I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
1
2
,
No
.
2
,
N
o
v
e
m
b
er
201
8
,
p
p
.
4
6
8
~
4
75
I
SS
N:
2502
-
4752
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ee
cs
.
v
1
2
.i
2
.
p
p
468
-
4
7
5
468
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e.
co
m/jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
ijeec
s
Ev
a
lua
tion o
f
CN
N,
Alex
n
et
a
nd
Go
o
g
leNet
for
Frui
t
Recog
nition
Nur
Azida
M
uh
a
m
m
a
d
1
,
Am
e
lin
a
Ab N
a
s
ir
2
,
Z
a
ida
h I
b
ra
hi
m
3
,
Nurba
it
y
Sa
bri
4
1,
2,
3
F
a
c
u
l
ty
o
f
Co
m
p
u
ter an
d
M
a
th
e
m
a
ti
c
a
l
S
c
ien
c
e
s,
Un
iv
e
rsiti
T
e
k
n
o
lo
g
i
M
A
RA
,
S
h
a
h
A
la
m
,
S
e
la
n
g
o
r,
M
a
lay
sia
4
F
a
c
u
lt
y
o
f
Co
m
p
u
ter an
d
M
a
th
e
m
a
ti
c
a
l
S
c
ien
c
e
s,
Un
iv
e
rsiti
T
e
k
n
o
lo
g
i
M
A
RA
,
Ca
m
p
u
s Ja
sin
,
M
e
l
a
k
a
,
M
a
la
y
si
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
y
1
9
,
2
0
1
8
R
ev
i
s
ed
J
u
n
3
0
,
2
0
1
8
A
cc
ep
ted
Ju
l
11
,
2
0
1
8
F
ru
it
re
c
o
g
n
it
i
o
n
is
u
se
f
u
l
f
o
r
a
u
to
m
a
ti
c
f
ru
it
h
a
rv
e
stin
g
.
F
ru
it
r
e
c
o
g
n
it
io
n
a
p
p
li
c
a
ti
o
n
c
a
n
re
d
u
c
e
o
r
m
i
n
im
ize
h
u
m
a
n
in
terv
e
n
ti
o
n
d
u
rin
g
f
ru
it
h
a
rv
e
stin
g
o
p
e
ra
ti
o
n
.
H
o
w
e
v
e
r,
in
c
o
m
p
u
ter
v
isio
n
,
f
ru
it
re
c
o
g
n
it
i
o
n
is
v
e
r
y
c
h
a
ll
e
n
g
in
g
b
e
c
a
u
se
o
f
si
m
il
a
r
s
h
a
p
e
s,
c
o
lo
rs
a
n
d
tex
tu
re
s
a
m
o
n
g
v
a
rio
u
s
f
ru
it
s.
Ill
u
m
in
a
ti
o
n
s
c
h
a
n
g
e
s
d
u
e
to
w
e
a
th
e
r
c
o
n
d
it
io
n
a
lso
lea
d
to
a
c
h
a
ll
e
n
g
in
g
tas
k
f
o
r
f
ru
it
re
c
o
g
n
it
io
n
.
T
h
u
s,
t
h
is
p
a
p
e
r
ten
d
s t
o
i
n
v
e
stig
a
te t
h
e
p
e
rf
o
r
m
a
n
c
e
o
f
b
a
sic
Co
n
v
o
lu
ti
o
n
a
l
Ne
u
ra
l
Ne
t
w
o
rk
(CNN
),
Ale
x
n
e
t
a
n
d
G
o
o
g
len
e
t
in
re
c
o
g
n
izi
n
g
n
i
n
e
d
if
fe
re
n
t
ty
p
e
s
o
f
f
ru
it
s
f
ro
m
a
p
u
b
li
c
ly
a
v
a
il
a
b
le
d
a
tas
e
t.
T
h
e
e
x
p
e
ri
m
e
n
tal
re
su
lt
s
i
n
d
ica
te
t
h
a
t
a
ll
t
h
e
se
tec
h
n
iq
u
e
s
p
ro
d
u
c
e
e
x
c
e
ll
e
n
t
re
c
o
g
n
it
io
n
a
c
c
u
ra
c
y
,
b
u
t
b
a
sic
CNN
a
c
h
iev
e
s
th
e
f
a
ste
s
t
re
c
o
g
n
it
io
n
re
su
lt
c
o
m
p
a
re
d
w
it
h
A
le
x
n
e
t
a
n
d
G
o
o
g
len
e
t.
K
ey
w
o
r
d
s
:
A
le
x
n
et
C
NN
Fru
i
t
R
ec
o
g
n
itio
n
Go
o
g
len
et
Neu
r
al
Net
w
o
r
k
Co
p
y
rig
h
t
©
2
0
1
8
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Nu
r
A
zid
a
Mu
h
a
m
m
ad
,
Facu
lt
y
o
f
C
o
m
p
u
ter
an
d
Ma
t
h
e
m
a
tical
Scie
n
ce
s
,
Un
i
v
er
s
iti T
ek
n
o
lo
g
i M
A
R
A
,
S
h
a
h
A
la
m
,
Selan
g
o
r
,
Ma
la
y
s
i
a
.
E
m
ail:
az
id
a
m
u
h
a
m
m
ad
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Fru
it
r
ec
o
g
n
itio
n
is
u
s
ef
u
l
f
o
r
au
to
m
a
tic
f
r
u
it
h
ar
v
e
s
ti
n
g
.
B
y
h
a
v
i
n
g
f
r
u
it
r
ec
o
g
n
itio
n
ap
p
licatio
n
,
i
t
ca
n
r
ed
u
ce
o
r
m
i
n
i
m
ize
h
u
m
a
n
in
ter
v
en
tio
n
i
n
th
eir
f
r
u
it
h
a
r
v
esti
n
g
o
p
er
ati
o
n
s
.
Fr
u
it
r
ec
o
g
n
i
tio
n
s
y
s
te
m
w
i
ll
au
to
m
at
icall
y
d
etec
t
a
n
d
in
s
p
ec
t
th
e
f
r
u
it
f
o
r
h
ar
v
es
tin
g
w
it
h
i
n
t
h
e
i
m
a
g
e.
T
h
e
i
m
p
l
e
m
en
tatio
n
o
f
f
r
u
it
r
ec
o
g
n
itio
n
ap
p
licatio
n
ca
n
a
ls
o
in
cr
ea
s
e
t
h
e
v
al
u
e
o
f
p
r
o
d
u
cts to
t
h
e
co
n
s
u
m
er
s
[
1
]
.
I
n
ad
d
itio
n
,
it
ca
n
r
ed
u
ce
th
e
o
p
er
atio
n
ti
m
e
an
d
h
ar
v
e
s
tin
g
co
s
t.
Fru
it
r
ec
o
g
n
it
io
n
ap
p
licatio
n
is
also
u
s
e
f
u
l
f
o
r
f
r
u
it
d
is
ea
s
e
d
etec
tio
n
in
t
h
e
ea
r
l
y
s
ta
g
e.
Fo
r
clas
s
ic
al
ap
p
r
o
ac
h
,
th
e
d
etec
tio
n
a
n
d
id
en
ti
f
icatio
n
o
f
f
r
u
it
d
is
ea
s
e
is
b
ased
o
n
h
u
m
a
n
n
ak
ed
e
y
e
s
w
h
ic
h
is
ti
m
e
co
n
s
u
m
in
g
an
d
co
s
tl
y
[
2
]
.
T
h
r
o
u
g
h
au
to
m
atic
f
r
u
it
r
ec
o
g
n
it
io
n
p
r
o
ce
s
s
,
it
ca
n
f
ac
ilit
ate
t
h
e
co
n
tr
o
l o
f
f
r
u
i
t d
is
ea
s
es a
s
th
e
d
is
ea
s
e
ca
n
b
e
av
o
id
ed
b
y
ap
p
r
o
p
r
iate
s
p
r
in
k
li
n
g
o
f
p
esti
cid
es.
Var
io
u
s
r
esear
c
h
es
o
n
f
r
u
it
r
ec
o
g
n
itio
n
b
ased
o
n
i
m
a
g
es
h
av
e
b
ee
n
p
er
f
o
r
m
ed
.
M
u
lt
i
p
le
f
ea
t
u
r
e
b
ased
an
al
y
s
i
s
th
at
i
n
cl
u
d
e
co
lo
r
,
s
h
ap
e
an
d
tex
tu
r
e
h
av
e
b
ee
n
ap
p
lied
to
r
ec
o
g
n
ize
s
ix
d
if
f
er
en
t
t
y
p
es
o
f
f
r
u
it
s
t
h
at
ar
e
r
ea
d
ap
p
le,
b
an
an
a,
l
y
ch
ee
,
o
r
a
n
g
e,
p
i
n
ea
p
p
le
an
d
p
o
m
e
g
r
an
ate
[
3
]
.
T
h
e
r
esear
ch
er
s
h
av
e
u
s
ed
th
e
L
o
g
Gab
o
r
f
ilter
to
r
ec
o
g
n
ize
th
e
te
x
tu
r
e
o
f
a
f
r
u
it.
T
h
e
h
u
e
h
as
b
ee
n
ca
lcu
lated
f
o
r
co
lo
r
an
d
s
h
ap
e
w
a
s
b
ein
g
an
al
y
ze
d
b
y
co
u
n
ti
n
g
t
h
e
p
er
i
m
eter
a
n
d
ar
ea
p
ix
el
s
.
I
n
ad
d
itio
n
,
th
e
A
r
ti
f
icial
Neu
r
al
Net
w
o
r
k
(
ANN)
w
a
s
b
ein
g
u
s
ed
f
o
r
th
e
clas
s
i
f
i
ca
tio
n
an
d
it a
ch
iev
e
s
ab
o
u
t
9
0
% c
lass
i
f
icatio
n
ac
c
u
r
ac
y
.
T
h
e
u
s
e
o
f
d
ee
p
lear
n
i
n
g
h
a
s
d
r
a
m
atica
ll
y
i
m
p
r
o
v
e
s
t
h
e
p
er
f
o
r
m
an
ce
o
f
o
b
j
ec
t
d
etec
tio
n
,
s
p
ee
c
h
r
ec
o
g
n
itio
n
,
v
is
u
al
o
b
j
ec
t
r
ec
o
g
n
itio
n
a
n
d
m
a
n
y
o
th
er
d
o
m
ain
s
li
k
e
g
e
n
o
m
ic
s
a
n
d
d
r
u
g
d
is
co
v
er
y
[
1
]
.
Dee
p
lear
n
in
g
is
a
clas
s
o
f
m
ac
h
in
e
lear
n
i
n
g
al
g
o
r
ith
m
s
t
h
at
u
s
e
s
m
u
ltip
le
la
y
er
s
th
a
t
co
n
ta
in
n
o
n
lin
ea
r
p
r
o
ce
s
s
in
g
u
n
i
ts
.
C
o
n
v
o
lu
t
io
n
al
Neu
r
al
Net
w
o
r
k
s
(
C
NNs)
ar
e
clas
s
if
i
ed
as
a
d
ee
p
lear
n
i
n
g
al
g
o
r
ith
m
[
2
]
.
I
t
p
r
o
v
id
es
s
u
cc
e
s
s
f
u
l
r
es
u
lts
i
n
ar
ea
s
o
f
i
m
ag
e
r
ec
o
g
n
itio
n
an
d
cla
s
s
i
f
i
ca
ti
o
n
.
B
esid
es
th
a
t,
A
le
x
n
et
an
d
Go
o
g
le
n
et
ar
e
p
r
e
-
tr
ain
ed
C
NN
m
o
d
els
t
h
at
h
av
e
p
r
o
d
u
ce
d
v
er
y
g
o
o
d
r
esu
lts
f
o
r
th
e
p
as
t
f
e
w
p
ast
y
ea
r
s
[
3
]
.
A
lex
n
et
is
t
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
E
va
lu
a
tio
n
o
f CN
N
,
A
lexn
et
a
n
d
Go
o
g
leN
et
fo
r
F
r
u
it R
ec
o
g
n
itio
n
(
N
u
r
A
z
id
a
Mu
h
a
mma
d
)
469
w
i
n
n
er
o
f
I
m
ag
e
Net
L
ar
g
e
S
ca
le
Vis
u
al
R
ec
o
g
n
itio
n
C
h
al
len
g
e
(
I
L
S
VR
C
)
in
2
0
1
2
w
h
i
le
Go
o
g
len
et
is
t
h
e
w
i
n
n
er
in
2
0
1
4
[
4
]
.
T
h
ese
m
o
d
els
s
h
o
w
b
ig
i
m
p
ac
t
s
o
n
i
m
ag
e
r
ec
o
g
n
it
io
n
an
d
clas
s
i
f
i
ca
tio
n
tas
k
s
as
t
h
e
y
p
r
o
d
u
ce
o
u
ts
tan
d
in
g
p
er
f
o
r
m
an
ce
.
As
a
r
esu
lt,
C
NN
m
o
d
els
w
er
e
w
id
el
y
u
s
ed
in
th
e
f
ield
o
f
co
m
p
u
ter
v
is
io
n
[
5
].
As
t
h
e
C
NN
m
o
d
el
g
o
es
d
ee
p
er
in
t
h
eir
co
n
v
o
l
u
tio
n
ar
ch
i
te
ctu
r
e,
it
ca
n
r
ea
c
h
a
lo
w
er
id
en
ti
f
icatio
n
er
r
o
r
r
ate
co
m
p
ar
ed
to
th
e
h
u
m
an
’s
e
y
e
s
.
T
h
u
s
,
t
h
e
C
NN
m
o
d
el
w
as
i
m
p
le
m
en
ted
f
o
r
f
r
u
it
an
d
v
e
g
etab
les
class
i
f
icatio
n
as
it
p
r
o
d
u
ce
s
g
r
ea
t
r
esu
l
ts
f
o
r
o
th
er
o
b
j
ec
t
r
ec
o
g
n
itio
n
ap
p
licatio
n
s
.
Ho
w
e
v
er
,
i
n
co
m
p
u
ter
v
is
io
n
,
th
e
f
r
u
it
clas
s
if
icatio
n
g
iv
e
s
ch
a
llen
g
e
s
in
i
m
a
g
e
r
ec
o
g
n
itio
n
b
ec
au
s
e
o
f
th
e
s
i
m
ila
r
s
h
ap
es,
co
lo
r
s
an
d
tex
t
u
r
es
a
m
o
n
g
v
ar
io
u
s
f
r
u
i
ts
[
6
]
.
T
h
e
ch
a
n
g
e
s
i
n
t
h
e
lo
ca
tio
n
a
n
d
e
y
e
-
s
ig
h
t
v
ie
w
o
f
t
h
e
f
r
u
it
s
also
lead
to
t
h
i
s
is
s
u
e.
B
esid
es
th
a
t
,
i
n
t
h
e
s
u
p
er
m
ar
k
et,
t
h
e
s
ta
f
f
s
till
r
eq
u
ir
e
s
to
w
ei
g
h
th
e
s
e
lli
n
g
f
r
u
i
t
w
h
ich
ef
f
ec
t
s
th
e
co
s
t
o
f
lab
o
r
,
tim
e
a
n
d
th
e
e
f
f
icie
n
c
y
i
s
lo
w
[
7
]
.
T
h
u
s
,
t
h
e
m
ai
n
o
b
j
ec
tiv
e
o
f
th
i
s
r
esear
ch
i
s
to
in
v
e
s
ti
g
ate
t
h
e
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
p
er
f
o
r
m
an
ce
o
f
b
asic
C
NN,
A
le
x
n
e
t
an
d
Go
o
g
len
et
i
n
r
ec
o
g
n
izi
n
g
f
r
u
it
i
m
ag
e
s
to
s
ee
w
h
et
h
er
th
e
r
es
u
lts
w
ill ac
h
iev
e
m
o
r
e
th
at
9
0
% a
cc
u
r
ac
y
o
r
n
o
t.
2.
RE
L
AT
E
D
WO
RK
Fo
r
th
e
p
as
t
f
e
w
y
ea
r
s
,
m
a
n
y
r
esear
ch
er
s
h
a
v
e
b
ee
n
w
o
r
k
i
n
g
o
n
d
e
v
elo
p
in
g
f
r
u
i
t
r
ec
o
g
n
itio
n
a
n
d
class
i
f
icatio
n
ap
p
r
o
ac
h
es
.
W
C
Sen
g
an
d
S
H
Mir
is
ae
e
[
1
1
]
d
ev
elo
p
ed
a
f
r
u
it
r
ec
o
g
n
it
io
n
s
y
s
te
m
t
h
at
co
m
b
i
n
e
f
ea
t
u
r
e
s
li
k
es
co
lo
r
,
s
ize
a
n
d
s
h
ap
e
b
ased
.
T
h
e
y
u
s
ed
t
h
e
n
e
ar
est
n
ei
g
h
b
o
r
class
i
f
icat
io
n
.
T
h
e
r
esu
lt
s
h
o
w
ed
a
g
o
o
d
p
er
f
o
r
m
a
n
ce
f
o
r
s
i
n
g
le
f
r
u
it
r
ec
o
g
n
itio
n
o
n
l
y
b
u
t
is
n
o
t
s
u
i
tab
le
to
u
s
e
f
o
r
f
r
u
i
t
r
ec
o
g
n
i
tio
n
t
h
at
ar
e
in
a
b
u
n
c
h
.
An
ef
f
icie
n
t
f
u
s
io
n
o
f
tex
tu
r
e
an
d
co
lo
r
f
o
r
f
r
u
it
t
y
p
e
r
ec
o
g
n
itio
n
h
a
s
b
ee
n
p
r
o
p
o
s
ed
.
Ho
w
e
v
er
,
th
e
r
esu
lt
of
t
h
e
r
ec
o
g
n
itio
n
r
ate
is
n
o
t
v
er
y
e
n
co
u
r
a
g
i
n
g
[
1
]
.
L
ec
u
n
,
B
en
g
io
a
n
d
Hin
to
n
[
8
]
p
r
o
p
o
s
ed
f
r
u
it
r
ec
o
g
n
itio
n
u
s
i
n
g
C
NN.
I
t
i
n
v
o
lv
ed
w
it
h
o
u
t
f
ea
tu
r
e
ex
tr
ac
ti
o
n
an
d
th
e
i
n
p
u
t
i
m
ag
e
s
w
er
e
d
ir
ec
tl
y
e
n
ter
ed
i
n
to
th
e
n
e
t
w
o
r
k
.
T
h
e
r
esu
lt
s
s
h
o
w
ed
t
h
at
t
h
e
r
ec
o
g
n
i
tio
n
r
ate
is
i
m
p
r
o
v
ed
a
n
d
it
is
s
u
itab
le
to
id
en
tify
m
u
ltip
le
t
y
p
es o
f
f
r
u
it
s
.
Du
e
to
t
h
e
r
is
i
n
g
v
al
u
es
o
f
a
g
r
icu
lt
u
r
al
s
u
p
p
lies
s
u
c
h
as
a
g
r
o
ch
e
m
ical
s
,
w
ater
ir
r
ig
a
tio
n
a
n
d
p
o
w
er
h
as
lead
to
th
e
ag
r
ic
u
lt
u
r
e
in
d
u
s
tr
y
a
s
o
n
e
o
f
th
e
m
o
s
t
co
s
t
-
d
e
m
an
d
i
n
g
ar
ea
s
.
A
f
r
u
i
t
d
etec
tio
n
s
y
s
te
m
b
y
u
s
i
n
g
d
ee
p
n
e
u
r
al
n
et
w
o
r
k
s
i
s
p
r
o
p
o
s
ed
in
[
9
]
.
T
h
e
p
u
r
p
o
s
e
o
f
th
eir
p
ap
er
is
to
b
u
ild
a
n
ac
c
u
r
ate,
f
a
s
t
a
n
d
r
eliab
le
f
r
u
it
d
etec
t
io
n
s
y
s
te
m
w
h
ic
h
is
a
n
i
m
p
o
r
tan
t
ele
m
e
n
t
o
f
an
a
u
to
n
o
m
o
u
s
ag
r
ic
u
lt
u
r
a
l
r
o
b
o
tic
p
latf
o
r
m
.
T
h
ey
ad
ap
t
th
e
tec
h
n
iq
u
e
o
f
Fas
ter
R
e
g
io
n
-
b
ased
C
N
N
(
R
-
C
NN)
f
o
r
th
e
f
r
u
it
d
etec
tio
n
b
y
u
s
i
n
g
i
m
a
g
er
y
o
b
tain
ed
f
r
o
m
t
w
o
m
o
d
alities
w
h
ic
h
i
s
co
lo
r
(
R
B
G)
a
n
d
Ne
ar
-
I
n
f
r
ar
ed
(
NI
R
)
.
T
h
e
y
p
er
f
o
r
m
ed
f
in
e
-
t
u
n
in
g
o
f
VGG1
6
n
et
w
o
r
k
b
ased
o
n
p
r
e
-
tr
ain
ed
I
m
a
g
eNe
t
m
o
d
el.
T
h
e
co
m
b
i
n
atio
n
o
f
R
GB
a
n
d
NI
R
m
u
l
ti
-
m
o
d
al
i
s
r
etr
ain
ed
to
p
er
f
o
r
m
t
h
e
d
ete
ctio
n
o
f
s
e
v
en
t
y
p
es
o
f
fr
u
it
s
.
As
a
r
es
u
lt,
th
e
ac
c
u
r
ac
y
is
i
m
p
r
o
v
ed
a
n
d
it
i
s
f
aster
to
b
e
d
ep
lo
y
ed
to
r
ec
o
g
n
ize
a
n
e
w
f
r
u
it
t
y
p
e
.
I
t
ta
k
es
o
n
l
y
f
o
u
r
h
o
u
r
s
to
an
n
o
tat
e
an
d
tr
ain
t
h
e
n
e
w
m
o
d
el
p
er
f
r
u
it.
2
.
1
CNN
(
Co
nv
o
lutio
na
l N
eura
l N
et
w
o
rk
)
T
h
e
ar
ch
itectu
r
e
o
f
C
NN
is
s
tr
u
ctu
r
ed
as
a
s
er
ies
o
f
la
y
er
s
,
th
at
co
n
s
i
s
ts
o
f
th
r
ee
la
y
er
s
w
h
ic
h
ar
e
co
n
v
o
lv
e
la
y
er
,
p
o
o
lin
g
la
y
er
an
d
R
ec
tif
ied
L
i
n
ea
r
u
n
it
(
R
eL
u
)
[
1
0
]
.
C
o
n
v
o
lv
e
la
y
er
ex
t
r
ac
ts
f
ea
tu
r
e
s
o
f
an
i
m
a
g
e
u
s
i
n
g
f
ilter
a
n
d
i
m
a
g
e
p
atch
t
h
at
s
tr
id
es
o
v
er
t
h
e
i
n
p
u
t
i
m
a
g
e.
R
e
L
u
la
y
er
r
ep
lac
es
all
n
e
g
ati
v
e
p
i
x
el
v
alu
e
s
i
n
t
h
e
f
ea
tu
r
e
m
ap
w
it
h
ze
r
o
w
h
ile
p
o
o
lin
g
la
y
er
al
lo
w
s
th
e
f
ea
t
u
r
e
m
ap
to
b
e
d
o
w
n
-
s
a
m
p
led
a
f
ter
R
eL
u
la
y
er
to
r
ed
u
ce
th
e
d
i
m
e
n
s
io
n
alit
y
.
M
ax
p
o
o
lin
g
co
m
p
u
tes
t
h
e
m
a
x
i
m
u
m
lo
c
al
o
f
f
ea
t
u
r
e
m
ap
.
Neig
h
b
o
r
in
g
p
o
o
lin
g
tak
e
s
i
n
p
u
t
f
r
o
m
f
ea
t
u
r
e
m
ap
s
t
h
at
a
r
e
s
h
i
f
ted
o
r
s
tr
id
e
b
y
m
o
r
e
th
an
o
n
e
r
o
w
s
o
r
co
lu
m
n
s
.
F
ig
u
r
e
1
s
h
o
w
s
t
h
e
a
r
ch
itect
u
r
e
o
f
a
C
NN.
Fig
u
r
e
1
.
An
ill
u
s
tr
atio
n
o
f
C
NN
la
y
er
s
[
1
0
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l
.
1
2
,
No
.
2
,
No
v
e
m
b
er
201
8
:
4
6
8
–
4
7
5
470
2
.
2
Alex
Net
A
le
x
n
et
is
al
s
o
k
n
o
w
n
as
tr
a
n
s
f
er
lear
n
i
n
g
m
o
d
el
w
h
er
e
k
n
o
w
led
g
e
is
lear
n
t
f
r
o
m
tr
ai
n
in
g
lar
g
e
a
m
o
u
n
t
o
f
d
ata
s
ets.
A
le
x
Net
w
o
n
th
e
I
m
ag
e
Net
L
ar
g
e
Sc
ale
Vis
u
al
R
ec
o
g
n
i
tio
n
C
h
al
l
en
g
e
(
I
L
SV
R
C
)
i
n
2
0
1
2
.
I
t
co
n
s
is
ts
o
f
2
5
la
y
e
r
s
th
at
co
m
b
i
n
e
a
f
e
w
s
tack
s
o
f
co
n
v
o
lu
tio
n
al
la
y
er
s
a
n
d
f
u
ll
y
co
n
n
ec
ted
la
y
er
s
[
1
3
]
.
A
n
ill
u
s
tr
atio
n
o
f
t
h
e
ar
ch
itect
u
r
e
o
f
A
le
x
Net
i
s
s
h
o
w
n
i
n
Fi
g
u
r
e
2
.
Fig
u
r
e
2
.
An
ill
u
s
tr
atio
n
o
f
A
l
ex
Net
la
y
er
s
[
1
4
]
2
.
3
G
o
o
g
leNe
t
Go
o
g
len
e
t
(
a
.
k
.
a.
I
n
c
e
p
ti
o
n
V1
)
is
th
e
w
in
n
er
o
f
th
e
I
L
SVR
C
2
0
1
4
c
o
m
p
eti
ti
o
n
f
r
o
m
Go
o
g
l
e.
I
t
ac
h
iev
e
d
a
t
o
p
-
5
e
r
r
o
r
r
a
te
o
f
6
.
6
7
%
[
1
5
]
.
T
h
is
w
as
v
er
y
clo
s
e
t
o
h
u
m
an
lev
el
p
er
f
o
r
m
an
ce
w
h
ich
th
e
o
r
g
an
ize
r
s
o
f
th
e
ch
al
len
g
e
w
er
e
f
o
r
c
ed
t
o
ev
a
lu
at
e.
A
s
it
t
u
r
n
s
o
u
t
,
th
is
w
as
ac
tu
a
lly
r
at
h
er
h
a
r
d
t
o
d
o
an
d
r
e
q
u
i
r
e
d
s
o
m
e
h
u
m
an
tr
ain
in
g
in
o
r
d
e
r
t
o
p
e
r
f
o
r
m
th
e
task
.
T
h
e
h
u
m
an
ex
p
er
t
(
A
n
d
r
e
j
Ka
r
p
ath
y
)
w
a
s
ab
l
e
t
o
ac
h
iev
e
a
t
o
p
-
5
er
r
o
r
r
at
e
o
f
5
.
1
%
(
s
in
g
le
m
o
d
el
)
an
d
3
.
6
%
(
en
s
em
b
le
)
.
T
h
e
n
etw
o
r
k
u
s
e
d
C
NN
in
s
p
i
r
e
d
b
y
L
eNe
t
b
u
t
im
p
lem
en
te
d
a
n
o
v
el
e
lem
en
t
w
h
ich
is
d
u
b
b
e
d
a
n
in
ce
p
ti
o
n
m
o
d
u
le
.
I
t
u
s
ed
b
atch
n
o
r
m
aliz
ati
o
n
,
im
ag
e
d
is
to
r
ti
o
n
s
an
d
R
MS
p
r
o
p
.
T
h
is
m
o
d
el
is
b
as
e
d
o
n
s
ev
e
r
al
v
e
r
y
s
m
all
co
n
v
o
lu
ti
o
n
s
in
o
r
d
er
t
o
d
r
as
tic
ally
r
e
d
u
ce
th
e
n
u
m
b
er
o
f
p
a
r
am
et
er
s
.
T
h
e
ir
ar
ch
ite
ctu
r
e
c
o
n
s
is
t
ed
o
f
2
2
l
ay
er
s
o
f
d
e
ep
C
N
N
b
u
t
th
e
n
u
m
b
er
o
f
p
a
r
am
ete
r
s
is
r
e
d
u
c
ed
f
r
o
m
6
0
m
illi
o
n
(
A
lex
Ne
t)
to
4
m
illi
o
n
(
G
o
o
g
l
en
et
)
.
A
n
il
lu
s
tr
ati
o
n
o
f
t
h
e
la
y
er
s
in
Go
o
g
leNe
t is s
h
o
w
n
in
F
ig
u
r
e
3
.
Fig
u
r
e
3
.
An
ill
u
s
tr
atio
n
o
f
th
e
la
y
er
s
o
f
Go
o
g
leNe
t [
1
5
]
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
E
va
lu
a
tio
n
o
f CN
N
,
A
lexn
et
a
n
d
Go
o
g
leN
et
fo
r
F
r
u
it R
ec
o
g
n
itio
n
(
N
u
r
A
z
id
a
Mu
h
a
mma
d
)
471
3.
RE
S
E
ARCH
M
E
T
H
O
D
I
n
th
i
s
s
t
u
d
y
,
M
A
T
L
A
B
2
0
1
8
a
is
u
s
ed
to
p
er
f
o
r
m
t
h
e
ex
p
er
i
m
e
n
ts
.
I
n
o
r
d
er
to
co
m
p
ar
e
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
t
h
r
ee
t
y
p
e
s
o
f
d
ee
p
lear
n
i
n
g
m
o
d
el
s
,
a
s
et
o
f
f
r
u
i
t
i
m
ag
e
s
ar
e
o
b
tain
ed
f
r
o
m
th
e
GitHu
b
[
12
]
w
h
ic
h
is
a
f
r
ee
l
y
a
v
ailab
le
d
ataset
.
T
h
e
d
ataset
co
n
s
i
s
ts
o
f
4
9
0
0
tr
ain
in
g
i
m
ag
e
s
an
d
1
6
4
0
v
alid
atio
n
i
m
ag
e
s
.
I
n
ad
d
itio
n
,
it
is
d
iv
id
ed
i
n
to
9
clas
s
es
o
f
f
r
u
it
i
m
a
g
es
w
h
ich
ar
e
k
i
w
i,
b
an
an
a,
s
tr
a
w
b
er
r
y
,
s
alak
,
p
o
m
e
g
r
an
ate,
p
in
ea
p
p
l
e,
m
a
n
d
ar
in
s
,
d
ates,
l
i
m
es
an
d
c
ar
am
b
u
la.
T
h
e
i
m
ag
e
s
co
n
s
i
s
t
o
f
f
r
a
m
e
s
th
a
t
w
er
e
r
o
tated
b
y
p
o
s
itio
n
.
T
ab
le
1
s
h
o
w
s
th
e
lis
t
o
f
t
h
e
n
u
m
b
er
o
f
i
n
s
ta
n
ce
s
o
f
ea
c
h
cla
s
s
u
s
ed
f
o
r
tr
ai
n
i
n
g
as
w
ell
as
tes
tin
g
p
u
r
p
o
s
es.
F
ig
u
r
e
4
s
h
o
w
s
s
o
m
e
o
f
th
e
s
p
ec
i
m
en
i
m
a
g
es
f
o
r
ea
c
h
c
lass
.
T
h
e
s
ize
o
f
ea
c
h
i
m
a
g
e
is
1
0
0
b
y
1
0
0
p
ix
els.
T
ab
le
1
.
T
h
e
n
u
m
b
er
o
f
i
m
a
g
e
s
f
o
r
tr
ain
i
n
g
a
n
d
v
alid
atio
n
[
1
2
]
N
o
o
f
c
l
a
ss
L
a
b
e
l
N
u
mb
e
r
o
f
T
r
a
i
n
i
n
g
I
mag
e
s
N
u
mb
e
r
o
f
V
a
l
i
d
a
t
i
o
n
I
mag
e
s
1
P
o
me
g
r
a
n
a
t
e
2
4
6
82
2
S
a
l
a
k
4
9
0
1
6
2
3
B
a
n
a
n
a
4
9
0
1
6
6
4
P
i
n
e
a
p
p
l
e
4
9
0
1
6
6
5
M
a
n
d
a
r
i
n
s
4
9
0
1
6
6
6
D
a
t
e
s
4
9
0
1
6
6
7
L
i
m
e
s
4
9
0
1
6
6
8
C
a
r
a
m
b
u
l
a
4
9
0
1
6
6
9
S
t
r
a
w
b
e
r
r
y
4
9
2
1
6
4
Cla
ss
-
1
(Po
m
e
g
r
a
n
a
te
)
Cla
ss
-
6
(Da
te
)
Cla
ss
-
2
(S
a
la
k
)
Cla
ss
-
7
(Li
m
e
)
Cla
ss
-
3
(B
a
n
a
n
a
)
Cla
ss
-
8
(Ca
r
a
m
b
u
l
a
)
Cla
ss
-
4
(Pi
n
e
a
p
p
le)
Cla
ss
-
9
(S
tr
a
w
b
e
r
r
y
)
Cla
ss
-
5
(M
a
n
d
a
r
in
e
)
Fig
u
r
e
4
.
Sp
ec
i
m
en
i
m
a
g
es
f
o
r
ea
ch
t
y
p
e
o
f
f
r
u
it
s
u
s
ed
i
n
th
e
ex
p
er
i
m
e
n
t
4.
RE
SU
L
T
S
A
ND
AN
AL
Y
SI
S
4
.
1
CNN
(
Co
nv
o
lutio
na
l N
eu
ra
l N
et
w
o
rk
)
Fo
r
th
e
ex
p
er
i
m
e
n
t
u
s
in
g
C
N
N,
th
e
s
ize
o
f
th
e
in
p
u
t
i
m
ag
e
is
s
et
to
1
0
0
b
y
1
0
0
b
y
3
p
ix
els
d
u
e
to
th
e
m
e
m
o
r
y
co
n
s
tr
ai
n
t
o
f
th
e
co
m
p
u
ter
u
s
ed
.
T
h
e
i
m
a
g
e
o
n
l
y
d
i
s
p
la
y
s
o
n
e
d
ata
s
et
b
y
id
en
ti
f
y
in
g
w
h
ic
h
ca
teg
o
r
ies
it
is
.
I
f
it
i
s
tr
u
e,
th
en
t
h
e
te
x
t
w
il
l
b
e
d
is
p
la
y
ed
i
n
g
r
ee
n
co
lo
r
,
b
u
t
i
f
it
is
f
al
s
e
,
th
e
te
x
t
i
s
i
n
b
lu
e
co
lo
r
.
I
n
th
is
ca
s
e,
t
h
e
o
u
tp
u
t
is
g
r
ee
n
(
tr
u
e)
,
w
h
ic
h
is
s
tr
a
w
b
er
r
y
.
Fig
u
r
e
5
s
h
o
w
s
th
e
co
d
in
g
f
o
r
t
h
e
ex
ec
u
t
io
n
o
f
C
NN
f
o
r
an
i
m
a
g
e
o
f
a
s
tr
a
w
b
er
r
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l
.
1
2
,
No
.
2
,
No
v
e
m
b
er
201
8
:
4
6
8
–
4
7
5
472
lab
els =
class
i
f
y
(
n
et,
i
m
d
s
_
tes
t)
;
ii =
r
an
d
i(
4
9
0
)
;
i
m
=
i
m
r
ea
d
(
i
m
d
s
_
test
.
F
iles
{i
i})
;
i
m
s
h
o
w
(
i
m
)
;
if
lab
els(ii)
==
i
m
d
s
_
test
.
L
ab
els(ii)
co
lo
r
T
ex
t =
'
g
'
;
else
co
lo
r
T
ex
t =
'
r
';
en
d
titl
e(
ch
ar
(
lab
els(ii)
)
,
'C
o
lo
r
'
,
co
l
o
r
T
ex
t)
;
Fig
u
r
e
5
.
T
h
e
s
am
p
le
co
d
in
g
a
n
d
r
esu
lt
f
o
r
an
i
m
ag
e
o
f
s
tr
a
wb
er
r
y
C
NN
ta
k
es
th
e
r
a
w
co
lo
r
i
m
a
g
e
an
d
t
h
e
f
ea
tu
r
e
s
ar
e
au
to
m
ati
ca
ll
y
e
x
tr
ac
ted
b
y
t
h
e
la
y
er
s
.
A
s
tac
k
o
f
C
NN
co
n
s
i
s
t o
f
co
n
v
o
lv
e
la
y
e
r
,
p
o
o
lin
g
la
y
er
an
d
R
e
L
u
la
y
e
r
w
h
ile
ad
d
itio
n
al
s
tack
o
f
la
y
er
s
ca
n
b
e
ad
d
ed
to
co
m
p
ar
e
t
h
e
p
er
f
o
r
m
an
ce
.
T
h
e
s
iz
e
i
n
co
n
v
o
l
v
e
la
y
er
an
d
th
e
v
al
u
e
o
f
s
tr
id
e
i
n
t
h
e
p
o
o
lin
g
la
y
er
r
ep
r
esen
t t
h
e
n
u
m
b
er
o
f
co
lu
m
n
to
b
e
s
k
ip
p
ed
f
o
r
th
e
s
lid
in
g
w
in
d
o
w
th
at
ca
n
ch
an
g
e
as
th
e
s
e
v
al
u
es
c
an
ef
f
ec
t
th
e
r
es
u
l
t
o
f
th
e
r
ec
o
g
n
itio
n
p
er
f
o
r
m
a
n
c
e.
B
esid
es
th
at,
th
e
v
al
u
es
o
f
ma
xe
p
o
ch
s
r
ep
r
esen
t
th
e
n
u
m
b
er
o
f
iter
atio
n
f
o
r
th
e
tr
ai
n
in
g
p
r
o
ce
s
s
a
n
d
in
i
tia
l
lear
n
i
n
g
r
ate
t
h
at
r
ep
r
ese
n
t
t
h
e
v
a
lu
e
o
f
t
h
e
w
ei
g
h
t
to
b
e
ad
j
u
s
ted
d
u
r
in
g
th
e
tr
ain
i
n
g
p
r
o
ce
s
s
,
ca
n
b
e
ch
a
n
g
ed
to
v
ie
w
t
h
eir
ef
f
ec
t to
th
e
r
ec
o
g
n
itio
n
r
ate.
Nex
t
is
t
h
e
v
alid
atio
n
ac
cu
r
a
c
y
w
h
ic
h
is
1
0
0
%,
th
at
m
a
k
es
th
e
f
in
al
ac
c
u
r
ac
y
is
1
.
T
h
e
ti
m
e
to
d
is
p
la
y
t
h
e
o
u
tp
u
t i
m
a
g
e
o
n
l
y
tak
es 5
s
ec
o
n
d
s
.
T
h
e
tr
ai
n
i
n
g
o
p
tio
n
n
ee
d
to
b
e
s
p
ec
i
f
ied
f
o
r
C
NN.
An
ep
o
ch
i
s
a
f
u
ll
tr
ai
n
i
n
g
c
y
cle
o
f
t
h
e
e
n
tire
d
ataset.
T
h
e
m
ax
i
m
u
m
n
u
m
b
er
o
f
ep
o
ch
s
f
o
r
d
ef
in
e
d
f
o
r
C
NN
i
n
t
h
i
s
ex
p
er
i
m
e
n
t is 1
0
w
it
h
in
i
tial le
ar
n
in
g
r
ate
is
0
.
0
0
1
.
T
h
e
f
r
eq
u
en
c
y
o
f
C
NN
i
s
3
0
iter
atio
n
s
.
Fig
u
r
e
6
s
h
o
w
s
t
h
e
tr
ain
i
n
g
a
n
d
v
alid
atio
n
p
r
o
g
r
e
s
s
es o
f
C
NN.
Fig
u
r
e
6
.
T
h
e
r
esu
lts
o
f
C
NN
4
.
2
.
Alex
Net
A
le
x
Net
i
s
also
ca
lled
as
t
r
an
s
f
er
lear
n
in
g
m
o
d
el,
w
h
ich
is
t
h
e
k
n
o
w
led
g
e
lear
n
t
f
r
o
m
tr
ai
n
in
g
lar
g
e
a
m
o
u
n
t
o
f
d
atab
ase.
Fo
r
th
i
s
ex
p
er
i
m
n
et,
th
e
s
ize
o
f
ea
ch
i
m
ag
e
i
s
2
2
7
b
y
2
2
7
p
ix
els.
T
h
e
im
a
g
e
d
is
p
la
y
s
f
o
u
r
d
ataset
o
f
f
r
u
it
s
w
it
h
th
e
ir
p
r
ed
icted
la
b
els.
Fig
u
r
e
7
s
h
o
w
s
s
o
m
e
s
a
m
p
le
r
esu
lt
s
p
r
o
d
u
ce
d
b
y
A
le
x
n
et
w
h
ic
h
is
s
tr
a
w
b
er
r
y
,
m
an
d
ar
i
n
e,
d
ates a
n
d
li
m
es.
A
le
x
n
et
co
n
s
i
s
ts
o
f
la
y
er
s
tr
a
n
s
f
er
w
it
h
a
f
u
ll
y
co
n
n
ec
ted
l
a
y
er
,
s
o
f
t
m
a
x
la
y
er
a
n
d
a
cla
s
s
i
f
icatio
n
o
u
tp
u
t
la
y
er
;
b
y
s
p
ec
i
f
y
i
n
g
th
e
o
p
tio
n
s
o
f
th
e
n
e
w
f
u
ll
y
co
n
n
ec
ted
to
th
e
n
e
w
d
ata.
B
y
s
p
ec
if
y
in
g
th
e
tr
ai
n
in
g
o
p
tio
n
s
,
tr
an
s
f
er
lear
n
i
n
g
k
ee
p
s
th
e
v
al
u
e
s
o
f
t
h
e
p
ar
a
m
et
er
s
f
r
o
m
t
h
e
p
r
ev
io
u
s
la
y
er
s
o
f
th
e
p
r
etr
ain
ed
n
et
w
o
r
k
.
T
h
e
i
n
it
ial
lear
n
in
g
r
ate
is
s
et
to
a
s
m
all
v
al
u
e
to
s
lo
w
d
o
w
n
t
h
e
tr
an
s
f
er
la
y
er
.
B
esid
es
t
h
at,
t
h
e
v
alu
e
s
o
f
m
ax
i
m
u
m
ep
o
ch
s
t
h
at
r
ep
r
esen
t
th
e
n
u
m
b
er
o
f
iter
atio
n
f
o
r
th
e
tr
ai
n
i
n
g
p
r
o
ce
s
s
an
d
in
itia
l
lear
n
i
n
g
r
ate
th
at
r
ep
r
esen
ts
t
h
e
v
alu
e
o
f
th
e
w
e
ig
h
ts
to
b
e
ad
j
u
s
ted
d
u
r
in
g
tr
ai
n
i
n
g
p
r
o
ce
s
s
is
s
et
to
0
.
0
0
0
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
E
va
lu
a
tio
n
o
f CN
N
,
A
lexn
et
a
n
d
Go
o
g
leN
et
fo
r
F
r
u
it R
ec
o
g
n
itio
n
(
N
u
r
A
z
id
a
Mu
h
a
mma
d
)
473
T
h
e
tr
ain
in
g
p
r
o
ce
s
s
to
o
k
ab
o
u
t
1
7
6
m
i
n
u
tes
an
d
8
s
ec
o
n
d
s
.
I
n
th
is
ex
p
er
i
m
e
n
t,
th
e
m
ax
i
m
u
m
n
u
m
b
er
o
f
ep
o
ch
s
f
o
r
A
lex
n
et
is
6
an
d
th
e
m
a
x
i
m
u
m
n
u
m
b
er
o
f
iter
atio
n
i
s
1
7
4
6
.
Fi
g
u
r
e
8
s
h
o
w
s
th
e
d
etail
in
f
o
r
m
atio
n
o
f
th
e
r
es
u
lt
s
o
f
Alex
n
e
t
w
h
er
e
it a
c
h
iev
e
s
1
0
0
% a
cc
u
r
ac
y
f
o
r
f
r
u
it r
ec
o
g
n
itio
n
.
Fi
g
u
r
e
7
.
T
h
e
r
esu
lts
o
f
A
le
x
n
et
Fig
u
r
e
8
.
T
h
e
d
etail
r
esu
lts
o
f
A
le
x
n
et
4
.
3
.
G
o
o
g
leNe
t
T
h
e
s
ize
o
f
an
i
m
ag
e
in
t
h
e
in
p
u
t
la
y
er
o
f
Go
o
g
len
e
t
is
2
2
4
b
y
2
2
4
.
T
h
e
r
esu
lt
f
o
r
th
e
i
m
a
g
e
is
d
is
p
la
y
ed
w
it
h
t
h
e
p
r
ed
icted
lab
el
(
b
an
an
a)
an
d
p
r
o
b
ab
ilit
ies
w
it
h
t
h
e
la
b
el
w
h
ich
is
9
0
.
6
%.
Fi
g
u
r
e
9
s
h
o
w
s
th
e
r
esu
lt p
r
o
d
u
ce
d
b
y
Go
o
g
le
n
et
w
h
er
e
it d
is
p
la
y
s
t
h
e
n
a
m
e
o
f
th
e
f
r
u
it
w
i
th
t
h
e
p
r
ed
icted
p
r
o
b
a
b
ilit
y
.
Fi
g
u
r
e
9
.
T
h
e
r
esu
lt o
f
Go
o
g
le
n
et
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l
.
1
2
,
No
.
2
,
No
v
e
m
b
er
201
8
:
4
6
8
–
4
7
5
474
Go
o
g
len
et
d
is
p
la
y
s
t
h
e
to
p
f
i
v
e
p
r
ed
icted
lab
el
an
d
th
e
p
r
o
b
ab
ilit
ies.
Fig
u
r
e
1
0
s
h
o
w
s
th
e
to
p
5
p
r
ed
ictio
n
s
an
d
p
r
o
b
ab
ilit
y
o
f
an
i
m
a
g
e
o
f
a
b
a
n
an
a.
T
h
e
v
alid
atio
n
ac
c
u
r
ac
y
i
s
1
0
0
%,
t
h
at
m
a
k
es
t
h
e
f
i
n
al
ac
cu
r
ac
y
as
1
.
T
h
e
tr
ain
i
n
g
ti
m
e
to
o
k
ab
o
u
t
4
8
7
m
i
n
u
tes
a
n
d
4
9
s
ec
o
n
d
s
to
co
m
p
lete
t
h
e
p
r
o
ce
s
s
.
On
th
e
o
th
er
h
an
d
,
it
n
ee
d
s
a
h
ig
h
p
o
w
er
co
m
p
u
ter
o
r
lap
to
p
t
o
co
m
p
lete
th
e
ex
ec
u
tio
n
in
a
r
elativ
el
y
s
h
o
r
t
ti
m
e.
I
n
th
is
e
x
p
er
i
m
e
n
t,
t
h
e
m
a
x
i
m
u
m
n
u
m
b
er
o
f
ep
o
ch
s
f
o
r
Go
o
g
le
n
et
is
6
an
d
th
e
m
a
x
i
m
u
m
n
u
m
b
er
o
f
i
ter
atio
n
is
1746.
Fig
u
r
e
1
1
s
h
o
w
s
t
h
e
d
et
ail
r
esu
lt
s
o
f
Go
o
g
leNe
t.
Fig
u
r
e
1
0
.
T
h
e
to
p
5
p
r
ed
ictio
n
o
f
p
r
o
b
ab
ilit
y
o
f
an
i
m
a
g
e
o
f
a
b
an
an
a
Fig
u
r
e
1
1
.
T
h
e
d
etail
r
esu
lts
o
f
Go
o
g
le
n
et
T
ab
le
2
lis
ts
t
h
e
o
v
er
all
r
es
u
lt
s
o
f
f
r
u
it
r
ec
o
g
n
i
tio
n
u
s
i
n
g
C
NN,
Alex
Ne
t
an
d
Go
o
g
le
n
et.
B
y
lo
o
k
in
g
at
T
ab
le
2
,
w
e
ca
n
s
ee
t
h
at
a
ll
th
e
s
e
th
r
ee
m
o
d
els
p
r
o
d
u
ce
th
e
s
a
m
e
p
er
f
ec
t
ac
c
u
r
ac
y
wh
ich
is
1
.
B
u
t
t
h
e
r
u
n
ti
m
e
r
eq
u
ir
ed
b
y
C
NN
i
s
th
e
lo
w
e
s
t
w
h
ile
Go
o
g
len
e
t
r
eq
u
ir
es
t
h
e
lo
n
g
es
t
ti
m
e.
T
h
is
is
d
u
e
to
t
h
e
ar
ch
itect
u
r
e
o
f
t
h
e
m
o
d
els
wh
er
e
C
N
N
h
as
th
e
s
m
alles
t
n
u
m
b
er
o
f
la
y
er
s
w
h
ile
Go
o
g
l
en
et
h
as
th
e
lar
g
e
s
t
n
u
m
b
er
o
f
la
y
er
s
.
T
ab
le
2
.
T
h
e
p
er
f
o
r
m
an
ce
co
m
p
ar
i
s
o
n
b
et
w
ee
n
C
NN,
A
le
x
n
et
an
d
Go
o
g
le
n
et
C
N
N
A
l
e
x
n
e
t
G
o
o
g
l
e
n
e
t
I
n
p
u
t
si
z
e
1
0
0
1
0
0
3
2
2
7
2
2
7
3
2
2
4
2
2
4
3
I
mag
e
d
i
s
p
l
a
y
1
4
1
Ex
t
r
a
f
e
a
t
u
r
e
s
No
No
D
i
sp
l
a
y
t
o
p
p
r
e
d
i
c
t
i
o
n
A
c
c
u
r
acy
1
1
1
R
u
n
t
i
me
5
se
c
o
n
d
1
7
6
m
i
n
8
se
c
o
n
d
s
4
8
7
m
i
n
4
9
se
c
o
n
d
s
Ep
o
c
h
10
6
6
F
r
e
q
u
e
n
c
y
3
0
i
t
e
r
a
t
i
o
n
3
i
t
e
r
a
t
i
o
n
3
i
t
e
r
a
t
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
E
va
lu
a
tio
n
o
f CN
N
,
A
lexn
et
a
n
d
Go
o
g
leN
et
fo
r
F
r
u
it R
ec
o
g
n
itio
n
(
N
u
r
A
z
id
a
Mu
h
a
mma
d
)
475
5.
CO
NCLU
SI
O
N
I
n
th
is
p
ap
er
,
w
e
ev
a
lu
ate
th
e
r
ec
o
g
n
itio
n
p
er
f
o
r
m
a
n
ce
o
f
C
NN,
Alex
n
et
a
n
d
Go
o
g
le
n
et
f
o
r
n
in
e
d
if
f
er
e
n
t
t
y
p
e
s
o
f
f
r
u
i
ts
.
T
h
e
ex
p
er
i
m
e
n
tal
r
es
u
lt
s
s
h
o
w
th
at
t
h
e
t
h
r
ee
m
o
d
els
p
r
o
d
u
c
e
a
p
er
f
ec
t
1
0
0
%
r
ec
o
g
n
itio
n
ac
cu
r
ac
y
b
u
t
w
ith
d
if
f
er
en
t
r
an
g
e
o
f
r
u
n
ti
m
e.
C
NN
m
o
d
el
s
ee
m
s
to
b
e
th
e
b
est
ch
o
ice
f
o
r
th
i
s
ex
p
er
i
m
e
n
t
s
in
ce
i
t
is
v
er
y
ac
cu
r
ate
an
d
f
ast.
Fo
r
f
u
tu
r
e
wo
r
k
,
w
e
w
i
ll
i
n
v
e
s
ti
g
ate
o
t
h
er
f
r
u
i
t
d
atasets
w
it
h
m
o
r
e
f
r
u
i
t t
y
p
es a
n
d
in
v
o
lv
e
f
r
u
its
i
n
a
b
u
n
ch
.
ACK
NO
WL
E
D
G
E
M
E
NT
T
h
e
au
th
o
r
s
w
o
u
ld
li
k
e
to
t
h
an
k
Fac
u
lt
y
o
f
C
o
m
p
u
ter
an
d
Ma
th
e
m
atica
l
Scien
ce
s
,
Un
i
v
er
s
iti
T
ek
n
o
lo
g
i M
A
R
A
,
S
h
a
h
A
la
m
,
Sela
n
g
o
r
,
f
o
r
s
p
o
n
s
o
r
i
n
g
t
h
is
r
esear
ch
.
RE
F
E
R
E
NC
E
S
[1
]
L
e
c
u
n
,
Y.
Be
n
g
io
,
a
n
d
G
.
Hin
t
o
n
.
(2
0
1
5
)
.
“
De
e
p
lea
rn
in
g
”
,
Na
tu
re
,
v
o
l.
5
2
1
,
n
o
.
7
5
5
3
,
4
3
6
–
4
4
4
.
[2
]
S
tu
a
rt
J.
Ru
ss
e
ll
,
P
e
ter
No
rv
ig
.
(1
9
9
5
).
A
rti
f
icia
l
In
telli
g
e
n
c
e
-
A
M
o
d
e
rn
A
p
p
ro
a
c
h
.
Ne
w
J
e
r
se
y
,
S
i
m
o
n
a
n
d
S
c
h
u
ste
r
C
o
m
p
a
n
y
.
A
rt
icia
lIn
telli
g
e
n
c
e
-
A
M
o
d
e
rn
A
p
p
ro
a
c
h
.
[3
]
G
.
Zh
e
n
g
(2
0
1
7
).
Fru
it
a
n
d
Ve
g
e
ta
b
les
Cla
ss
if
ica
ti
o
n
S
y
ste
m
Us
in
g
Im
a
g
e
S
a
li
e
n
c
y
a
n
d
C
o
n
v
o
lu
ti
o
n
a
l
Ne
u
r
a
l
Ne
two
rk
.
IEE
E
3
rd
.
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
a
n
d
M
e
c
h
a
tro
n
ics
En
g
in
e
e
rin
g
Co
n
f
e
re
n
c
e
.
p
p
.
6
1
3
–
6
1
7
.
[4
]
Ba
ll
e
ste
r,
P
.
,
&
A
ra
u
jo
,
R.
M
.
(
2
0
1
6
).
On
t
h
e
P
e
rf
o
rm
a
n
c
e
o
f
G
o
o
g
LeN
e
t
a
n
d
A
le
x
Ne
t
A
p
p
li
e
d
to
S
k
e
tch
e
s.
P
r
o
c
e
e
d
in
g
s o
f
th
e
3
0
t
h
Co
n
f
e
re
n
c
e
o
n
A
rti
f
icia
l
In
telli
g
e
n
c
e
(
AA
A
I
2
0
1
6
),
1
1
2
4
–
1
1
2
8
.
[5
]
Ja
n
a
,
S
.
,
Ba
sa
k
,
S
.
&
P
a
re
k
h
,
R.
(
2
0
1
7
).
Au
to
ma
ti
c
Fru
it
Rec
o
g
n
i
ti
o
n
fro
m
N
a
tu
r
a
l
Im
a
g
e
s
u
sin
g
Co
lo
r
a
n
d
T
e
x
tu
re
Fea
tu
re
s
.
Co
n
f
e
re
n
c
e
o
n
De
v
ice
s
f
o
r
In
teg
ra
ted
Circu
it
.
2
3
–
2
4
.
[6
]
S
a
b
ri,
N.
,
Ib
ra
h
im
,
Z.
,
S
y
a
h
lan
,
S
.
,
Ja
m
il
,
N.,
&
M
a
n
g
sh
o
r,
N.
N.
A
.
(2
0
1
7
).
Pa
lm
Oil
Fre
sh
Fr
u
it
Bu
n
c
h
Ri
p
e
n
e
ss
Gr
a
d
in
g
I
d
e
n
ti
fi
c
a
ti
o
n
Us
i
n
g
C
o
l
o
r F
e
a
t
u
re
s
.
Jo
u
r
n
a
l
o
f
F
u
n
d
a
m
e
n
tal
a
n
d
A
p
p
li
e
d
S
c
ien
c
e
s
,
9
(4
S
),
5
6
3
-
5
7
9
.
[7
]
Ib
ra
h
im
,
Z.
,
S
a
b
ri,
N.,
&
M
a
n
g
sh
o
r,
N.
N.
A
.
(2
0
1
8
)
.
L
e
a
f
Re
c
o
g
n
it
io
n
u
si
n
g
T
e
x
tu
re
F
e
a
tu
re
s
fo
r
He
rb
a
l
P
lan
t
Id
e
n
ti
f
ica
ti
o
n
.
[8
]
Ib
ra
h
im
,
Z.
,
Ka
sira
n
,
Z
.
,
Isa
,
D.,
&
S
a
b
ri
,
N.
(
2
0
1
6
)
.
M
u
lt
i
-
sc
rip
t
T
e
x
t
De
tec
ti
o
n
a
n
d
Cla
ss
if
ic
a
ti
o
n
fro
m
N
a
tu
r
a
l
S
c
e
n
e
s
.
In
I
n
tern
a
ti
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
o
f
t
Co
m
p
u
ti
n
g
in
Da
ta S
c
i
e
n
c
e
(p
p
.
2
0
0
-
2
1
0
)
.
S
p
rin
g
e
r,
S
in
g
a
p
o
re
.
[9
]
S
h
u
k
la,
D.,
&
De
sa
i,
A
.
(2
0
1
7
).
Rec
o
g
n
it
i
o
n
o
f
fru
it
s
u
sin
g
h
y
b
rid
fea
tu
r
e
s
a
n
d
ma
c
h
in
e
lea
rn
i
n
g
.
I
n
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
C
o
m
p
u
ti
n
g
,
A
n
a
l
y
ti
c
s
a
n
d
S
e
c
u
rit
y
T
re
n
d
s,
CA
S
T
2
0
1
6
,
5
7
2
–
5
7
7
.
h
tt
p
s:/
/d
o
i.
o
rg
/1
0
.
1
1
0
9
/CA
S
T
.
2
0
1
6
.
7
9
1
5
0
3
3
[1
0
]
Ho
u
,
L
.
,
W
u
,
Q.,
S
u
n
,
Q.,
Ya
n
g
,
H.,
&
L
i,
P
.
(2
0
1
6
).
Fru
it
re
c
o
g
n
it
io
n
b
a
se
d
o
n
c
o
n
v
o
lu
ti
o
n
n
e
u
ra
l
n
e
two
rk
.
2
0
1
6
1
2
t
h
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Na
tu
ra
l
Co
m
p
u
tati
o
n
,
F
u
z
z
y
S
y
ste
m
s
a
n
d
Kn
o
w
led
g
e
Disc
o
v
e
r
y
,
ICNC
-
F
S
KD
2
0
1
6
,
1
8
–
2
2
.
h
tt
p
s://
d
o
i.
o
rg
/1
0
.
1
1
0
9
/F
S
KD
.
2
0
1
6
.
7
6
0
3
1
4
4
[1
1
]
Na
sk
a
r,
S
.
(2
0
1
5
).
A
F
ru
it
Re
c
o
g
n
it
io
n
T
e
c
h
n
iq
u
e
u
si
n
g
M
u
l
ti
p
le
F
e
a
tu
re
s
a
n
d
A
rti
f
icia
l
N
e
u
ra
l
Ne
t
w
o
rk
,
In
tern
a
ti
o
n
a
l
Jo
u
rn
a
l
o
f
Co
m
p
u
ter A
p
p
li
c
a
ti
o
n
s.
1
1
6
(2
0
),
2
3
–
2
8
.
[1
2
]
L
i,
P
.
,
L
e
e
,
S
.
H.,
&
Hs
u
,
H.
Y
.
(2
0
1
1
).
Re
v
iew
o
n
f
ru
it
h
a
rv
e
stin
g
m
e
th
o
d
f
o
r
p
o
ten
t
ial
u
se
o
f
a
u
to
m
a
ti
c
f
ru
it
h
a
rv
e
stin
g
s
y
ste
m
s.
P
r
o
c
e
d
ia E
n
g
in
e
e
rin
g
,
2
3
,
3
5
1
-
3
6
6
.
h
tt
p
s://
d
o
i
.
o
rg
/1
0
.
1
0
1
6
/j
.
p
r
o
e
n
g
.
2
0
1
1
.
1
1
.
2
5
1
4
[1
3
]
Du
b
e
y
,
S
.
R.
,
&
Ja
lal,
A
.
S
.
(
2
0
1
2
).
De
tec
ti
o
n
a
n
d
Clas
sif
ica
ti
o
n
o
f
A
p
p
le F
ru
it
Dise
a
se
s
[1
4
]
Us
in
g
Co
m
p
lete
L
o
c
a
l
Bin
a
r
y
P
a
tt
e
rn
s
.
T
h
ird
In
ter
n
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Co
m
p
u
te
r
a
n
d
Co
m
m
u
n
ica
ti
o
n
T
e
c
h
n
o
lo
g
y
,
346
–
3
5
1
.
[1
5
]
S
a
,
I.
,
G
e
,
Z.
,
Da
y
o
u
b
,
F
.
,
Up
c
ro
f
t,
B.
,
P
e
re
z
,
T
.
,
&
M
c
Co
o
l,
C.
(2
0
1
6
).
De
e
p
f
ru
it
s:
A
f
ru
it
d
e
tec
ti
o
n
sy
ste
m
u
sin
g
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
s.
S
e
n
so
rs (S
w
it
z
e
rl
a
n
d
),
1
6
(8
).
[1
6
]
A
.
Kriz
h
e
v
s
k
y
,
I.
S
u
tsk
e
v
e
r
a
n
d
G
.
E.
Hin
to
n
.
(
2
0
1
2
)
.
Im
a
g
e
Ne
t
Clas
sif
ic
a
ti
o
n
w
it
h
De
e
p
Co
n
v
o
l
u
ti
o
n
a
l
Ne
tw
o
rk
s.
A
d
v
a
n
c
e
s in
Ne
u
ra
l
In
f
o
rm
a
ti
o
n
P
r
o
c
e
ss
in
g
S
y
ste
m
s 2
5
,
2
0
1
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.