if
ier
s
lik
e
SVM,
k
-
n
ea
r
est
n
eig
h
b
o
r
s
(k
-
NN)
,
an
d
r
an
d
o
m
f
o
r
ests
[
1
1
]
.
W
h
ile
th
ese
m
eth
o
d
s
o
f
f
er
ed
s
o
m
e
s
u
cc
ess
in
co
n
tr
o
lled
en
v
ir
o
n
m
en
ts
,
th
ey
wer
e
lim
ited
b
y
th
eir
d
e
p
en
d
en
cy
o
n
e
x
p
er
t
-
d
r
iv
en
f
ea
tu
r
e
en
g
in
ee
r
in
g
an
d
th
ei
r
s
en
s
itiv
ity
to
illu
m
in
atio
n
,
n
o
is
e,
an
d
b
ac
k
g
r
o
u
n
d
v
a
r
iatio
n
s
[
1
2
]
.
T
h
e
in
tr
o
d
u
ctio
n
o
f
d
ee
p
lea
r
n
in
g
,
p
ar
ticu
lar
l
y
C
NN
,
m
ar
k
ed
a
s
ig
n
if
ican
t
a
d
v
an
ce
m
e
n
t
in
p
lan
t
d
is
ea
s
e
r
ec
o
g
n
itio
n
.
Mo
h
an
ty
et
a
l.
[
2
]
d
em
o
n
s
tr
ated
th
e
ef
f
ec
tiv
en
ess
o
f
d
ee
p
C
NNs
o
n
th
e
Plan
tVillag
e
d
ataset,
ac
h
iev
in
g
s
u
p
e
r
io
r
p
e
r
f
o
r
m
a
n
ce
co
m
p
ar
ed
t
o
tr
ad
iti
o
n
al
m
eth
o
d
s
.
T
r
an
s
f
er
lear
n
i
n
g
f
u
r
th
er
im
p
r
o
v
ed
class
if
icatio
n
ac
cu
r
ac
y
b
y
lev
er
ag
in
g
p
r
etr
ain
ed
m
o
d
els
s
u
ch
as
VGG,
R
e
s
Net,
an
d
I
n
ce
p
tio
n
[
1
3
]
,
[
6
]
.
Fer
en
tin
o
s
[
1
4
]
r
e
p
o
r
ted
cla
s
s
if
icatio
n
ac
cu
r
ac
ies
ex
ce
ed
in
g
9
9
%
u
s
in
g
C
NNs
ac
r
o
s
s
m
u
ltip
le
cr
o
p
s
,
v
alid
atin
g
th
e
s
ca
lab
ilit
y
o
f
d
ee
p
lear
n
in
g
.
Ho
wev
er
,
th
e
s
e
m
eth
o
d
s
ar
e
co
m
p
u
tatio
n
ally
ex
p
en
s
iv
e
an
d
r
eq
u
ir
e
lar
g
e
lab
eled
d
atasets
,
lim
itin
g
th
eir
ap
p
licab
ilit
y
in
r
eso
u
r
ce
-
co
n
s
tr
ain
e
d
ag
r
ic
u
ltu
r
al
s
ettin
g
s
.
T
o
ad
d
r
ess
th
ese
ch
allen
g
es,
r
esear
ch
er
s
h
av
e
ex
p
lo
r
ed
h
y
b
r
id
ap
p
r
o
ac
h
es
th
at
co
m
b
i
n
e
d
e
ep
f
ea
tu
r
e
ex
tr
ac
tio
n
with
class
ical
m
ac
h
in
e
lear
n
in
g
class
if
ier
s
.
Fo
r
in
s
tan
ce
,
C
NNs
h
av
e
b
e
en
u
s
ed
as
f
ea
t
u
r
e
ex
tr
ac
to
r
s
,
with
th
e
ex
tr
ac
te
d
em
b
ed
d
i
n
g
s
class
if
ied
u
s
in
g
SVMs
o
r
R
an
d
o
m
Fo
r
e
s
ts
,
o
f
ten
y
ield
in
g
im
p
r
o
v
e
d
g
en
er
aliza
tio
n
co
m
p
ar
ed
to
en
d
-
to
-
en
d
C
NN
class
i
f
ier
s
[
1
5
]
.
Hy
b
r
i
d
m
o
d
els h
av
e
also
b
ee
n
ap
p
lied
in
o
th
er
d
o
m
ain
s
,
s
u
ch
as
m
e
d
ical
im
ag
in
g
,
wh
er
e
co
m
b
in
i
n
g
d
e
ep
f
ea
tu
r
es
with
SVMs
en
h
an
ce
s
r
o
b
u
s
tn
ess
o
n
s
m
all
d
atasets
[
1
6
]
.
M
o
r
e
r
ec
en
tly
,
ViT
s
h
av
e
e
m
er
g
ed
as
p
o
wer
f
u
l
f
ea
t
u
r
e
ex
tr
ac
t
o
r
s
d
u
e
to
t
h
eir
ab
ilit
y
to
ca
p
tu
r
e
g
lo
b
al
d
ep
e
n
d
en
cies
v
ia
s
elf
-
atten
tio
n
[
7
]
.
Stu
d
ies
h
av
e
s
h
o
wn
th
at
ViT
f
ea
tu
r
es,
wh
en
p
air
ed
with
class
if
ier
s
lik
e
SV
M
o
r
k
-
NN,
ac
h
iev
e
co
m
p
etitiv
e
p
er
f
o
r
m
a
n
ce
in
im
ag
e
class
if
icatio
n
task
s
[
1
7
]
.
Desp
ite
th
ese
ad
v
an
ce
s
,
th
er
e
is
lim
ited
r
e
s
ea
r
ch
o
n
lev
er
a
g
in
g
ViT
-
b
ased
f
ea
tu
r
e
ex
t
r
a
ctio
n
with
d
im
en
s
io
n
ality
r
e
d
u
ctio
n
an
d
class
ical
clas
s
if
ier
s
f
o
r
ag
r
icu
ltu
r
al
d
atasets
.
Mo
s
t
s
tu
d
ies
eith
er
r
ely
s
o
lely
o
n
C
NN
-
b
ased
en
d
-
to
-
en
d
m
o
d
e
ls
o
r
u
s
e
ViT
with
o
u
t
o
p
tim
izin
g
th
e
f
ea
tu
r
e
s
p
ac
e
f
o
r
d
o
wn
s
tr
ea
m
lear
n
in
g
.
T
h
is
g
ap
m
o
tiv
ates
th
e
p
r
esen
t
wo
r
k
,
wh
ich
in
teg
r
ates
a
p
r
etr
ain
ed
ViT
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
,
PC
A
f
o
r
d
im
en
s
io
n
ality
r
ed
u
ctio
n
,
an
d
a
tu
n
ed
SVM
f
o
r
class
if
icatio
n
.
B
y
co
m
b
in
in
g
th
e
s
tr
en
g
th
s
o
f
d
ee
p
an
d
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
Tr
a
n
s
fo
r
mer
-
b
a
s
ed
h
yb
r
id
cla
s
s
ifica
tio
n
fo
r
p
la
n
t le
a
f d
is
ea
s
e
d
etec
tio
n
…
(
V
ija
ya
la
ksh
mi
S
.
A
b
b
ig
eri
)
1401
class
ical
lear
n
in
g
,
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
m
eta
-
lear
n
in
g
f
r
a
m
ewo
r
k
ad
d
r
ess
es
b
o
th
ef
f
ic
ien
cy
an
d
ac
c
u
r
ac
y
,
m
ak
in
g
it su
itab
le
f
o
r
r
ea
l
-
wo
r
ld
s
m
ar
t f
ar
m
in
g
ap
p
licatio
n
s
.
R
ec
en
t
s
tu
d
ies
h
av
e
also
ex
p
l
o
r
ed
in
tellig
en
t
p
r
ed
ictiv
e
m
o
d
elin
g
f
r
am
ewo
r
k
s
an
d
h
y
b
r
id
ar
tific
ial
in
tellig
en
ce
tech
n
iq
u
es
f
o
r
co
m
p
lex
d
ec
is
io
n
-
m
a
k
in
g
s
y
s
tem
s
.
T
h
ese
wo
r
k
s
d
em
o
n
s
tr
ate
th
e
g
r
o
win
g
p
o
ten
tial
o
f
in
teg
r
atin
g
m
ac
h
i
n
e
lear
n
in
g
m
o
d
els
with
o
p
ti
m
izatio
n
an
d
in
tellig
en
t
co
m
p
u
tin
g
s
tr
ateg
ies
to
im
p
r
o
v
e
p
r
ed
ictiv
e
p
er
f
o
r
m
a
n
ce
in
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
.
I
n
s
p
ir
ed
b
y
th
ese
ap
p
r
o
ac
h
es,
th
is
s
tu
d
y
in
v
esti
g
ates
a
h
y
b
r
id
f
r
am
ew
o
r
k
co
m
b
in
in
g
tr
a
n
s
f
o
r
m
er
-
b
ased
d
ee
p
r
ep
r
esen
tatio
n
s
with
class
ical
m
ac
h
in
e
lear
n
in
g
class
if
icatio
n
f
o
r
p
lan
t d
is
ea
s
e
d
etec
tio
n
.
3.
M
E
T
H
O
D
T
h
e
p
r
o
p
o
s
ed
f
r
a
m
ewo
r
k
c
o
m
b
in
es
th
e
s
tr
en
g
th
s
o
f
d
ee
p
lear
n
in
g
f
ea
tu
r
e
ex
tr
ac
tio
n
a
n
d
class
ical
m
ac
h
in
e
lear
n
in
g
class
if
icatio
n
.
Sp
ec
if
ically
,
a
p
r
etr
ain
ed
V
iT
is
em
p
lo
y
ed
f
o
r
h
i
g
h
-
lev
el
f
ea
tu
r
e
ex
tr
ac
tio
n
,
f
o
llo
wed
b
y
PC
A
f
o
r
d
im
e
n
s
io
n
ality
r
ed
u
ctio
n
,
a
n
d
a
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
(
SV
C
)
f
o
r
f
in
al
d
is
ea
s
e
class
if
icatio
n
.
3
.
1
.
Da
t
a
s
et
T
h
e
ex
p
e
r
im
en
ts
wer
e
co
n
d
u
cted
o
n
t
h
e
b
ea
n
leaf
lesi
o
n
s
d
ataset
av
ailab
le
o
n
Kag
g
le
[
1
8
]
,
wh
ic
h
co
n
tain
s
th
r
ee
class
es:
an
g
u
l
ar
leaf
s
p
o
t,
b
ea
n
r
u
s
t,
a
n
d
h
ea
lth
y
leav
es.
T
h
e
d
ataset
was
p
ar
titi
o
n
ed
in
to
tr
ain
in
g
(
9
7
4
im
a
g
es),
v
alid
a
tio
n
(
1
3
3
im
ag
es),
a
n
d
test
in
g
(
6
0
im
a
g
es)
s
ets.
T
h
is
s
p
li
t
en
s
u
r
es
s
u
f
f
icien
t
s
am
p
les
f
o
r
m
o
d
el
tr
ain
in
g
,
h
y
p
er
p
ar
am
eter
tu
n
in
g
,
an
d
u
n
b
i
ased
ev
alu
atio
n
.
T
o
im
p
r
o
v
e
t
h
e
r
eliab
ilit
y
o
f
th
e
ev
alu
atio
n
an
d
r
e
d
u
ce
p
o
ten
ti
al
b
ias
ca
u
s
ed
b
y
lim
ited
s
am
p
les,
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es
wer
e
ap
p
lied
d
u
r
in
g
tr
ai
n
in
g
.
Fu
r
t
h
er
m
o
r
e,
th
e
d
ataset
s
p
lit
wa
s
ca
r
ef
u
lly
d
esig
n
ed
to
m
ain
tain
cl
ass
b
alan
ce
ac
r
o
s
s
tr
ain
in
g
,
v
alid
atio
n
,
an
d
test
in
g
s
ets.
Alth
o
u
g
h
th
e
d
ataset
s
ize
is
r
elativ
ely
s
m
all
co
m
p
ar
ed
to
lar
g
e
-
s
ca
le
v
is
io
n
d
atasets
,
it
is
wid
ely
u
s
ed
in
p
lan
t
d
is
ea
s
e
d
etec
tio
n
r
esear
ch
an
d
p
r
o
v
id
es
a
r
ea
li
s
tic
b
en
ch
m
ar
k
f
o
r
ev
alu
atin
g
m
o
d
el
g
e
n
er
aliza
tio
n
.
3
.
2
.
P
re
pro
ce
s
s
ing
E
ac
h
im
ag
e
was
r
esized
to
2
2
4
×2
2
4
p
ix
els
to
m
atch
th
e
in
p
u
t
s
ize
o
f
th
e
ViT
m
o
d
el.
Pix
el
v
alu
es
wer
e
n
o
r
m
alize
d
to
t
h
e
r
an
g
e
[
0
,
1
]
t
o
s
tab
ilize
tr
ain
in
g
.
Data
au
g
m
en
tatio
n
tech
n
iq
u
e
s
,
in
clu
d
in
g
r
a
n
d
o
m
f
lip
s
,
b
r
ig
h
tn
ess
ad
ju
s
tm
en
ts
,
co
n
tr
ast
v
ar
iatio
n
s
,
an
d
s
atu
r
a
tio
n
ch
an
g
es,
wer
e
ap
p
lied
to
im
p
r
o
v
e
r
o
b
u
s
tn
ess
ag
ain
s
t e
n
v
ir
o
n
m
en
tal
v
ar
iatio
n
s
s
u
ch
as illu
m
in
atio
n
an
d
o
r
i
en
tatio
n
.
3
.
3
.
ViT
f
o
r
f
e
a
t
ure
ex
t
r
a
ct
io
n
A
p
r
etr
ain
ed
ViT
-
B
1
6
m
o
d
el
was
u
s
ed
as
th
e
b
ac
k
b
o
n
e
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
.
Un
lik
e
C
NNs,
ViT
s
o
p
er
ate
b
y
d
iv
i
d
in
g
an
im
ag
e
in
to
p
atch
es
an
d
ap
p
ly
in
g
a
tr
an
s
f
o
r
m
er
en
c
o
d
er
to
ca
p
tu
r
e
g
lo
b
al
d
ep
en
d
e
n
cies.
T
h
e
p
r
etr
ain
e
d
weig
h
ts
f
r
o
m
I
m
ag
eNe
t
p
r
o
v
id
ed
tr
an
s
f
er
a
b
le
r
ep
r
esen
tat
io
n
s
,
s
ig
n
if
ican
tly
r
ed
u
cin
g
tr
ain
in
g
tim
e.
A
d
e
n
s
e
f
ea
tu
r
e
lay
er
was
a
d
d
ed
,
p
r
o
d
u
cin
g
6
4
-
d
im
e
n
s
io
n
al
e
m
b
ed
d
in
g
s
f
o
r
ea
ch
im
ag
e.
T
h
ese
em
b
e
d
d
in
g
s
f
o
r
m
th
e
in
p
u
t f
o
r
th
e
d
im
e
n
s
io
n
ality
r
ed
u
ctio
n
s
tag
e.
3
.
4
.
P
CA
T
o
ad
d
r
ess
th
e
cu
r
s
e
o
f
d
im
e
n
s
io
n
ality
an
d
m
in
im
ize
c
o
m
p
u
tatio
n
al
o
v
e
r
h
ea
d
,
PC
A
wa
s
ap
p
lied
to
th
e
ex
tr
ac
ted
f
ea
t
u
r
es.
T
h
e
n
u
m
b
er
o
f
co
m
p
o
n
en
ts
was
ch
o
s
en
to
r
etain
9
8
%
v
ar
ian
ce
,
r
ed
u
cin
g
t
h
e
f
ea
tu
r
e
v
ec
to
r
f
r
o
m
6
4
d
im
en
s
io
n
s
to
4
1
d
im
e
n
s
io
n
s
.
T
h
is
s
tep
n
o
t
o
n
ly
d
ec
r
ea
s
es
tr
ain
in
g
tim
e
b
u
t
also
r
ed
u
ce
s
th
e
r
is
k
o
f
o
v
er
f
itti
n
g
wh
ile
p
r
eser
v
in
g
d
is
cr
im
in
ativ
e
i
n
f
o
r
m
atio
n
.
3
.
5
.
SVM
cla
s
s
if
ier
T
h
e
r
ed
u
ce
d
f
ea
tu
r
es
wer
e
cl
ass
if
ied
u
s
in
g
an
SVM
with
an
R
B
F
k
er
n
el,
o
p
tim
ized
with
Op
tu
n
a
-
b
ased
h
y
p
e
r
p
ar
am
ete
r
tu
n
in
g
.
SVM
was
ch
o
s
en
d
u
e
to
its
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
in
h
ig
h
-
d
im
en
s
io
n
al
f
ea
tu
r
e
s
p
ac
es
an
d
ab
ilit
y
to
h
an
d
le
s
m
all
d
atasets
ef
f
ec
tiv
ely
.
T
h
e
m
o
d
el
o
u
t
p
u
ts
class
lab
els
c
o
r
r
esp
o
n
d
in
g
to
t
h
e
leaf
d
is
ea
s
e
ca
teg
o
r
ies.
3.
6
.
E
nd
-
to
-
end pip
eline
T
h
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
in
t
eg
r
ates
m
u
ltip
le
s
tag
es
in
to
a
s
ea
m
less
p
ip
elin
e
f
o
r
p
lan
t
d
is
ea
s
e
d
etec
tio
n
.
E
ac
h
s
tag
e
co
n
tr
ib
u
tes
to
im
p
r
o
v
in
g
m
o
d
el
r
o
b
u
s
tn
ess
,
co
m
p
u
tatio
n
al
ef
f
icien
cy
,
an
d
class
if
icatio
n
ac
cu
r
ac
y
.
T
h
e
o
v
e
r
all
m
eth
o
d
o
lo
g
y
is
s
u
m
m
a
r
ized
i
n
Fig
u
r
e
1
.
T
h
e
SVM
o
u
tp
u
ts
th
e
f
in
al
d
is
ea
s
e
lab
el:
an
g
u
lar
leaf
s
p
o
t,
b
ea
n
r
u
s
t,
o
r
h
ea
lth
y
.
T
h
ese
p
r
ed
ictio
n
s
c
an
b
e
in
teg
r
ated
i
n
to
d
ec
is
io
n
s
u
p
p
o
r
t
s
y
s
tem
s
f
o
r
f
ar
m
er
s
,
en
a
b
lin
g
tim
ely
d
is
ea
s
e
m
an
ag
em
en
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.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
3
9
9
-
1
4
0
6
1402
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
ex
p
e
r
im
en
tal
ev
al
u
atio
n
o
f
th
e
p
r
o
p
o
s
ed
h
y
b
r
i
d
f
r
a
m
ewo
r
k
was
co
n
d
u
cted
o
n
th
e
b
ea
n
leaf
d
is
ea
s
e
d
ataset
u
s
in
g
s
tr
u
ctu
r
ed
tr
ain
in
g
,
v
alid
ati
o
n
,
a
n
d
test
in
g
p
r
o
to
co
l.
T
h
is
s
ec
tio
n
p
r
esen
ts
a
co
m
p
r
eh
e
n
s
iv
e
a
n
aly
s
is
o
f
th
e
m
o
d
el’
s
p
e
r
f
o
r
m
an
ce
th
r
o
u
g
h
b
o
th
q
u
an
titativ
e
m
etr
i
cs
an
d
q
u
alitativ
e
in
s
ig
h
ts
.
Key
asp
ec
ts
,
in
cl
u
d
i
n
g
o
v
er
all
class
if
icatio
n
ac
c
u
r
ac
y
,
lear
n
in
g
b
eh
a
v
io
r
,
co
n
f
u
s
io
n
m
atr
i
x
an
al
y
s
is
,
an
d
m
is
class
if
icatio
n
p
atter
n
s
,
ar
e
d
is
cu
s
s
ed
in
d
etail
t
o
p
r
o
v
id
e
a
d
ee
p
er
u
n
d
er
s
t
an
d
in
g
o
f
m
o
d
el
p
er
f
o
r
m
an
ce
.
Fu
r
th
er
m
o
r
e,
a
co
m
p
ar
ativ
e
ev
alu
atio
n
wi
th
co
n
v
en
tio
n
al
C
NN
m
o
d
els
an
d
p
r
etr
ain
ed
ar
ch
itectu
r
es
d
em
o
n
s
tr
ates
th
e
ef
f
ec
tiv
en
ess
o
f
th
e
p
r
o
p
o
s
ed
ViT
–
PC
A
–
SVM
ap
p
r
o
ac
h
.
T
h
e
r
esu
lts
in
d
icate
th
at
th
e
in
teg
r
atio
n
o
f
tr
an
s
f
o
r
m
er
-
b
ased
f
ea
tu
r
e
e
x
tr
ac
tio
n
with
d
im
en
s
io
n
ality
r
e
d
u
ctio
n
an
d
class
ical
class
if
icatio
n
im
p
r
o
v
es
b
o
th
ac
cu
r
ac
y
an
d
co
m
p
u
tatio
n
al
ef
f
icien
cy
.
I
n
ad
d
itio
n
,
th
e
p
r
ac
ti
ca
l
ap
p
licab
ilit
y
o
f
th
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
in
r
ea
l
-
wo
r
ld
a
g
r
icu
ltu
r
al
s
ce
n
a
r
io
s
is
d
is
cu
s
s
ed
,
h
ig
h
lig
h
tin
g
its
p
o
ten
tial
f
o
r
d
ep
lo
y
m
e
n
t in
s
m
ar
t f
a
r
m
in
g
s
y
s
tem
s
f
o
r
ea
r
ly
a
n
d
r
eliab
le
p
lan
t d
is
ea
s
e
d
etec
tio
n
.
4
.
1
.
E
x
perim
ent
a
l
s
et
up
T
h
e
p
r
o
p
o
s
ed
h
y
b
r
id
p
ip
elin
e
was
ev
alu
ated
o
n
a
b
ea
n
leaf
d
is
ea
s
e
d
ataset
co
m
p
r
is
in
g
th
r
ee
class
es
:
an
g
u
lar
leaf
s
p
o
t,
b
ea
n
r
u
s
t,
an
d
h
ea
lth
y
.
T
h
e
d
ataset
was p
ar
titi
o
n
ed
in
to
9
7
4
im
ag
es f
o
r
t
r
ain
in
g
,
1
3
3
im
ag
es
f
o
r
v
alid
atio
n
,
an
d
6
0
im
ag
es
f
o
r
test
in
g
.
All
ex
p
er
im
e
n
ts
wer
e
co
n
d
u
cted
u
s
in
g
T
en
s
o
r
Flo
w
with
a
d
is
tr
ib
u
ted
tr
ain
in
g
s
tr
ateg
y
(
Mir
r
o
r
ed
Stra
teg
y
)
o
n
a
GPU
-
en
ab
led
en
v
ir
o
n
m
en
t.
I
m
a
g
es
wer
e
r
esized
to
2
2
4
×2
2
4
p
i
x
els,
n
o
r
m
alize
d
,
an
d
au
g
m
en
ted
th
r
o
u
g
h
r
a
n
d
o
m
f
lip
s
,
c
o
n
tr
ast,
b
r
ig
h
tn
ess
,
an
d
s
atu
r
atio
n
ad
ju
s
tm
en
ts
to
im
p
r
o
v
e
m
o
d
el
r
o
b
u
s
tn
ess
.
Fig
u
r
e
1
.
Pro
p
o
s
ed
s
y
s
tem
ar
c
h
itectu
r
e
f
o
r
leaf
d
is
ea
s
e
class
i
f
icatio
n
u
s
in
g
ViT
-
PC
A
-
SVM
h
y
b
r
i
d
ap
p
r
o
ac
h
4
.
2
.
F
e
a
t
ure
ex
t
r
a
ct
io
n a
nd
dim
ens
io
na
lity
re
du
ct
io
n
T
h
e
v
is
io
n
tr
an
s
f
o
r
m
e
r
(
ViT
-
B
1
6
)
was
em
p
lo
y
ed
as
a
p
r
etr
ain
ed
f
ea
tu
r
e
ex
tr
ac
t
o
r
.
I
t
g
en
er
ated
64
-
d
im
e
n
s
io
n
al
f
ea
tu
r
e
v
ec
t
o
r
s
f
o
r
ea
c
h
im
ag
e
.
T
o
m
i
tig
ate
th
e
cu
r
s
e
o
f
d
im
e
n
s
io
n
ality
an
d
r
ed
u
ce
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
Tr
a
n
s
fo
r
mer
-
b
a
s
ed
h
yb
r
id
cla
s
s
ifica
tio
n
fo
r
p
la
n
t le
a
f d
is
ea
s
e
d
etec
tio
n
…
(
V
ija
ya
la
ksh
mi
S
.
A
b
b
ig
eri
)
1403
co
m
p
u
tatio
n
al
co
s
t,
PC
A
was
ap
p
lied
,
c
o
m
p
r
ess
in
g
th
e
f
ea
tu
r
e
s
p
ac
e
t
o
4
1
p
r
in
cip
al
co
m
p
o
n
en
ts
wh
ile
p
r
eser
v
in
g
9
8
%
o
f
th
e
o
r
ig
i
n
al
v
ar
ian
ce
.
T
h
is
r
ed
u
ctio
n
ac
ce
ler
ated
th
e
s
u
b
s
eq
u
e
n
t
class
if
icatio
n
s
tag
e
with
o
u
t sig
n
if
ican
t p
er
f
o
r
m
a
n
ce
lo
s
s
.
4
.
3
.
Cla
s
s
if
ica
t
io
n
perf
o
r
m
a
nce
A
s
u
p
p
o
r
t
v
ec
to
r
class
if
ier
with
an
R
B
F
k
er
n
el,
o
p
tim
ized
u
s
in
g
th
e
Op
tu
n
a
f
r
am
ewo
r
k
,
was
tr
ain
ed
o
n
th
e
PC
A
-
tr
an
s
f
o
r
m
ed
f
ea
tu
r
es.
T
h
e
p
r
o
p
o
s
ed
h
y
b
r
id
ap
p
r
o
ac
h
ac
h
ie
v
ed
s
tr
o
n
g
class
if
icatio
n
r
esu
lts
o
n
t
h
e
test
s
et:
−
Acc
u
r
ac
y
: 9
2
.
0
%
−
W
eig
h
ted
F1
-
Sco
r
e:
0
.
9
2
−
Me
an
s
q
u
ar
ed
e
r
r
o
r
(
MSE
)
: 0
.
0
4
5
T
h
ese
r
esu
lts
co
n
f
ir
m
th
at
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
p
ip
elin
e
s
u
c
ce
s
s
f
u
lly
in
teg
r
ates
d
ee
p
f
ea
t
u
r
e
ex
tr
ac
tio
n
with
class
ical
m
ac
h
in
e
lear
n
in
g
f
o
r
p
r
ec
is
e
p
lan
t d
is
ea
s
e
class
if
ica
tio
n
.
4
.
4
.
Cla
s
s
-
wis
e
a
na
ly
s
is
T
ab
le
1
s
h
o
ws
th
e
d
etailed
class
if
icatio
n
m
etr
ics
f
o
r
ea
ch
ca
teg
o
r
y
.
T
h
e
h
ea
lth
y
class
ac
h
iev
ed
th
e
h
ig
h
est
p
er
f
o
r
m
an
ce
with
a
p
r
ec
is
io
n
o
f
0
.
9
5
,
r
ec
all
o
f
1
.
0
0
,
an
d
F1
-
s
co
r
e
o
f
0
.
9
8
.
T
h
e
b
ea
n
r
u
s
t
class
f
o
llo
wed
clo
s
ely
with
0
.
8
3
p
r
ec
is
io
n
an
d
0
.
9
5
r
ec
all.
An
g
u
lar
leaf
s
p
o
t
ac
h
iev
e
d
1
.
0
0
p
r
ec
is
io
n
b
u
t
s
lig
h
tly
lo
wer
r
ec
all
(
0
.
8
0
)
,
in
d
icatin
g
s
o
m
e
co
n
f
u
s
io
n
with
v
is
u
ally
s
im
ilar
d
is
ea
s
e
p
atter
n
s
.
T
ab
le
1
.
C
lass
if
icatio
n
m
etr
ics f
o
r
ea
c
h
ca
teg
o
r
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F
1
-
sc
o
r
e
S
u
p
p
o
r
t
A
n
g
u
l
a
r
_
l
e
a
f
_
s
p
o
t
1
.
0
0
0
.
8
0
0
.
8
9
20
B
e
a
n
_
r
u
s
t
0
.
8
3
0
.
9
5
0
.
8
8
20
H
e
a
l
t
h
y
0
.
9
5
1
.
0
0
0
.
9
8
20
A
c
c
u
r
a
c
y
0
.
9
2
60
M
a
c
r
o
a
v
g
0
.
9
3
0
.
9
2
0
.
9
2
60
W
e
i
g
h
t
e
d
a
v
g
0
.
9
3
0
.
9
2
0
.
9
2
60
4
.
5
.
Co
nfusi
o
n
m
a
t
rix
ins
ig
hts
T
h
e
co
n
f
u
s
io
n
m
atr
ix
in
Fig
u
r
e
2
,
p
r
o
v
id
es
a
d
etailed
v
iew
o
f
th
e
class
-
wis
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
ViT
–
PC
A
–
SVM
p
i
p
elin
e.
T
h
e
m
o
d
el
d
em
o
n
s
tr
ated
p
er
f
ec
t
class
if
icatio
n
f
o
r
th
e
h
ea
lth
y
class
,
ac
h
iev
in
g
1
0
0
%
p
r
ec
is
io
n
an
d
r
ec
all
with
all
2
0
s
am
p
les
co
r
r
ec
tly
id
en
tifie
d
.
Fo
r
b
e
an
r
u
s
t,
1
9
o
u
t
o
f
2
0
s
am
p
les
wer
e
c
o
r
r
ec
tly
cl
ass
if
ied
,
with
a
s
in
g
le
in
s
tan
c
e
m
is
class
if
ied
as
h
ea
lth
y
.
T
h
e
an
g
u
lar
leaf
s
p
o
t
class
s
h
o
wed
s
lig
h
tly
lo
wer
p
er
f
o
r
m
a
n
ce
,
with
1
6
co
r
r
ec
t
p
r
ed
ictio
n
s
an
d
4
m
is
class
if
ica
tio
n
s
as
b
ea
n
r
u
s
t.
T
h
is
in
d
icate
s
th
at
wh
ile
th
e
m
o
d
el
is
h
ig
h
ly
ef
f
ec
tiv
e
o
v
er
all,
s
u
b
tle
v
is
u
al
s
im
ilar
ities
b
etwe
en
an
g
u
lar
leaf
s
p
o
t a
n
d
b
ea
n
r
u
s
t le
s
io
n
s
co
n
t
r
ib
u
te
to
o
cc
asio
n
al
er
r
o
r
s
.
Fig
u
r
e
2
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
th
e
p
r
o
p
o
s
ed
ViT
–
PC
A
–
SVM
p
ip
elin
e
o
n
th
e
test
d
ataset
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.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
3
9
9
-
1
4
0
6
1404
Ov
er
all,
th
e
p
e
r
-
class
ac
cu
r
ac
ies
r
em
ain
ed
co
n
s
is
ten
tly
h
ig
h
,
with
r
ec
all
v
al
u
es
o
f
0
.
8
0
f
o
r
an
g
u
lar
leaf
s
p
o
t,
0
.
9
5
f
o
r
b
ea
n
r
u
s
t,
an
d
1
.
0
0
f
o
r
h
ea
lth
y
.
T
h
ese
r
esu
lts
h
ig
h
lig
h
t
th
e
m
o
d
el’
s
a
b
ilit
y
to
d
is
tin
g
u
is
h
b
etwe
en
h
ea
lth
y
an
d
d
is
ea
s
ed
leav
es
with
s
tr
o
n
g
r
eliab
ilit
y
,
wh
ile
also
r
ev
ea
lin
g
th
e
n
ee
d
f
o
r
f
u
r
th
er
f
in
e
-
tu
n
in
g
to
b
etter
s
ep
ar
ate
clo
s
el
y
r
elate
d
d
is
ea
s
e
s
y
m
p
to
m
s
.
4
.
6
.
Co
m
pa
ra
t
iv
e
ev
a
lua
t
io
n
T
o
v
alid
ate
th
e
ef
f
ec
tiv
en
ess
o
f
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
p
ip
eli
n
e
(
ViT
+PC
A+
S
VM
)
,
we
co
m
p
ar
ed
it
with
ex
is
tin
g
ap
p
r
o
ac
h
es
r
ep
o
r
ted
in
liter
atu
r
e,
in
clu
d
in
g
C
NN
b
aselin
es,
tr
an
s
f
er
lear
n
in
g
with
VGG1
6
/R
esNet,
an
d
tr
ad
itio
n
al
SVM
clas
s
if
ier
s
wh
ich
is
li
s
ted
in
T
ab
le
2
.
T
h
e
r
esu
lts
co
n
f
ir
m
p
r
io
r
f
in
d
i
n
g
s
th
at
tr
an
s
f
er
lear
n
in
g
m
o
d
els
s
u
ch
as
VGG1
6
an
d
R
esNe
t
1
8
im
p
r
o
v
e
class
if
icatio
n
p
er
f
o
r
m
an
ce
f
o
r
p
lan
t
d
is
ea
s
es
[
1
9
]
,
[
6
]
.
Ho
wev
e
r
,
t
h
eir
co
m
p
u
tatio
n
al
o
v
er
h
ea
d
m
ak
es
th
em
less
s
u
itab
le
f
o
r
e
d
g
e
o
r
m
o
b
ile
-
b
ased
d
ep
lo
y
m
e
n
ts
.
T
r
ad
itio
n
al
SV
Ms
with
h
an
d
cr
af
ted
o
r
r
aw
p
i
x
el
f
ea
tu
r
es
f
ail
to
g
e
n
er
alize
ef
f
ec
tiv
ely
,
y
ield
i
n
g
s
ig
n
if
ican
tly
lo
wer
ac
cu
r
ac
y
[
1
2
]
.
T
ab
le
2
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
o
f
d
i
f
f
er
en
t
m
eth
o
d
s
o
n
p
l
an
t d
is
ea
s
e
d
atasets
M
e
t
h
o
d
A
c
c
u
r
a
c
y
(
%)
W
e
i
g
h
t
e
d
F
1
-
s
c
o
r
e
R
e
f
e
r
e
n
c
e
C
N
N
(
S
h
a
l
l
o
w
,
3
C
o
n
v
l
a
y
e
r
s)
8
4
.
5
0
.
8
3
[
2
]
V
G
G
1
6
(
Tr
a
n
sf
e
r
l
e
a
r
n
i
n
g
)
8
8
.
7
0
.
8
7
[
1
8
]
R
e
sN
e
t
1
8
(
Tr
a
n
sf
e
r
l
e
a
r
n
i
n
g
)
9
0
.
1
0
.
8
9
[
6
]
S
V
M
(
R
a
w
p
i
x
e
l
f
e
a
t
u
r
e
s
)
7
5
.
4
0
.
7
4
[
1
2
]
Pr
o
p
o
sed
V
i
T
+PC
A
+S
V
M
9
2
.
0
0
.
9
2
O
u
r
w
o
r
k
B
y
co
m
b
in
in
g
tr
a
n
s
f
o
r
m
er
-
b
a
s
ed
f
ea
tu
r
e
e
x
tr
ac
tio
n
ViT
with
d
im
en
s
io
n
ality
r
ed
u
ctio
n
P
C
A
an
d
a
lig
h
tweig
h
t
SVM
class
if
ier
,
o
u
r
ap
p
r
o
ac
h
ac
h
ie
v
es
th
e
b
est
tr
ad
e
-
o
f
f
b
etwe
en
ac
cu
r
a
cy
an
d
e
f
f
icien
cy
,
m
ak
in
g
it
h
ig
h
ly
p
r
o
m
is
in
g
f
o
r
r
ea
l
-
wo
r
ld
s
m
ar
t
f
ar
m
i
n
g
ap
p
licatio
n
s
.
Fig
u
r
e
3
s
h
o
ws
p
er
f
o
r
m
an
c
e
co
m
p
ar
is
o
n
ch
a
r
t
s
h
o
win
g
h
p
r
o
p
o
s
ed
ViT
+PC
A+
SV
M
p
ip
elin
e
o
u
tp
er
f
o
r
m
s
tr
ad
itio
n
al
C
NNs,
p
r
etr
ain
ed
n
etwo
r
k
s
(
VGG1
6
,
R
esNet1
8
)
,
an
d
b
aselin
e
SVM
class
if
ier
s
.
Fig
u
r
e
3
.
C
o
m
p
a
r
ativ
e
p
e
r
f
o
r
m
an
ce
o
f
p
r
o
p
o
s
ed
an
d
ex
is
tin
g
m
eth
o
d
s
5.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
in
tr
o
d
u
ce
d
a
h
y
b
r
i
d
f
r
am
ewo
r
k
f
o
r
p
la
n
t
leaf
d
is
ea
s
e
d
etec
tio
n
,
co
m
b
in
in
g
V
iT
f
ea
tu
r
e
ex
tr
ac
tio
n
,
PC
A
f
o
r
d
im
en
s
io
n
ality
r
ed
u
ctio
n
,
an
d
a
tu
n
ed
SVM
f
o
r
class
if
icatio
n
.
T
h
e
p
ip
elin
e
ac
h
iev
ed
9
2
%
ac
cu
r
ac
y
an
d
a
weig
h
t
ed
F1
-
s
co
r
e
o
f
0
.
9
2
,
o
u
t
p
er
f
o
r
m
in
g
co
n
v
e
n
tio
n
al
C
NN
-
b
ased
an
d
tr
an
s
f
er
lear
n
in
g
ap
p
r
o
ac
h
es r
ep
o
r
ted
i
n
ea
r
lier
wo
r
k
s
.
T
h
e
u
s
e
o
f
ViT
e
n
s
u
r
ed
r
o
b
u
s
t
f
ea
tu
r
e
r
ep
r
esen
tatio
n
,
PC
A
ad
d
r
ess
ed
th
e
cu
r
s
e
o
f
d
im
en
s
io
n
ality
,
an
d
SVM
p
r
o
v
id
ed
s
tr
o
n
g
g
e
n
er
aliza
tio
n
,
m
a
k
in
g
th
e
f
r
am
ewo
r
k
b
o
th
ef
f
icien
t
an
d
ef
f
e
ctiv
e.
C
o
m
p
ar
ativ
e
an
aly
s
is
ag
ain
s
t
VGG1
6
,
R
es
Net1
8
,
an
d
b
aselin
e
C
NN
class
if
ier
s
co
n
f
ir
m
ed
th
e
s
u
p
er
io
r
ity
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
in
ter
m
s
o
f
ac
cu
r
ac
y
an
d
co
m
p
u
tatio
n
al
e
f
f
icien
cy
.
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
Tr
a
n
s
fo
r
mer
-
b
a
s
ed
h
yb
r
id
cla
s
s
ifica
tio
n
fo
r
p
la
n
t le
a
f d
is
ea
s
e
d
etec
tio
n
…
(
V
ija
ya
la
ksh
mi
S
.
A
b
b
ig
eri
)
1405
T
h
e
p
r
o
p
o
s
ed
h
y
b
r
id
ViT
–
PC
A
–
SVM
f
r
am
ewo
r
k
d
em
o
n
s
tr
ates
s
tr
o
n
g
p
er
f
o
r
m
a
n
ce
f
o
r
p
lan
t
leaf
d
is
ea
s
e
d
etec
tio
n
,
ac
h
iev
i
n
g
an
ac
cu
r
ac
y
o
f
9
2
%
o
n
th
e
B
ea
n
L
ea
f
L
esio
n
s
d
ataset.
T
h
e
in
teg
r
atio
n
o
f
tr
an
s
f
o
r
m
er
-
b
ased
f
ea
tu
r
e
e
x
t
r
ac
tio
n
with
d
im
en
s
io
n
ality
r
ed
u
ctio
n
an
d
class
ical
class
if
i
ca
tio
n
p
r
o
v
id
es
a
n
ef
f
ec
tiv
e
b
alan
ce
b
etwe
en
ac
c
u
r
ac
y
a
n
d
c
o
m
p
u
tatio
n
al
ef
f
ic
ien
cy
.
Fu
tu
r
e
wo
r
k
will
f
o
cu
s
o
n
e
v
alu
atin
g
t
h
e
m
o
d
el
o
n
lar
g
er
m
u
lti
-
cr
o
p
d
atasets
an
d
ex
p
lo
r
in
g
r
ea
l
-
tim
e
d
ep
lo
y
m
en
t
f
o
r
m
o
b
ile
-
b
ase
d
s
m
ar
t
ag
r
icu
ltu
r
e
ap
p
licatio
n
s
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
t
r
ib
u
to
r
R
o
les
T
a
x
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
i
d
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Vijay
alak
s
h
m
i S
.
Ab
b
ig
er
i
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Gee
th
a
D
.
Dev
an
ag
a
v
i
✓
✓
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
All a
u
th
o
r
s
d
ec
lar
e
th
at
th
ey
h
av
e
n
o
c
o
n
f
licts
o
f
in
te
r
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
o
p
en
l
y
av
ailab
le
in
Ka
g
g
le,
"Bean
le
af
lesi
o
n
s
d
ata
s
et,
"
[
On
lin
e]
at
h
ttp
s
:
//w
w
w
.
ka
g
g
le.
co
m/d
a
ta
s
ets/
a
d
va
yp
r
a
s
a
d
/b
ea
n
-
lea
f
-
lesi
o
n
s
-
d
a
t
a
s
et
.
RE
F
E
R
E
NC
E
S
[
1
]
F
o
o
d
a
n
d
A
g
r
i
c
u
l
t
u
r
e
O
r
g
a
n
i
z
a
t
i
o
n
(
F
A
O
)
,
“
C
l
i
ma
t
e
c
h
a
n
g
e
f
a
n
s
sp
r
e
a
d
o
f
p
e
s
t
s a
n
d
t
h
r
e
a
t
e
n
s
p
l
a
n
t
s
a
n
d
c
r
o
p
s,
n
e
w
F
A
O
st
u
d
y
,
”
Fo
o
d
a
n
d
A
g
ri
c
u
l
t
u
re
O
rg
a
n
i
z
a
t
i
o
n
(
FAO
)
,
2
0
2
1
.
h
t
t
p
s
:
/
/
w
w
w
.
f
a
o
.
o
r
g
/
n
e
w
sr
o
o
m
/
d
e
t
a
i
l
/
C
l
i
ma
t
e
-
c
h
a
n
g
e
-
f
a
n
s
-
sp
r
e
a
d
-
of
-
p
e
st
s
-
a
n
d
-
t
h
r
e
a
t
e
n
s
-
p
l
a
n
t
s
-
a
n
d
-
c
r
o
p
s
-
n
e
w
-
F
A
O
-
st
u
d
y
/
e
n
(
a
c
c
e
ss
e
d
J
u
n
.
0
2
,
2
0
2
1
)
.
[
2
]
S
.
P
.
M
o
h
a
n
t
y
,
D
.
P
.
H
u
g
h
e
s,
a
n
d
M
.
S
a
l
a
t
h
é
,
“
U
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
f
o
r
i
m
a
g
e
-
b
a
s
e
d
p
l
a
n
t
d
i
se
a
se
d
e
t
e
c
t
i
o
n
,
”
F
ro
n
t
i
e
rs
i
n
Pl
a
n
t
S
c
i
e
n
c
e
,
v
o
l
.
7
,
p
.
1
4
1
9
,
2
0
1
6
,
d
o
i
:
1
0
.
3
3
8
9
/
f
p
l
s
.
2
0
1
6
.
0
1
4
1
9
.
[
3
]
Y
.
Le
C
u
n
,
Y
.
B
e
n
g
i
o
,
a
n
d
G
.
H
i
n
t
o
n
,
“
D
e
e
p
l
e
a
r
n
i
n
g
,
”
N
a
t
u
re
,
v
o
l
.
5
2
1
,
n
o
.
7
5
5
3
,
p
p
.
4
3
6
–
4
4
4
,
M
a
y
2
0
1
5
,
d
o
i
:
1
0
.
1
0
3
8
/
n
a
t
u
r
e
1
4
5
3
9
.
[
4
]
A
.
K
r
i
z
h
e
v
s
k
y
,
I
.
S
u
t
s
k
e
v
e
r
,
a
n
d
G
.
E.
H
i
n
t
o
n
,
“
I
mag
e
N
e
t
c
l
a
ssi
f
i
c
a
t
i
o
n
w
i
t
h
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
,
”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
o
f
t
h
e
A
C
M
,
v
o
l
.
6
0
,
n
o
.
6
,
p
p
.
8
4
–
9
0
,
M
a
y
2
0
1
7
,
d
o
i
:
1
0
.
1
1
4
5
/
3
0
6
5
3
8
6
.
[
5
]
K
.
H
e
,
X
.
Zh
a
n
g
,
S
.
R
e
n
,
a
n
d
J.
S
u
n
,
“
D
e
e
p
r
e
s
i
d
u
a
l
l
e
a
r
n
i
n
g
f
o
r
i
ma
g
e
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
P
a
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
,
J
u
n
.
2
0
1
6
,
v
o
l
.
2
0
1
6
-
D
e
c
e
m
b
e
r
,
p
p
.
7
7
0
–
7
7
8
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
9
0
.
[
6
]
E.
C
.
T
o
o
,
L.
Y
u
j
i
a
n
,
S
.
N
j
u
k
i
,
a
n
d
L.
Y
i
n
g
c
h
u
n
,
“
A
c
o
m
p
a
r
a
t
i
v
e
s
t
u
d
y
o
f
f
i
n
e
-
t
u
n
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
s
f
o
r
p
l
a
n
t
d
i
s
e
a
s
e
i
d
e
n
t
i
f
i
c
a
t
i
o
n
,
”
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
6
1
,
p
p
.
2
7
2
–
2
7
9
,
J
u
n
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
a
g
.
2
0
1
8
.
0
3
.
0
3
2
.
[
7
]
A
.
D
o
s
o
v
i
t
s
k
i
y
e
t
a
l
.
,
“
A
n
i
ma
g
e
i
s
w
o
r
t
h
1
6
×
1
6
w
o
r
d
s:
t
r
a
n
sf
o
r
m
e
r
s
f
o
r
i
m
a
g
e
r
e
c
o
g
n
i
t
i
o
n
a
t
s
c
a
l
e
,
”
I
C
L
R
2
0
2
1
-
9
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
L
e
a
r
n
i
n
g
Re
p
res
e
n
t
a
t
i
o
n
s
,
p
p
.
1
–
2
1
,
2
0
2
1
.
[
8
]
Z.
Li
u
e
t
a
l
.
,
“
S
w
i
n
t
r
a
n
sf
o
r
mer
:
h
i
e
r
a
r
c
h
i
c
a
l
v
i
si
o
n
t
r
a
n
sf
o
r
mer u
s
i
n
g
s
h
i
f
t
e
d
w
i
n
d
o
w
s,
”
i
n
Pr
o
c
e
e
d
i
n
g
s o
f
t
h
e
I
E
EE
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
Vi
s
i
o
n
,
2
0
2
1
,
p
p
.
9
9
9
2
–
1
0
0
0
2
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
4
8
9
2
2
.
2
0
2
1
.
0
0
9
8
6
.
[
9
]
I
.
T.
Jo
l
l
i
f
e
a
n
d
J.
C
a
d
i
ma,
“
P
r
i
n
c
i
p
a
l
c
o
m
p
o
n
e
n
t
a
n
a
l
y
si
s
:
A
r
e
v
i
e
w
a
n
d
r
e
c
e
n
t
d
e
v
e
l
o
p
m
e
n
t
s,
”
P
h
i
l
o
so
p
h
i
c
a
l
T
r
a
n
s
a
c
t
i
o
n
s o
f
t
h
e
Ro
y
a
l
S
o
c
i
e
t
y
A:
M
a
t
h
e
m
a
t
i
c
a
l
,
Ph
y
s
i
c
a
l
a
n
d
En
g
i
n
e
e
r
i
n
g
S
c
i
e
n
c
e
s
,
v
o
l
.
3
7
4
,
n
o
.
2
0
6
5
,
p
.
2
0
1
5
0
2
0
2
,
2
0
1
6
,
d
o
i
:
1
0
.
1
0
9
8
/
r
st
a
.
2
0
1
5
.
0
2
0
2
.
[
1
0
]
C
.
C
o
r
t
e
s
a
n
d
V
.
V
a
p
n
i
k
,
“
S
u
p
p
o
r
t
-
v
e
c
t
o
r
n
e
t
w
o
r
k
s,”
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
,
v
o
l
.
2
0
,
n
o
.
3
,
p
p
.
2
7
3
–
2
9
7
,
S
e
p
.
1
9
9
5
,
d
o
i
:
1
0
.
1
0
0
7
/
b
f
0
0
9
9
4
0
1
8
.
[
1
1
]
P
.
D
o
m
i
n
g
o
s,
“
A
f
e
w
u
s
e
f
u
l
t
h
i
n
g
s
t
o
k
n
o
w
a
b
o
u
t
mac
h
i
n
e
l
e
a
r
n
i
n
g
,
”
C
o
m
m
u
n
i
c
a
t
i
o
n
s
o
f
t
h
e
AC
M
,
v
o
l
.
5
5
,
n
o
.
1
0
,
p
p
.
7
8
–
8
7
,
2
0
1
2
,
d
o
i
:
1
0
.
1
1
4
5
/
2
3
4
7
7
3
6
.
2
3
4
7
7
5
5
.
[
1
2
]
P
.
R
e
v
a
t
h
i
a
n
d
M
.
H
e
m
a
l
a
t
h
a
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
c
o
t
t
o
n
l
e
a
f
s
p
o
t
d
i
s
e
a
s
e
s
u
si
n
g
i
m
a
g
e
p
r
o
c
e
ss
i
n
g
e
d
g
e
d
e
t
e
c
t
i
o
n
t
e
c
h
n
i
q
u
e
s,
”
i
n
I
EEE
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Em
e
r
g
i
n
g
T
re
n
d
s
i
n
S
c
i
e
n
c
e
E
n
g
i
n
e
e
ri
n
g
a
n
d
T
e
c
h
n
o
l
o
g
y
:
Re
c
e
n
t
Ad
v
a
n
c
e
m
e
n
t
s
o
n
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
ri
n
g
I
n
n
o
v
a
t
i
o
n
,
I
N
C
O
S
ET
2
0
1
2
,
2
0
1
2
,
p
p
.
1
6
9
–
1
7
3
,
d
o
i
:
1
0
.
1
1
0
9
/
i
n
c
o
se
t
.
2
0
1
2
.
6
5
1
3
9
0
0
.
[
1
3
]
K
.
S
i
m
o
n
y
a
n
a
n
d
A
.
Zi
ss
e
r
ma
n
,
“
V
e
r
y
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
s
f
o
r
l
a
r
g
e
-
s
c
a
l
e
i
ma
g
e
r
e
c
o
g
n
i
t
i
o
n
,
”
3
rd
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
L
e
a
rn
i
n
g
Re
p
res
e
n
t
a
t
i
o
n
s
,
I
C
L
R
2
0
1
5
-
C
o
n
f
e
re
n
c
e
T
r
a
c
k
P
ro
c
e
e
d
i
n
g
s
,
p
p
.
1
–
1
4
,
2
0
1
5
.
[
1
4
]
K
.
P
.
F
e
r
e
n
t
i
n
o
s
,
“
D
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
s
f
o
r
p
l
a
n
t
d
i
se
a
se
d
e
t
e
c
t
i
o
n
a
n
d
d
i
a
g
n
o
si
s,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
ro
n
i
c
s
i
n
A
g
r
i
c
u
l
t
u
r
e
,
v
o
l
.
1
4
5
,
p
p
.
3
1
1
–
3
1
8
,
F
e
b
.
2
0
1
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
a
g
.
2
0
1
8
.
0
1
.
0
0
9
.
[
1
5
]
H
.
S
.
D
e
y
,
M
.
K
a
m
a
l
,
a
n
d
A
.
R
.
N
u
r
u
z
z
a
ma
n
,
“
A
h
y
b
r
i
d
C
N
N
-
S
V
M
a
p
p
r
o
a
c
h
f
o
r
p
l
a
n
t
d
i
se
a
se
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EE
E
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
El
e
c
t
r
i
c
a
l
,
C
o
m
p
u
t
e
r
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
En
g
i
n
e
e
ri
n
g
(
E
C
C
E)
,
2
0
1
9
,
p
p
.
1
–
6.
[
1
6
]
Y
.
X
u
a
n
d
O
t
h
e
r
s,
“
H
y
b
r
i
d
d
e
e
p
f
e
a
t
u
r
e
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
c
l
a
ssi
f
i
e
r
a
p
p
r
o
a
c
h
f
o
r
b
r
e
a
st
c
a
n
c
e
r
h
i
s
t
o
l
o
g
y
i
ma
g
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
8
,
p
p
.
2
1
3
7
8
–
2
1
3
8
7
,
2
0
2
0
.
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.
1
6
,
No
.
3
,
J
u
n
e
20
2
6
:
1
3
9
9
-
1
4
0
6
1406
[
1
7
]
Z.
K
h
a
n
a
n
d
O
t
h
e
r
s,
“
V
i
si
o
n
t
r
a
n
sf
o
r
mers
f
o
r
sma
l
l
-
sc
a
l
e
i
ma
g
e
c
l
a
ssi
f
i
c
a
t
i
o
n
:
A
n
e
m
p
i
r
i
c
a
l
st
u
d
y
,
”
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
2
,
n
o
.
4
,
p
.
2
1
5
2
,
2
0
2
2
.
[
1
8
]
K
a
g
g
l
e
,
“
B
e
a
n
l
e
a
f
l
e
s
i
o
n
s
d
a
t
a
s
e
t
,
”
k
a
g
g
l
e
.
c
o
m
/
.
h
t
t
p
s
:
/
/
w
w
w
.
k
a
g
g
l
e
.
c
o
m
/
d
a
t
a
se
t
s/
a
d
v
a
y
p
r
a
sa
d
/
b
e
a
n
-
l
e
a
f
-
l
e
si
o
n
s
-
d
a
t
a
s
e
t
(
a
c
c
e
s
se
d
F
e
b
.
0
9
,
2
0
2
6
)
.
[
1
9
]
R
.
B
.
P
a
d
a
m
a
t
a
a
n
d
S
.
K
.
A
t
l
u
r
i
,
“
T
o
ma
t
o
c
r
o
p
d
i
se
a
se
c
l
a
ssi
f
i
c
a
t
i
o
n
u
s
i
n
g
s
e
ma
n
t
i
c
s
e
g
me
n
t
a
t
i
o
n
a
l
g
o
r
i
t
h
m
i
n
d
e
e
p
l
e
a
r
n
i
n
g
,
”
Re
v
u
e
d
’
I
n
t
e
l
l
i
g
e
n
c
e
Art
i
f
i
c
i
e
l
l
e
,
v
o
l
.
3
7
,
n
o
.
2
,
p
p
.
4
1
5
–
4
2
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
8
2
8
0
/
r
i
a
.
3
7
0
2
1
8
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Vij
a
y
a
l
a
k
s
h
m
i
S
.
A
b
b
i
g
e
r
i
S
c
h
o
o
l
o
f
C
o
m
p
u
te
r
S
c
ie
n
c
e
a
n
d
En
g
i
n
e
e
r
i
n
g
,
R
EV
A
Un
i
v
e
rs
i
t
y
,
Be
n
g
a
l
u
r
u
,
I
n
d
ia
.
Sh
e
c
a
n
b
e
c
o
n
tac
te
d
a
t
e
m
a
il
:
v
i
ja
y
a
la
x
m
isa
@
g
m
a
i
l.
c
o
m
.
G
e
e
th
a
D
.
De
v
a
n
a
g
a
v
i
re
c
e
iv
e
d
h
e
r
P
h
.
D.
in
2
0
1
4
.
S
h
e
is cu
rre
n
tl
y
a
p
ro
fe
ss
o
r
a
t
REVA
Un
iv
e
rsity
.
S
h
e
h
a
s
2
6
y
e
a
rs
o
f
tea
c
h
in
g
e
x
p
e
rien
c
e
.
He
r
re
se
a
rc
h
in
tere
st
in
c
lu
d
e
s
wire
les
s
se
n
so
r
n
e
two
rk
s,
n
e
two
r
k
se
c
u
rit
y
,
a
n
d
c
o
m
p
u
ter
n
e
two
r
k
s.
S
h
e
h
a
s
g
o
o
d
n
u
m
b
e
r
o
f
p
u
b
li
c
a
ti
o
n
s
i
n
re
p
u
ted
in
tern
a
ti
o
n
a
l
jo
u
rn
a
ls.
S
h
e
h
a
s
g
u
i
d
e
d
4
P
h
.
D.
sc
h
o
lars
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
g
e
e
th
a
@re
v
a
.
e
d
u
.
i
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.