I
AE
S
I
n
t
e
r
n
at
ion
al
Jou
r
n
al
of
Ar
t
if
icial
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
Vol.
14
,
No.
5
,
Oc
tober
20
25
,
pp
.
3734
~
3743
I
S
S
N:
2252
-
8938
,
DO
I
:
10
.
11591/i
jai
.
v
14
.i
5
.
pp
37
34
-
3743
3734
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
ai
.
iaes
c
or
e
.
c
om
C
om
p
a
r
is
on
of
HS
V
-
c
ol
o
r
an
d
A
N
N
-
HS
V
-
c
ol
or
s
e
gm
e
n
t
a
t
io
n
f
or
d
e
t
e
c
t
in
g soyb
e
an
a
d
u
lt
e
r
at
io
n
F
ar
id
Ra
k
h
m
at
Abad
i
1
,
2
,
Ru
d
iat
i
E
vi
M
as
it
h
oh
1
,
L
il
ik
S
u
t
iars
o
1
,
S
r
i
Rahayoe
1
1
D
e
pa
r
tm
e
nt
of
A
gr
ic
ul
tu
r
a
l
a
nd B
io
s
ys
te
m
s
E
ngi
ne
e
r
in
g, F
a
c
ul
ty
of
A
gr
ic
ul
tu
r
a
l
T
e
c
hnol
ogy, Unive
r
s
it
a
s
G
a
dj
a
h M
a
da
,
Y
og
ya
ka
r
ta
,
I
ndon
e
s
ia
2
B
e
a
n
C
r
op A
s
s
e
mbl
y a
nd T
e
s
ti
ng C
e
nt
e
r
(
B
a
la
i
P
e
r
a
ki
ta
n da
n
P
e
nguj
ia
n T
a
na
ma
n A
n
e
ka
K
a
c
a
ng)
,
M
a
la
ng, I
ndone
s
ia
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
S
e
p
17,
2024
R
e
vis
e
d
J
ul
10,
2025
Ac
c
e
pted
Aug
6,
2025
So
y
b
e
an
s
are
an
i
mp
o
rt
a
n
t
fo
o
d
cro
p
,
b
u
t
t
h
ei
r
q
u
al
i
t
y
i
s
o
ft
en
co
m
p
ro
m
i
s
e
d
b
y
c
o
n
t
ami
n
at
i
o
n
w
i
t
h
o
t
h
er
ma
t
eri
a
l
s
,
a
p
ro
ce
s
s
k
n
o
w
n
as
a
d
u
l
t
era
t
i
o
n
.
Co
n
v
en
t
i
o
n
a
l
met
h
o
d
s
f
o
r
d
e
t
ect
i
n
g
ad
u
l
t
erat
i
o
n
are
s
l
o
w
;
t
h
eref
o
re,
t
h
ere
i
s
a
n
eed
fo
r
rap
i
d
an
d
n
o
n
-
i
n
v
a
s
i
v
e
al
t
ern
a
t
i
v
es
.
T
h
i
s
s
t
u
d
y
ai
med
t
o
as
s
e
s
s
t
h
e
cap
ab
i
l
i
t
y
o
f
hue
-
s
at
u
rat
i
o
n
-
v
a
l
u
e
(
H
SV
)
c
o
l
o
r
s
e
g
men
t
at
i
o
n
an
d
i
t
s
co
mb
i
n
a
t
i
o
n
w
i
t
h
art
i
f
i
ci
a
l
n
eu
ra
l
n
e
t
w
o
rk
s
(A
N
N
)
t
o
i
d
en
t
i
fy
ad
u
l
t
erat
i
o
n
i
n
s
o
y
b
ea
n
s
am
p
l
e
s
.
T
h
i
s
res
earc
h
emp
l
o
y
e
d
i
ma
g
e
p
r
o
c
es
s
i
n
g
an
d
mach
i
n
e
l
earn
i
n
g
t
o
s
e
g
men
t
s
o
y
b
ean
s
m
i
x
e
d
w
i
t
h
ad
u
l
t
eran
t
s
at
co
n
cen
t
rat
i
o
n
s
o
f
5
%
,
1
0
%
,
1
5
%
,
2
0
%
,
an
d
2
5
%
.
T
h
e
H
SV
met
h
o
d
s
u
cce
s
s
fu
l
l
y
d
i
s
t
i
n
g
u
i
s
h
e
d
s
o
y
b
ea
n
s
an
d
o
t
h
er
mat
er
i
al
s
,
b
u
t
s
o
me
ch
al
l
en
g
es
w
ere
o
b
s
erv
e
d
i
n
s
h
ad
o
w
reg
i
o
n
s
an
d
area
s
w
i
t
h
s
i
m
i
l
ar
co
l
o
rs
.
T
h
e
H
S
V
-
A
N
N
m
o
d
e
l
w
i
t
h
s
i
x
h
i
d
d
en
l
ay
er
s
p
erfo
rm
ed
w
e
l
l
w
i
t
h
a
ca
l
i
b
rat
i
o
n
accu
rac
y
o
f
R²
v
al
u
e
o
f
0
.
9
7
a
n
d
ro
o
t
-
mean
-
s
q
u
are
erro
r
(RMSE
)
o
f
2
.
1
6
%
,
w
h
i
ch
p
r
o
v
i
d
ed
mo
re
d
e
t
ai
l
ed
s
eg
me
n
t
a
t
i
o
n
,
al
t
h
o
u
g
h
i
t
s
t
i
l
l
h
ad
s
o
me
p
ro
b
l
em
s
i
n
s
h
a
d
o
w
reg
i
o
n
s
an
d
u
n
d
et
ec
t
ed
co
r
n
emb
ry
o
p
ar
t
s
.
T
h
e
v
al
i
d
a
t
i
o
n
res
u
l
t
s
i
n
d
i
ca
t
ed
t
h
at
t
h
e
H
SV
mo
d
e
l
h
ad
an
R²
v
al
u
e
o
f
0
.
9
8
an
d
RMSE
o
f
4
.
4
8
%
,
w
h
i
l
e
t
h
e
H
S
V
-
A
N
N
m
o
d
e
l
h
ad
an
R²
v
al
u
e
o
f
0
.
9
6
an
d
RMSE
o
f
1
.
3
%
.
Bo
t
h
mo
d
el
s
w
ere
cap
ab
l
e
o
f
p
red
i
ct
i
n
g
t
h
e
l
ev
e
l
s
o
f
ad
u
l
t
erat
i
o
n
,
an
d
t
h
e
H
SV
-
A
N
N
mo
d
el
p
ro
v
ed
t
o
b
e
mo
re
accu
rat
e.
It
i
s
c
o
n
c
l
u
d
ed
t
h
at
b
o
t
h
met
h
o
d
s
are
effi
ci
en
t
;
h
o
w
ev
er,
t
h
ere
i
s
a
n
eed
fo
r
mo
re
w
o
rk
o
n
m
o
d
el
i
n
g
an
d
s
amp
l
i
n
g
t
o
i
n
crea
s
e
t
h
e
s
eg
me
n
t
a
t
i
o
n
p
rec
i
s
i
o
n
an
d
d
ecrea
s
e
t
h
e
b
i
a
s
es
,
es
p
ec
i
al
l
y
i
n
t
h
e
s
h
a
d
o
w
an
d
o
v
er
l
ap
p
ed
co
l
o
r.
K
e
y
w
o
r
d
s
:
Ar
ti
f
icia
l
ne
ur
a
l
ne
twor
k
HSV
s
e
gmenta
ti
on
I
mage
pr
oc
e
s
s
ing
Non
-
de
s
tr
uc
ti
ve
tes
ti
ng
S
oybe
a
n
a
dult
e
r
a
ti
on
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e.
C
or
r
e
s
pon
din
g
A
u
th
or
:
R
udiati
E
vi
M
a
s
it
hoh
De
pa
r
tm
e
nt
of
Ag
r
icultur
a
l
a
nd
B
ios
ys
tems
E
nginee
r
ing,
F
a
c
ult
y
of
Ag
r
icultur
a
l
T
e
c
hnology
Unive
r
s
it
a
s
Ga
djah
M
a
da
Yogya
ka
r
ta
5528
1
,
I
ndone
s
ia
E
mail:
e
vi@ugm
.
a
c
.
id
1.
I
NT
RODU
C
T
I
ON
S
oybe
a
ns
a
r
e
a
vit
a
l
s
taple
f
ood
a
nd
e
c
onomi
c
c
o
mm
odit
y
in
As
ia,
whe
r
e
they
a
r
e
wide
ly
c
ons
umed
in
va
r
ious
f
or
ms
,
s
uc
h
a
s
tempe
h,
tof
u,
s
oy
s
a
uc
e
,
s
oy
mi
lk,
a
nd
li
ve
s
tock
f
e
e
d
[
1
]
.
W
it
h
the
lar
ge
de
mand
f
or
s
oybe
a
ns
f
or
c
ons
umpt
ion,
e
ns
ur
ing
the
qua
li
ty
a
nd
pur
it
y
of
s
oybe
a
ns
dur
ing
pr
oduc
t
ion,
s
tor
a
ge
,
a
nd
dis
tr
ibut
ion
ha
s
be
c
ome
incr
e
a
s
ingl
y
c
r
it
ica
l.
How
e
ve
r
,
s
oybe
a
n
c
omm
odit
ies
s
ometim
e
s
e
x
pe
r
ienc
e
a
dult
e
r
a
ti
on,
whic
h
is
int
e
nt
ional
o
r
un
int
e
nti
ona
l
mi
xing
with
othe
r
mate
r
ials
s
uc
h
a
s
c
or
n
,
gr
e
e
n
b
e
a
ns
,
or
e
ve
n
im
pur
it
ies
li
ke
s
a
nd
dur
ing
pos
t
-
ha
r
ve
s
t
a
nd
dis
tr
ibut
ion
pr
oc
e
s
s
e
s
.
T
his
a
dult
e
r
a
ti
on
not
only
dim
ini
s
he
s
the
qua
li
ty
o
f
s
oybe
a
ns
but
a
ls
o
s
igni
f
ica
ntl
y
im
pa
c
ts
their
ma
r
ke
t
va
lue,
pos
ing
c
ha
ll
e
nge
s
to
c
ons
umer
s
a
nd
pr
oduc
e
r
s
[
2]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
C
ompar
is
on
of
HSV
-
c
olor
and
A
N
N
-
HSV
-
c
olor
s
e
gme
ntat
ion
for
de
tec
ti
ng
…
(
F
ar
id
R
ahmat
A
badi
)
3735
E
f
f
e
c
ti
ve
methods
f
or
d
e
tec
ti
ng
a
dult
e
r
a
ti
on
a
r
e
e
s
s
e
nti
a
l
to
a
ddr
e
s
s
thi
s
is
s
ue
.
T
r
a
dit
ional
a
ppr
oa
c
he
s
,
s
uc
h
a
s
vis
ua
l
ins
pe
c
ti
on
or
manua
l
gr
a
ding,
a
r
e
o
f
ten
ti
me
-
c
ons
umi
ng
a
nd
pr
one
to
hum
a
n
e
r
r
or
.
Adulter
a
ti
on
may
a
lt
e
r
na
te
qua
li
ty,
including
f
lav
or
,
whic
h
c
a
n
be
de
tec
ted
u
s
ing
s
e
ns
or
y
or
e
lec
tr
o
nic
nos
e
[
3]
.
I
n
r
e
c
e
nt
ye
a
r
s
,
non
-
de
s
tr
uc
ti
ve
tec
hniques
f
or
de
tec
ti
ng
qua
l
it
y,
including
a
dult
e
r
a
ti
on
in
f
o
od,
ha
ve
be
e
n
incr
e
a
s
ing,
whic
h
include
the
us
e
of
s
pe
c
tr
os
c
opy
[
4]
a
nd
c
omput
e
r
vis
ion
[
5
]
.
Ne
a
r
inf
r
a
r
e
d
s
pe
c
tr
os
c
opy
(
NI
R
S
)
wa
s
us
e
d
to
de
tec
t
s
oybe
a
n
a
dult
e
r
a
ti
on
[
6]
,
p
r
e
dict
s
oybe
a
n
c
he
mi
c
a
ls
[
7]
,
[
8]
,
a
nd
s
oybe
a
n
c
olor
c
las
s
if
ica
ti
on
[
9]
.
Although
a
c
c
ur
a
t
e
,
NI
R
S
ha
s
li
mi
tations
,
s
uc
h
a
s
the
r
e
latively
hig
h
pr
ice
of
ins
tr
uments
,
whic
h
make
s
them
una
f
f
or
da
ble
f
or
s
mall
indus
tr
ies
.
T
he
r
e
f
or
e
,
f
indi
ng
c
he
a
pe
r
a
nd
mor
e
a
f
f
or
da
ble
methods
f
o
r
de
tec
ti
ng
s
oybe
a
n
a
dult
e
r
a
ti
on
is
im
por
tant
.
Adva
nc
e
ments
in
im
a
ge
p
r
oc
e
s
s
ing
a
nd
mac
hine
lea
r
ning
tec
hnologi
e
s
ha
ve
p
r
ovided
p
r
omi
s
ing
a
lt
e
r
na
ti
ve
s
f
or
non
-
de
s
tr
uc
ti
ve
a
nd
e
f
f
icie
nt
q
ua
li
ty
a
s
s
e
s
s
ment
[
10]
.
F
or
ins
tanc
e
,
im
a
ge
pr
oc
e
s
s
ing
tec
hniques
ha
ve
be
e
n
s
uc
c
e
s
s
f
ull
y
e
mpl
oye
d
to
g
r
a
de
s
oybe
a
n
qua
li
ty
[
11
]
a
nd
to
de
tec
t
s
oybe
a
n
da
mage
[
12]
.
T
o
im
p
r
ove
model
a
c
c
ur
a
c
y
in
de
tec
ti
on
o
r
c
l
a
s
s
if
ica
ti
on,
mac
hine
lea
r
ning
ha
s
be
e
n
e
mpl
oye
d
[
13]
,
f
or
ins
tanc
e
,
de
e
p
ne
ur
a
l
ne
twor
ks
(
NN
)
us
e
d
to
de
t
e
c
t
a
dult
e
r
a
ti
on
in
S
or
ghum
[
14]
or
c
onvolut
ion
a
l
ne
ur
a
l
ne
twor
ks
us
e
d
to
de
tec
t
a
dult
e
r
a
ti
on
in
f
ood
[
15]
.
M
or
e
ove
r
,
s
oybe
a
n
c
ult
ivar
s
we
r
e
c
las
s
if
ied
us
ing
a
r
ti
f
icia
l
ne
ur
a
l
ne
twor
k
s
(
AN
N)
[
16]
.
T
he
im
a
ge
pr
oc
e
s
s
ing
tec
hnique
us
ua
ll
y
e
mpl
oys
c
olor
pa
r
a
mete
r
s
s
tor
e
d
in
a
n
im
a
ge
in
d
ig
it
a
l
da
ta
in
matr
ix
c
omponents
or
c
ha
nne
ls
s
uc
h
a
s
r
e
d
-
gr
e
e
n
-
blue
(
R
GB
)
or
hue
-
s
a
tur
a
ti
on
-
va
lue
(
HSV)
[
17]
.
Hue
(
H)
indi
c
a
tes
the
main
c
olor
s
,
s
uc
h
a
s
r
e
d,
or
a
nge
,
gr
e
e
n,
with
0~360
°
mea
s
ur
e
,
s
a
tur
a
ti
on
(
S
)
indi
c
a
ti
ng
the
de
pth
of
c
olo
r
,
f
or
e
xa
mpl
e
,
da
r
k
r
e
d
a
nd
l
ight
r
e
d,
mea
s
ur
e
d
in
pe
r
c
e
ntage
f
r
om
0%
to
f
u
ll
y
s
a
tur
a
te
d
100%
,
a
nd
va
lue
(
V)
indi
c
a
tes
the
de
gr
e
e
o
f
li
ght
a
nd
da
r
k
c
olor
,
us
ua
ll
y
mea
s
ur
e
d
in
pe
r
c
e
ntage
f
r
om
blac
k
0%
to
white
100%
[
18
]
.
I
n
HSV
c
olo
r
s
pa
c
e
,
va
lue
(
V)
i
s
the
a
ve
r
a
ge
of
R
GB
s
ignals
[
19]
.
I
n
the
im
a
ge
pr
oc
e
s
s
ing
method,
c
olor
pa
r
a
mete
r
s
c
a
n
be
us
e
d
f
or
s
e
gment
a
ti
on.
S
e
gmenta
ti
on
is
a
tec
hnique
f
o
r
s
e
pa
r
a
ti
ng
da
ta
in
digi
tal
i
mage
s
int
o
s
e
ve
r
a
l
pa
r
ts
o
r
s
e
gments
,
us
ua
ll
y
us
e
d
to
s
e
pa
r
a
te
the
ba
c
kgr
ound
f
r
om
the
o
bs
e
r
ve
d
ob
jec
t
,
e
na
bli
ng
pr
e
c
is
e
a
na
lys
is
of
a
dult
e
r
a
ti
on
or
c
ontamination.
T
he
HSV
c
olor
model
is
e
quival
e
nt
to
human
thi
nking,
making
HSV
a
n
idea
l
c
hoice
f
or
im
a
ge
s
e
gmenta
ti
on
[
20]
.
S
e
ve
r
a
l
r
e
s
e
a
r
c
he
r
s
we
r
e
a
ble
to
us
e
f
or
f
ood
a
ppli
c
a
ti
ons
,
s
uc
h
a
s
in
o
li
ve
oil
[
21
]
,
c
oc
onut
oil
[
22
]
,
or
be
e
f
[
23
]
.
I
mage
s
e
gmenta
ti
on
in
f
ood
is
mor
e
c
ompl
e
x
a
s
it
a
im
s
to
r
e
c
ognize
e
a
c
h
ing
r
e
dient
c
a
tegor
y
a
s
we
ll
a
s
it
s
pixel
-
wis
e
loca
ti
ons
in
the
f
ood
i
mage
[
24]
.
De
e
p
lea
r
ning
a
r
e
a
ble
to
lea
r
n
c
ompl
e
x
f
e
a
tu
r
e
s
f
r
om
uns
tr
uc
tur
e
d
da
ta
e
na
ble
c
omput
e
r
s
to
make
inf
or
mative
de
c
is
ions
ba
s
e
d
on
r
a
w
da
ta;
thus
r
e
s
e
a
r
c
h
e
r
s
ha
ve
us
e
d
AN
N
to
e
xtr
a
c
t
a
nd
lea
r
n
c
ompl
e
x
inf
or
m
a
ti
on
[
25]
.
W
he
n
c
ombi
ne
d
with
HSV
c
olor
pa
r
a
mete
r
s
,
AN
Ns
of
f
e
r
a
r
obus
t
a
ppr
oa
c
h
to
r
e
c
ognizing
pa
tt
e
r
ns
a
nd
making
p
r
e
dictions
[
26]
.
E
ve
n
though
the
ne
e
d
f
or
pr
e
c
is
e
a
nd
r
e
a
s
ona
bly
pr
ice
d
wa
ys
to
identi
f
y
a
dult
e
r
a
ted
s
oybe
a
ns
is
gr
owing,
c
onve
nti
ona
l
methods
a
r
e
s
ti
ll
inef
f
e
c
ti
v
e
,
a
nd
s
ophis
ti
c
a
ted
ins
tr
uments
li
ke
NI
R
s
pe
c
tr
os
c
opy
a
r
e
f
r
e
que
ntl
y
too
e
xpe
ns
ive
f
o
r
s
mall
bus
ines
s
e
s
.
F
e
w
s
tudi
e
s
c
ompar
e
the
e
f
f
ica
c
y
of
ba
s
ic
c
olor
-
ba
s
e
d
im
a
ge
s
e
gmenta
ti
on
(
HSV)
with
mor
e
s
ophis
ti
c
a
ted
methods
,
li
ke
HSV
in
c
onjunction
with
AN
N
,
in
ide
nti
f
ying
a
dult
e
r
a
nts
in
s
oybe
a
ns
.
T
he
r
e
f
or
e
,
thi
s
s
tudy
a
i
med
a
t
e
va
luating
the
potential
of
HSV
c
olor
i
t
s
e
lf
a
nd
HSV
-
AN
N
c
ombi
na
ti
on
methods
f
or
de
tec
ti
ng
a
d
ult
e
r
a
ti
on
in
s
oybe
a
n
s
a
mpl
e
s
.
B
y
de
tec
ti
ng
the
mi
xtur
e
s
of
s
oybe
a
n
a
nd
c
or
n,
mungbea
ns
,
a
nd
s
a
nd
a
t
va
r
yi
ng
c
onc
e
ntr
a
ti
ons
,
the
r
e
s
e
a
r
c
h
c
a
n
be
us
e
d
to
d
e
ve
lop
a
non
-
de
s
tr
uc
ti
ve
,
e
f
f
icie
nt,
a
nd
a
c
c
ur
a
te
method
f
or
identif
ying
a
dult
e
r
a
nts
in
s
oybe
a
n.
2.
M
E
T
HO
D
2.
1.
M
at
e
r
ial
s
T
he
mate
r
ials
c
ons
is
t
of
s
oybe
a
n
va
r
ieties
,
na
m
e
ly
Gr
oboga
n,
De
von
2,
De
tap
1
,
De
r
a
p
1
,
a
nd
De
ja
2,
ob
taine
d
f
r
om
the
M
a
lang
R
e
ge
nc
y
E
a
s
t
J
a
va
f
r
om
the
2
nd
p
lanting
s
e
a
s
on
of
2022.
T
hi
r
ty
gr
a
ms
of
whole
s
oybe
a
ns
a
nd
a
dult
e
r
a
nts
we
r
e
we
ighed
a
n
d
plac
e
d
in
a
blac
k
c
e
r
a
mi
c
c
up.
T
he
a
dult
e
r
a
nts
c
ons
is
t
of
c
or
n,
mungbea
n,
a
nd
s
a
nd
with
c
onc
e
ntr
a
ti
ons
of
5%
,
10%
,
15%
,
20%
,
a
nd
25
%
.
T
he
mi
x
tur
e
of
s
oybe
a
n
s
a
mpl
e
a
nd
a
dult
e
r
a
nts
we
r
e
plac
e
d
e
ve
nly
on
the
c
up,
s
o
the
ba
c
kgr
ound
wa
s
c
ove
r
e
d
a
t
0
.
8
c
m
thi
c
kne
s
s
.
2.
2
.
I
m
age
ac
q
u
is
it
ion
T
he
e
quipm
e
nt
us
e
d
in
thi
s
r
e
s
e
a
r
c
h
include
d
c
e
r
a
mi
c
c
ups
f
o
r
plac
ing
s
a
mpl
e
s
with
a
diame
ter
o
f
8.
5
c
m
a
nd
a
photo
box
32
×
32
×
32
c
m
3
e
quipped
with
3
-
wa
tt
L
E
D
li
ghti
ng.
A
c
e
ll
phone
c
a
mer
a
wa
s
s
us
e
d
to
c
a
ptur
e
im
a
ge
da
ta
(
S
G
-
A10)
with
13
M
P
r
e
s
olut
ion
(
4128
×
3096
maximum
pixels
)
a
nd
C
M
OS
s
e
ns
or
type.
T
he
c
a
mer
a
wa
s
he
ld
with
a
gr
ip
o
r
s
tanc
e
to
s
tabili
z
e
the
c
a
mer
a
’
s
pos
it
ion.
A
s
mar
t
s
e
ns
or
AS8
03
digi
tal
lux
mete
r
mea
s
ur
e
d
li
ght
int
e
ns
it
y
a
nd
r
oom
tempe
r
a
tur
e
.
I
mage
da
ta
wa
s
take
n
in
the
mi
ddle
pos
it
ion
of
the
photo
box
a
t
a
dis
tanc
e
(
x
)
o
f
20
c
m
with
the
i
m
a
ge
s
hooti
ng
dir
e
c
ti
on
ve
r
ti
c
a
ll
y
downw
a
r
ds
(
90
°
)
.
I
mage
s
we
r
e
take
n
a
t
a
n
a
ve
r
a
ge
r
oom
li
ght
int
e
ns
it
y
leve
l
of
95
lux
a
t
25
°
C
.
T
he
il
lus
tr
a
ti
on
of
im
a
ge
a
c
qui
s
it
ion
is
s
hown
in
F
igu
r
e
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
5
,
Oc
tober
20
25
:
37
34
-
3743
3736
2.
3.
I
m
age
p
r
oc
e
s
s
in
g
an
d
an
alys
is
T
he
im
a
ge
s
we
r
e
a
na
lyze
d
us
ing
P
ython
3.
12
.
1,
w
it
h
the
int
e
gr
a
ted
de
ve
lopm
e
nt
e
nvi
r
onment
(
I
D
E
)
of
vis
ua
l
c
ode
(
VSC
ode
)
.
Da
ta
p
r
e
-
pr
oc
e
s
s
ing
wa
s
c
a
r
r
ied
out
ini
ti
a
ll
y
by
de
ter
mi
ning
the
r
e
gion
o
f
int
e
r
e
s
t
(
R
OI
)
,
wh
e
r
e
the
im
a
ge
da
ta
wa
s
c
r
oppe
d
in
s
qua
r
e
a
t
1000
×
1000
pixels
,
whic
h
wa
s
then
r
e
duc
e
d
to
500
×
500
pixels
.
T
he
im
a
ge
dim
e
ns
ions
we
r
e
r
e
du
c
e
d
to
s
pe
e
d
up
pr
ogr
a
mm
ing
c
omput
a
ti
on
by
li
mi
ti
ng
the
wor
king
a
r
e
a
or
the
R
OI
.
T
o
obtain
the
HSV
input
s
,
s
a
mpl
ing
wa
s
c
a
r
r
ied
out
us
ing
GI
M
P
2.
10.
30
s
of
twa
r
e
.
Uppe
r
a
nd
lowe
r
HSV
pa
r
a
mete
r
s
we
r
e
s
a
mpl
e
d
f
r
om
200
s
a
mpl
e
s
us
ing
a
pu
r
pos
ive
s
a
mpl
ing
met
hod
f
r
om
im
a
ge
s
of
c
or
n
,
mungbea
n,
a
nd
s
a
nd.
T
he
HSV
pa
r
a
mete
r
va
lue
in
the
Ope
nC
V
ve
r
s
ion
wa
s
obtaine
d
by
the
f
or
mul
a
s
hown
in
(
1
)
-
(
3
)
.
F
igur
e
1
s
hows
s
c
he
matic
of
im
a
ge
a
c
quis
it
ion
a
nd
da
ta
e
xtr
a
c
ti
on
.
H
=
1
2
(
H
GI
M
P
)
(
1)
S
=
(
100
)
×
255
(
2)
V
=
(
V
G
I
M
P
100
)
×
255
(
3)
W
he
r
e
H
G
IM
P
,
S
G
IM
P
,
a
nd
V
G
IM
P
a
r
e
the
HSV
pa
r
a
m
e
ter
va
lue
in
G
I
M
P
s
of
twa
r
e
.
F
igur
e
1.
I
l
lus
tr
a
ti
on
of
im
a
ge
a
c
quis
it
ion
a
nd
da
ta
e
xtr
a
c
ti
on
T
he
c
olor
of
s
oybe
a
n
a
nd
other
im
pu
r
it
ies
wa
s
d
e
ter
mi
ne
d
us
ing
HSV
c
olor
im
a
ge
s
e
gmenta
ti
on.
T
he
c
v2.
inr
a
nge
f
unc
ti
on
pe
r
f
or
ms
th
r
e
s
holdi
ng,
p
r
oduc
ing
a
gr
a
ys
c
a
le
s
e
gmente
d
im
a
ge
,
whic
h
f
un
c
ti
ons
a
s
a
mas
k
f
or
c
e
r
tain
de
s
ir
e
d
pa
r
ts
f
r
om
the
im
a
ge
.
M
e
a
nwhile,
in
the
AN
N
-
HSV
model,
the
HSV
f
e
a
tur
e
s
that
ha
ve
be
e
n
e
xtr
a
c
ted
we
r
e
then
a
r
r
a
nge
d
in
a
n
A
NN
input
da
ta
mat
r
ix
c
ons
is
ti
ng
of
indepe
nde
nt
va
r
iable
s
(X
ij
)
a
nd
de
pe
nde
nt
va
r
iable
s
(
Y
i
)
.
Af
ter
be
ing
modele
d,
pr
e
dictions
we
r
e
made
by
pe
r
f
or
mi
ng
a
matr
ix
tr
a
ns
f
or
mation
ba
c
k
to
the
or
igi
na
l
im
a
ge
s
ize
wit
h
the
“
np”
r
e
s
ha
pe
c
omm
a
nd.
T
he
p
r
e
diction
da
ta
ha
s
unde
r
gone
a
th
r
e
s
holdi
ng
pr
oc
e
s
s
that
c
las
s
if
ies
dir
t
a
nd
s
oybe
a
ns
int
o
binar
y
va
lues
(
0
a
nd
1)
.
As
a
r
e
s
ult
,
the
dis
playe
d
output
a
ppe
a
r
s
a
s
a
s
e
gmente
d
im
a
ge
.
H
S
V
mo
de
li
ng
wa
s
c
a
r
r
i
e
d
ou
t
by
me
a
s
u
r
i
ng
ne
w
im
a
g
e
d
a
t
a
s
a
mp
le
s
r
a
n
do
ml
y
di
vi
d
e
d
i
nt
o
t
r
a
in
in
g
a
nd
t
e
s
t
da
t
a
(
F
i
gur
e
2
)
.
T
h
e
d
a
t
a
s
i
z
e
w
a
s
10
00,
div
id
e
d
in
to
70%
t
r
a
in
in
g
a
nd
30%
te
s
t
d
a
ta.
N
e
xt,
th
e
mo
de
l
wa
s
f
i
tt
e
d
u
s
i
ng
tr
a
i
ni
ng
a
nd
te
s
t
da
ta
to
pr
od
u
c
e
t
he
d
e
s
ir
e
d
pr
e
di
c
ti
on
e
q
u
a
ti
on.
S
e
gm
e
nt
a
ti
o
n
w
a
s
d
on
e
b
y
r
e
d
uc
in
g
th
e
s
ha
do
w
a
s
pe
c
t
a
nd
s
l
ic
in
g
t
h
e
s
a
m
e
c
o
lor
,
e
s
p
e
c
ia
ll
y
i
n
p
a
r
t
s
of
t
h
e
c
or
n
e
m
br
y
o
t
ha
t
h
a
v
e
s
i
mi
l
a
r
it
i
e
s
t
o
s
o
ybe
a
n
s
.
S
o,
th
e
e
mb
r
y
o
p
a
r
t
wa
s
de
te
r
mi
n
e
d
ba
s
e
d
on
a
c
a
lc
ul
a
t
io
n
of
20%
o
f
t
he
e
n
ti
r
e
c
or
n
[
2
7]
.
M
o
de
l
v
a
li
d
a
ti
o
n
w
a
s
c
a
r
r
i
e
d
out
u
s
i
ng
10
0
n
e
w
s
a
m
pl
e
s
of
p
ur
e
a
n
d
a
d
u
lt
e
r
a
t
e
d
s
oy
be
a
n
d
a
t
a
.
As
s
hown
in
F
igur
e
2
,
the
AN
N
wa
s
us
e
d
to
p
r
e
dict
a
dult
e
r
a
ti
on
[
28
]
e
mpl
oying
HSV
pa
r
a
mete
r
da
ta
a
s
input
.
T
he
da
ta
wa
s
divi
de
d
r
a
ndoml
y
int
o
tr
a
ini
ng
a
nd
tes
t
da
ta
us
e
d
to
buil
d
the
model.
T
he
AN
N
method
us
e
d
the
mul
ti
laye
r
pe
r
c
e
ptr
on
(
M
L
P
)
with
s
e
ve
r
a
l
hidden
laye
r
s
a
s
input
.
T
he
input
da
ta
wa
s
e
nter
e
d
a
s
the
input
laye
r
with
the
a
mount
of
da
ta
a
s
node
s
,
while
the
bi
na
r
y
output
da
ta
wa
s
a
s
the
output
l
a
ye
r
.
T
he
AN
N
a
r
c
hit
e
c
tur
e
is
i
ll
us
tr
a
ted
in
F
igu
r
e
3.
T
he
i
nc
omi
ng
input
da
ta
wa
s
given
a
r
a
ndom
we
ight
in
g,
whic
h
wa
s
then
e
nter
e
d
int
o
the
tr
a
ns
f
e
r
f
unc
ti
on
,
whe
r
e
bias
pa
r
a
mete
r
s
we
r
e
a
dde
d.
C
a
lcula
ti
ons
we
r
e
c
a
r
r
ied
out
r
e
pe
a
tedly
on
the
hidden
laye
r
a
nd
a
t
e
a
c
h
node
.
Ulti
mate
ly,
the
input
-
output
will
be
pr
oduc
e
d
i
n
binar
y
thr
ough
the
s
igm
oid
a
c
ti
va
ti
on
f
unc
ti
on.
AN
N
mo
de
li
ng
wa
s
c
a
r
r
ied
out
us
ing
P
ython
3
.
1
p
r
ogr
a
mm
ing
with
the
T
e
ns
or
F
low
li
br
a
r
y
.
Ne
xt
,
the
r
e
s
ult
ing
model
wa
s
s
tor
e
d
in
c
omput
e
r
memor
y.
Va
li
da
ti
on
wa
s
c
a
r
r
ied
out
us
ing
ne
w
s
a
mpl
e
s
a
nd
c
a
lcula
ted
model
pa
r
a
mete
r
s
in
R
2
a
nd
R
M
S
E
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
C
ompar
is
on
of
HSV
-
c
olor
and
A
N
N
-
HSV
-
c
olor
s
e
gme
ntat
ion
for
de
tec
ti
ng
…
(
F
ar
id
R
ahmat
A
badi
)
3737
F
igur
e
2.
De
ter
mi
na
ti
on
of
a
dult
e
r
a
ti
on
in
s
oybe
a
n
us
ing
HSV
model
a
nd
HSV
-
AN
N
model
F
igur
e
3.
A
ge
ne
r
a
l
model
f
or
AN
N
a
r
c
hit
e
c
tur
e
T
o
de
ter
mi
ne
the
pe
r
c
e
ntage
of
a
dult
e
r
a
ti
on
,
the
p
r
opor
ti
on
o
f
s
e
gmente
d
pixels
r
e
lative
to
the
tot
a
l
pixels
in
the
R
OI
wa
s
c
a
lcula
ted
us
ing
(
4
)
.
T
he
s
e
r
e
s
ult
s
we
r
e
then
c
ompar
e
d
with
the
a
c
tual
da
ta
to
obtain
the
c
oe
f
f
icie
nt
of
de
ter
mi
n
a
ti
on
(
R
²)
a
nd
r
oot
mea
n
s
qua
r
e
e
r
r
or
(
R
M
S
E
)
va
lues
.
T
he
be
s
t
model
wa
s
s
e
lec
ted
ba
s
e
d
on
the
lar
ge
s
t
R
²
a
nd
the
lowe
s
t
R
M
S
E
va
lue,
e
ns
ur
ing
opti
mal
pe
r
f
o
r
manc
e
in
qua
nti
f
ying
a
dult
e
r
a
ti
on.
T
his
pr
oc
e
s
s
a
li
gns
with
the
va
li
da
ti
o
n
a
nd
tes
ti
ng
pha
s
e
s
de
s
c
r
ib
e
d,
whe
r
e
R
²
a
nd
R
M
S
E
we
r
e
ke
y
metr
ics
f
or
e
va
luating
the
model's
a
c
c
ur
a
c
y.
A
d
u
l
te
r
a
nt
=
∑
p
i
i
j
R
O
I
×
100%
(
4)
W
he
r
e
is
the
pixel
in
the
s
a
mpl
e
im
a
ge
matr
ix.
3.
RE
S
UL
T
S
AN
D
DI
S
CU
S
S
I
ON
3.
1.
Hu
e
-
s
at
u
r
at
ion
-
val
u
e
s
e
gm
e
n
t
at
io
n
T
he
HSV
c
olor
pa
r
a
mete
r
r
e
s
ult
s
s
howe
d
s
igni
f
ica
nt
dif
f
e
r
e
nc
e
s
in
H,
S
,
a
nd
V
va
lues
whic
h
we
r
e
de
f
ined
a
s
lowe
r
a
nd
uppe
r
li
mi
ts
of
the
HSV
c
o
lor
pa
r
a
mete
r
s
f
o
r
s
oybe
a
ns
,
c
or
n
,
g
r
e
e
n
be
a
ns
,
a
nd
s
a
nd.
T
he
H
pa
r
a
mete
r
f
or
s
oybe
a
ns
r
a
nge
d
f
r
om
5
to
1
77,
S
r
a
nge
d
f
r
o
m
2
to
95,
a
nd
V
r
a
nge
d
f
r
om
59
to
126.
F
or
c
or
n
,
the
H
r
a
nge
d
f
r
om
12
to
23,
S
r
a
nge
d
f
r
om
136
to
245
,
a
nd
V
r
a
nge
d
f
r
om
174
to
233
.
T
he
H
pa
r
a
mete
r
f
or
gr
e
e
n
be
a
ns
r
a
nge
d
f
r
om
34
to
98,
S
r
a
nge
d
f
r
om
25
to
110,
a
nd
V
r
a
nge
d
f
r
om
117
to
191.
M
e
a
nw
hil
e
,
the
H
pa
r
a
mete
r
in
s
a
nd
r
a
nge
d
f
r
o
m
101
to
173,
S
r
a
nge
d
f
r
o
m
7
to
129
,
a
nd
V
r
a
ng
e
d
f
r
om
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
5
,
Oc
tober
20
25
:
37
34
-
3743
3738
60
to
183.
T
he
r
e
s
ult
s
s
howe
d
that
the
dif
f
e
r
e
nc
e
s
in
H,
S
,
a
nd
V
va
lues
f
or
s
oybe
a
ns
,
c
or
n
,
gr
e
e
n
be
a
ns
,
a
nd
s
oybe
a
ns
we
r
e
vis
ibl
e
f
o
r
the
3
-
dim
e
ns
ional
(
3D)
gr
a
phic
im
a
ge
a
s
pr
e
s
e
nted
in
F
igur
e
4.
T
he
di
f
f
e
r
e
nc
e
s
s
howe
d
a
potenc
y
in
dis
ti
nguis
hing
s
a
mpl
e
s
ba
s
e
d
on
HSV
c
olor
s
.
T
he
c
olo
r
c
ompos
it
ion
made
it
po
s
s
ibl
e
to
dif
f
e
r
e
nti
a
te
be
twe
e
n
s
oybe
a
n
a
nd
othe
r
im
pu
r
it
ie
s
,
including
c
or
n
,
g
r
e
e
n
be
a
ns
,
a
nd
s
a
nd.
How
e
v
e
r
,
s
ome
c
olor
a
r
e
a
s
ha
ve
c
olor
int
e
r
s
e
c
ti
ons
whe
r
e
c
e
r
tain
c
olor
a
r
e
a
s
in
s
oybe
a
ns
ha
d
the
s
a
me
va
lue
a
s
c
e
r
tain
a
r
e
a
s
in
the
mi
xtur
e
whic
h
c
a
us
e
d
bias
in
s
e
gmenta
ti
on.
I
n
HSV
s
e
gmenta
ti
on,
bias
r
e
duc
ti
on
wa
s
done
us
ing
im
a
ge
pr
oc
e
s
s
ing
f
unc
ti
ons
in
the
pr
og
r
a
mm
ing
l
ibr
a
r
y
(
Ope
nC
V)
.
M
e
a
nwhile,
in
the
HSV
-
AN
N
method,
the
inf
luenc
e
of
bias
wa
s
c
a
lcula
ted
in
the
t
r
a
ns
f
e
r
f
unc
ti
on
to
ob
tain
maximum
output
.
S
oybe
a
n
C
or
n
M
ungbe
a
n
S
a
nd
R
G
B
i
ma
ge
3D
H
S
V
i
ma
ge
s
F
igur
e
4.
R
GB
im
a
ge
s
a
nd
3D
HSV
im
a
ge
s
of
s
oy
be
a
n,
c
or
n,
mungbea
n,
a
nd
s
a
nd
3.
2.
H
ue
-
s
at
u
r
at
ion
-
val
u
e
s
e
gm
e
n
t
at
io
n
m
od
e
l
T
he
e
xa
mpl
e
r
e
s
ult
o
f
HSV
thr
e
s
holdi
ng
s
e
gmenta
ti
on
is
s
hown
in
F
igur
e
5.
T
he
i
mage
s
howe
d
that
a
dult
e
r
a
nts
s
uc
h
a
s
c
or
n,
gr
e
e
n
be
a
ns
,
a
nd
s
a
nd
we
r
e
dis
ti
nc
t;
howe
ve
r
,
pa
r
ts
of
a
dult
e
r
a
nts
we
r
e
not
de
tec
ted.
E
s
pe
c
ially
f
or
c
or
n,
only
the
e
ndos
pe
r
m
wa
s
s
e
gm
e
nted,
while
the
e
mbr
yo
wa
s
not
s
e
gmente
d,
whic
h
r
e
duc
e
d
the
s
e
gmente
d
vis
ua
l
a
ppe
a
r
a
nc
e
.
I
n
a
ddit
ion,
s
ha
dow
r
e
duc
ti
on
wa
s
c
a
r
r
ied
out
in
thi
s
s
e
gm
e
ntation
us
ing
the
c
v2.
in
r
a
nge
method
to
obtain
be
tt
e
r
r
e
s
ult
s
.
I
mage
p
r
oc
e
s
s
ing
c
a
n
be
c
a
r
r
ied
out
wit
h
s
e
v
e
r
a
l
f
unc
ti
ons
to
ge
t
a
be
tt
e
r
s
e
gmenta
ti
on
dis
play
[
29
]
.
T
he
a
dva
ntage
of
the
HSV
method
is
that
it
a
ll
o
ws
im
a
ge
pr
oc
e
s
s
ing
dir
e
c
tl
y
by
a
pplyi
ng
methods
in
pr
og
r
a
mm
ing
langua
ge
s
.
S
e
gmenta
ti
on
models
us
ing
HS
V
c
olor
pa
r
a
mete
r
s
pr
oduc
e
be
tt
e
r
outpu
t
than
thos
e
us
ing
other
c
olor
pa
r
a
mete
r
s
[
30]
.
F
igur
e
5.
E
xa
mpl
e
of
s
e
gmenta
ti
on
r
e
s
ult
s
us
ing
th
e
HSV
model
f
or
5%
a
dult
e
r
a
ti
on
in
s
oybe
a
n
3.
3.
H
ue
-
s
at
u
r
at
ion
-
val
u
e
-
ar
t
if
icial
n
e
u
r
al
n
e
t
w
or
k
s
s
e
gm
e
n
t
at
ion
m
od
e
l
HSV
-
ANN
modeling
wa
s
obtaine
d
us
ing
s
ix
h
id
de
n
laye
r
s
,
e
a
c
h
c
ons
is
ti
ng
of
16,
32
,
64,
64
,
32,
a
nd
16
node
s
,
r
e
s
ult
ing
in
a
n
a
c
c
ur
a
c
y
s
c
or
e
of
0.
97.
T
he
model
us
e
d
the
T
e
ns
or
F
low
li
br
a
r
y
in
P
yt
hon
with
f
unc
ti
ons
in
the
ha
r
d
f
unc
ti
on
c
las
s
s
uc
h
a
s
models
,
s
e
que
nti
a
l,
a
nd
De
ns
e
.
T
he
De
ns
e
pa
r
a
mete
r
s
in
the
input
include
d
the
number
of
hidden
laye
r
s
a
nd
the
r
e
c
ti
f
ied
a
c
ti
va
ti
on
f
unc
ti
on
(
R
e
LU
)
a
c
ti
va
ti
on,
while
the
output
us
e
d
“
s
igm
oid”
a
c
ti
va
ti
on.
At
the
mode
l
c
ompi
lation
s
teps
,
the
c
ompi
le(
)
f
unc
ti
on
is
us
e
d
f
or
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
C
ompar
is
on
of
HSV
-
c
olor
and
A
N
N
-
HSV
-
c
olor
s
e
gme
ntat
ion
for
de
tec
ti
ng
…
(
F
ar
id
R
ahmat
A
badi
)
3739
pa
r
a
mete
r
s
s
uc
h
a
s
opti
mi
z
e
r
=
"
a
da
m"
,
los
s
=
"
bin
a
r
y_c
r
os
s
e
ntr
opy"
(
s
ince
the
model
output
is
bin
a
r
y)
,
a
nd
metr
ics
=
[
"
a
c
c
ur
a
c
y"
]
.
Dur
ing
the
model
f
it
ti
ng
s
tep,
the
f
it
(
)
f
unc
ti
on
wa
s
e
mpl
oye
d
with
the
pa
r
a
mete
r
s
ba
tch_s
ize
=
10
a
nd
e
poc
hs
=
100.
T
he
HSV
-
AN
N
model
s
e
gment
a
ti
on
r
e
s
ult
e
d
in
mor
e
de
tailed
pa
tt
e
r
ns
;
none
thele
s
s
,
the
r
e
s
ult
s
s
ti
ll
e
xhibi
ted
a
f
e
w
inappr
opr
iate
pa
r
ts
.
S
ha
dow
r
e
gio
ns
r
e
maine
d
vis
ibl
e
in
the
a
dult
e
r
a
ted
s
a
mpl
e
s
e
gm
e
ntation.
T
hus
,
thes
e
r
e
gions
we
r
e
s
a
mpl
e
d
a
nd
us
e
d
a
s
a
s
u
btr
a
c
ti
on
f
a
c
tor
in
the
HSV
method
a
s
s
hown
in
F
igur
e
5.
I
n
c
ontr
a
s
t,
the
AN
N
-
HSV
model
de
ter
mi
ne
d
the
output
pixel
thr
ough
modeling
c
a
lcula
ti
ons
.
C
olor
e
d
s
li
c
e
s
be
twe
e
n
s
oybe
a
ns
a
nd
other
ingr
e
dients
we
r
e
obs
e
r
va
ble
in
the
r
e
s
ult
s
a
s
pr
e
s
e
nted
in
F
igu
r
e
6
.
Addi
ti
ona
ll
y,
s
ince
the
modeling
input
r
e
li
e
d
on
HSV
pa
r
a
m
e
ter
s
,
s
e
gmenta
ti
on
wa
s
li
mi
ted
to
r
e
gions
with
dis
ti
nc
t
f
e
a
tur
e
s
.
S
im
il
a
r
to
dir
e
c
t
HSV
s
e
gmenta
ti
on,
pa
r
ts
of
the
c
or
n
e
mbr
yo
c
ould
not
be
de
tec
ted
in
the
HSV
-
AN
N
method
.
I
nput
laye
r
ne
twor
k
c
a
lcula
ti
ons
c
a
n
be
im
pleme
nted
with
ha
r
dwa
r
e
in
r
e
a
l
-
ti
me;
howe
ve
r
,
be
c
a
us
e
e
a
c
h
ne
twor
k
is
f
u
ll
y
c
onne
c
ted,
it
is
les
s
s
uit
a
ble
f
or
3D
or
2D
im
a
ge
s
e
gmenta
ti
on
due
to
the
lar
ge
number
of
pa
r
a
mete
r
s
r
e
qui
r
e
d
[
31]
.
C
on
s
ider
ing
the
r
e
lations
hip
be
twe
e
n
pixels
,
objec
t
r
e
c
ognit
ion
us
ing
AN
N
is
les
s
e
f
f
e
c
ti
ve
[
32]
.
Ne
ve
r
thele
s
s
,
f
ur
ther
s
tudi
e
s
on
pixel
r
e
lations
hips
a
r
e
ne
c
e
s
s
a
r
y
t
o
e
va
luate
the
pe
r
f
or
manc
e
of
s
e
gmenta
ti
on
r
e
s
ult
s
;
thus
,
i
mpr
ove
ments
in
the
modeling
p
r
oc
e
s
s
,
s
uc
h
a
s
modi
f
ying
AN
N
input
s
,
a
r
e
r
e
qui
r
e
d.
An
AN
N
model
f
o
r
im
a
ge
s
e
gmenta
ti
on
c
a
n
be
de
ve
loped
by
c
ombi
ning
mul
ti
ple
pa
r
a
mete
r
s
a
nd
r
e
-
tr
a
ini
ng
laye
r
s
to
a
c
hi
e
ve
the
be
s
t
pe
r
f
or
manc
e
[
33
]
.
F
igur
e
6
s
hows
that
the
c
or
n
,
gr
e
e
n
be
a
ns
,
a
nd
s
a
nd
s
e
c
ti
ons
we
r
e
a
de
qua
tely
s
e
gmente
d
a
nd
vis
ua
l
ly
dis
ti
nguis
ha
ble
to
the
human
e
ye
.
How
e
ve
r
,
the
s
ha
dow
s
e
c
ti
ons
we
r
e
s
ti
ll
s
e
gmente
d,
incr
e
a
s
ing
the
number
of
pixels
c
ounted
a
s
pa
r
t
of
the
s
oybe
a
n.
T
his
is
s
ue
c
a
n
be
les
s
e
ne
d
by
c
a
li
br
a
ti
ng
the
m
ode
l
a
nd
e
xplor
ing
opti
mal
s
a
mpl
ing
of
c
olor
f
e
a
tur
e
s
.
F
igur
e
6.
E
xa
mpl
e
of
s
e
gmenta
ti
on
r
e
s
ult
s
us
ing
th
e
HSV
-
AN
N
model
f
or
5%
a
dult
e
r
a
ti
on
in
s
oybe
a
n
3.
4.
M
od
e
l
v
a
li
d
at
ion
Af
ter
the
c
a
li
br
a
ti
on
models
we
r
e
obtaine
d,
va
li
da
ti
on
models
we
r
e
ob
taine
d
us
ing
the
tes
t
da
tas
e
ts
.
F
igur
e
7
s
hows
a
c
tual
a
nd
pr
e
dicte
d
model
to
de
ter
mi
ne
a
dult
e
r
a
ti
on
ba
s
e
d
on
the
HSV
method
f
or
c
a
li
br
a
ti
on
a
nd
va
li
da
ti
on.
F
igur
e
7
(
a
)
s
hows
the
c
a
li
br
a
ti
on
model
us
ing
HSV
s
e
gmenta
ti
on,
whic
h
plot
s
the
a
c
tual
a
nd
pr
e
dicte
d
pe
r
c
e
ntage
s
of
a
dult
e
r
a
nts
in
s
oybe
a
n.
T
he
c
a
li
br
a
ti
on
model
a
c
hieve
d
a
c
oe
f
f
icie
nt
of
de
ter
mi
na
ti
on
(
R
2
)
of
0
.
95
indi
c
a
ti
ng
that
i
mage
f
e
a
tur
e
s
us
e
d
in
the
model
e
xplain
95%
of
the
va
r
i
a
ti
ons
in
pr
e
dicting
s
oybe
a
n
a
dult
e
r
a
ti
on,
a
nd
r
oot
mea
n
s
qua
r
e
of
r
e
gr
e
s
s
ion
(
R
M
S
E
)
of
18.
7
%
.
T
o
f
ur
ther
e
va
luate
the
pe
r
f
o
r
manc
e
o
f
the
HSV
c
a
li
br
a
ti
on
model
,
va
l
idation
wa
s
done
us
i
ng
tes
t
da
tas
e
ts
,
whic
h
a
c
hieve
d
a
n
R
2
of
0.
98
a
nd
R
M
S
E
o
f
4.
48
%
a
s
s
hown
in
F
igur
e
7
(
b
)
.
M
e
a
nwhile,
F
igur
e
8
s
hows
a
c
tual
a
nd
pr
e
dicte
d
model
to
de
ter
mi
ne
a
dult
e
r
a
ti
on
ba
s
e
d
on
the
HSV
-
AN
N
method
f
or
c
a
li
br
a
ti
on
a
nd
va
li
da
ti
on;
in
the
A
NN
-
HSV
s
e
gmenta
ti
on
model,
the
c
a
li
br
a
ti
on
model
a
c
hieve
d
R
2
of
0
.
97
a
nd
R
M
S
E
o
f
2
.
16%
a
s
pr
e
s
e
nted
in
F
igur
e
8
(
a
)
,
a
nd
the
va
li
da
ti
on
model
a
c
hieve
d
R
2
of
0
.
96
a
nd
R
M
S
E
o
f
1.
3
%
a
s
s
hown
in
F
igu
r
e
8
(
b)
.
T
his
s
howe
d
that
the
AN
N
-
HSV
model'
s
pe
r
f
or
manc
e
wa
s
be
tt
e
r
than
the
HS
V
model,
whic
h
ha
d
a
high
e
r
R
2
a
nd
lowe
r
R
M
S
E
.
T
he
s
e
gmenta
ti
on
pr
oc
e
s
s
by
thr
e
s
holdi
ng
us
ing
t
he
c
v2.
inr
a
nge
(
Ope
nC
V)
in
HSV
modeling
wa
s
les
s
ti
me
-
c
ons
umi
ng
s
ince
ther
e
w
a
s
no
loopi
ng
pr
oc
e
s
s
.
I
n
thi
s
c
a
s
e
,
the
HSV
s
e
gmenta
ti
on
pr
o
c
e
s
s
w
a
s
c
a
r
r
ied
out
wi
th
a
n
a
ve
r
a
ge
ti
me
o
f
0
.
034
s
e
c
onds
,
whic
h
wa
s
r
e
latively
f
a
s
t.
S
e
gmenta
ti
on
r
e
s
ult
s
with
ANN
-
HSV
we
r
e
13.
52
s
e
c
onds
,
whic
h
take
s
r
e
lati
ve
ly
longer
to
dis
play.
T
his
s
hows
the
a
dva
ntage
s
of
HSV
s
e
gmenta
ti
on
dir
e
c
tl
y,
e
s
pe
c
ially
in
it
s
u
ti
li
z
a
ti
on
f
or
gr
a
phi
c
a
l
us
e
r
int
e
r
f
a
c
e
(
GU
I
)
de
ve
lopm
e
nt
[
34
]
.
T
his
is
a
n
a
dva
ntage
obtaine
d
by
the
dir
e
c
t
HSV
s
e
gmenta
ti
on
method.
How
e
ve
r
,
p
r
e
diction
s
a
r
e
only
li
mi
ted
to
c
olor
pa
r
a
mete
r
s
,
s
e
ve
r
a
l
unmea
s
ur
e
d
r
e
lations
hip
s
a
nd
bias
e
s
in
the
s
e
gmenta
ti
on
pr
oc
e
s
s
may
be
r
uled
out,
r
e
duc
ing
a
c
c
ur
a
c
y.
I
n
ge
ne
r
a
l,
both
models
pe
r
f
o
r
m
we
ll
,
s
howing
thei
r
a
bil
it
y
to
p
r
e
dict
mate
r
ial
mi
xtur
e
s
a
s
a
dult
e
r
a
nts
.
T
he
r
e
s
ult
s
we
r
e
c
ompar
a
ble
with
the
f
indi
ngs
f
o
r
de
tec
ti
ng
f
r
a
ud
in
r
e
d
a
nd
blac
k
pe
ppe
r
[
35]
,
[
36]
a
nd
r
ice
[
37]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
5
,
Oc
tober
20
25
:
37
34
-
3743
3740
(
a
)
(
b)
F
igur
e
7.
Ac
tual
a
nd
p
r
e
dicte
d
model
to
de
ter
mi
ne
a
dult
e
r
a
ti
on
ba
s
e
d
on
the
HSV
method
f
or
(
a
)
c
a
li
br
a
ti
on
a
nd
(
b
)
va
li
da
ti
on
(
a
)
(
b)
F
igur
e
8.
Ac
tual
a
nd
p
r
e
dicte
d
model
to
de
ter
mi
ne
a
dult
e
r
a
ti
on
ba
s
e
d
on
the
HSV
-
AN
N
method
f
or
(
a
)
c
a
li
br
a
ti
on
a
nd
(
b
)
va
li
da
ti
on
4.
CONC
L
USI
ON
T
his
s
tudy
de
mons
tr
a
tes
that
HSV
c
olor
s
e
gmenta
ti
on
a
nd
HSV
-
AN
N
models
a
r
e
e
f
f
e
c
ti
ve
in
de
tec
ti
ng
s
oybe
a
n
a
dult
e
r
a
ti
on
a
nd
a
c
c
ur
a
tely
dis
ti
nguis
hing
be
twe
e
n
s
oybe
a
ns
a
nd
other
im
pur
it
ies
,
s
uc
h
a
s
c
or
n,
gr
e
e
n
be
a
ns
,
a
nd
s
a
nd.
T
he
HSV
method
pr
ovided
e
f
f
icie
nt
a
nd
f
a
s
t
s
e
gmenta
ti
on,
but
the
HSV
-
AN
N
model
p
r
ovided
mor
e
de
tailed
r
e
s
ult
s
;
howe
ve
r
,
it
ha
d
s
ome
c
ha
ll
e
nge
s
with
s
ha
do
w
r
e
gions
a
nd
unde
tec
ted
c
or
n
e
mbr
yo
pa
r
ts
.
B
oth
models
ha
d
a
s
tr
ong
pr
e
dictive
c
a
pa
bil
it
y
with
the
HS
V
-
AN
N
(
R
²=
0.
96,
R
M
S
E
=
1.
3%
)
model
be
ing
mo
r
e
a
c
c
ur
a
te
than
the
HSV
(
R
²=
0.
98
,
R
M
S
E
=
4.
48%
)
.
How
e
ve
r
,
the
dr
a
wba
c
k
of
the
AN
N
model’
s
f
ull
y
c
onne
c
ted
a
r
c
hit
e
c
tur
e
is
that
it
is
not
e
a
s
il
y
e
xtenda
ble
f
or
c
ompl
e
x
im
a
ge
s
e
gmenta
ti
on
due
to
c
omput
a
ti
ona
l
e
xpe
ns
e
s
.
F
utur
e
wo
r
k
s
hould
a
ls
o
a
im
to
im
pr
ove
mod
e
l
input
s
,
include
mor
e
pa
r
a
mete
r
s
,
a
nd
r
e
f
ine
s
a
mpl
ing
methods
to
incr
e
a
s
e
s
e
gmenta
ti
on
pr
e
c
is
ion
a
nd
o
ve
r
c
ome
bias
e
s
,
e
s
pe
c
ially
in
the
s
ha
dow
r
e
gions
a
nd
ove
r
l
a
pping
c
olor
f
e
a
tur
e
s
.
T
he
s
e
im
pr
ove
ments
c
ould
r
e
s
ult
in
mor
e
e
f
f
e
c
ti
ve
a
nd
non
-
de
s
tr
uc
ti
ve
a
ppr
oa
c
he
s
to
identi
f
ying
the
a
dult
e
r
a
ti
on
of
s
oybe
a
ns
a
nd
other
a
gr
icultur
a
l
pr
oduc
ts
.
AC
KNOWL
E
DGM
E
N
T
S
T
his
s
e
c
ti
on
s
hould
a
c
knowle
dge
indi
viduals
who
pr
ovided
pe
r
s
ona
l
a
s
s
is
tanc
e
to
the
wor
k
but
do
not
mee
t
the
c
r
it
e
r
ia
f
or
a
uthor
s
hip,
de
taili
ng
thei
r
c
ontr
ibut
ions
.
I
t
is
i
mper
a
ti
ve
to
obtain
c
ons
e
nt
f
r
om
a
ll
indi
viduals
li
s
ted
in
the
a
c
kno
wle
dgments
.
F
UN
DI
NG
I
NF
ORM
AT
I
ON
T
his
r
e
s
e
a
r
c
h
wa
s
f
unde
d
by
the
M
ini
s
ter
of
E
duc
a
ti
on,
C
ult
ur
e
,
R
e
s
e
a
r
c
h,
a
nd
T
e
c
hnology
of
I
ndone
s
ia,
gr
a
nt
numbe
r
048/
E
5/P
G.
02
.
00.
P
L
/2024
a
nd
2782/UN1/DI
T
L
I
T
/P
T
.
01
.
03/2024
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
C
ompar
is
on
of
HSV
-
c
olor
and
A
N
N
-
HSV
-
c
olor
s
e
gme
ntat
ion
for
de
tec
ti
ng
…
(
F
ar
id
R
ahmat
A
badi
)
3741
AU
T
HO
R
CONT
RI
B
U
T
I
ONS
S
T
AT
E
M
E
N
T
T
his
jour
na
l
us
e
s
the
C
ontr
ibut
o
r
R
oles
T
a
xo
nomy
(
C
R
e
diT
)
to
r
e
c
ognize
indi
vidual
a
uthor
c
ontr
ibut
ions
,
r
e
duc
e
a
utho
r
s
hip
dis
putes
,
a
nd
f
a
c
il
it
a
te
c
oll
a
bor
a
ti
on.
Nam
e
of
Au
t
h
or
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
F
a
r
id
R
a
k
hmat
Aba
di
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
R
udiati
E
vi
M
a
s
it
hoh
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
L
il
ik
S
uti
a
r
s
o
✓
✓
✓
✓
✓
S
r
i
R
a
ha
yoe
✓
✓
✓
✓
✓
✓
✓
C
:
C
onc
e
pt
ua
li
z
a
ti
on
M
:
M
e
th
odol
ogy
So
:
So
f
twa
r
e
Va
:
Va
li
da
ti
on
Fo
:
Fo
r
ma
l
a
na
ly
s
is
I
:
I
nve
s
ti
ga
ti
on
R
:
R
e
s
our
c
e
s
D
:
D
a
ta
C
ur
a
ti
on
O
:
W
r
it
in
g
-
O
r
ig
in
a
l
D
r
a
f
t
E
:
W
r
it
in
g
-
R
e
vi
e
w
&
E
di
ti
ng
Vi
:
Vi
s
ua
li
z
a
ti
on
Su
:
Su
pe
r
vi
s
io
n
P
:
P
r
oj
e
c
t
a
dmi
ni
s
tr
a
ti
on
Fu
:
Fu
ndi
ng a
c
qui
s
it
io
n
CONF
L
I
CT
OF
I
NT
E
RE
S
T
S
T
AT
E
M
E
N
T
Author
s
s
tate
no
c
onf
li
c
t
of
int
e
r
e
s
t.
DA
T
A
AV
AI
L
A
B
I
L
I
T
Y
T
he
da
ta
that
s
uppor
t
the
f
indi
ngs
of
thi
s
s
tudy
a
r
e
a
va
il
a
ble
on
r
e
que
s
t
f
r
om
the
c
o
r
r
e
s
ponding
a
uthor
,
[
R
E
M
]
.
T
he
da
ta,
whic
h
c
ontain
in
f
or
mation
that
c
ould
c
omp
r
omi
s
e
the
pr
ivac
y
of
r
e
s
e
a
r
c
h
pa
r
ti
c
ipants
,
a
r
e
not
publi
c
ly
a
va
il
a
ble
due
to
c
e
r
ta
in
r
e
s
tr
ictions
.
RE
F
E
RE
NC
E
S
[
1]
A
.
S
uda
r
ić
,
“
I
nt
r
oduc
to
r
y
c
ha
pt
e
r
:
s
oybe
a
n
-
qua
li
ty
a
nd
ut
il
iz
a
ti
on,”
in
Soy
be
an
fo
r
H
um
an
C
ons
um
pt
io
n
and
A
ni
m
al
F
e
e
d
,
A
. S
uda
r
ić
, E
d., R
ij
e
ka
:
I
nt
e
c
hO
pe
n, 2020, doi:
10.5772/i
nt
e
c
hope
n.93942.
[
2]
J
.
C
.
M
oor
e
,
J
.
S
pi
nk,
a
nd
M
.
L
ip
p,
“
D
e
ve
lo
pme
nt
a
nd
a
ppl
ic
a
ti
on
of
a
da
ta
ba
s
e
of
f
ood
in
g
r
e
di
e
nt
f
r
a
ud
a
nd
e
c
onomi
c
a
ll
y
mot
iv
a
te
d
a
dul
te
r
a
ti
on
f
r
om
1980
to
2010,”
J
our
nal
of
F
ood
Sc
ie
nc
e
,
vol
.
77,
no.
4,
2012,
doi
:
10.1111/
j.
1
750
-
3841.2012.02657.x.
[
3]
J
.
D
.
S
il
va
,
S
.
P
r
ude
nc
io
,
M
.
C
.
-
P
a
ni
z
z
i,
C
.
G
r
e
gor
ut
,
F
.
F
ons
e
c
a
,
a
nd
L
.
M
a
tt
os
o,
“
S
tu
dy
on
th
e
f
la
vour
of
s
oybe
a
n
c
ul
ti
va
r
s
by
s
e
ns
or
y
a
na
ly
s
i
s
a
nd
e
le
c
tr
oni
c
to
ngue
,”
I
nt
e
r
nat
io
nal
J
our
nal
of
F
ood
Sc
ie
nc
e
and
T
e
c
hnol
ogy
,
vol
.
47,
no.
8,
pp.
1630
–
16
38,
2012, doi:
10.1111/j
.1365
-
2621.2012.03013.x.
[
4]
K
.
M
.
S
ør
e
ns
e
n,
B
.
K
ha
ki
mov,
a
nd
S
.
B
.
E
nge
ls
e
n,
“
T
he
us
e
of
r
a
pi
d
s
pe
c
tr
os
c
opi
c
s
c
r
e
e
ni
ng
me
th
od
s
to
de
te
c
t
a
dul
te
r
a
ti
on
of
f
ood r
a
w
ma
te
r
ia
ls
a
nd i
ngr
e
di
e
nt
s
,”
C
u
r
r
e
nt
O
pi
ni
on i
n
F
ood
Sc
ie
nc
e
, vol
. 10, pp. 45
–
51, 2016, doi:
10.1016/j
.c
of
s
.2016.08.
001.
[
5]
P
.
V
it
hu
a
nd
J
.
A
.
M
os
e
s
,
“
M
a
c
hi
ne
vi
s
io
n
s
ys
t
e
m
f
or
f
ood
gr
a
in
qua
li
ty
e
va
lu
a
ti
on:
a
r
e
vi
e
w
,”
T
r
e
nds
in
F
ood
Sc
ie
nc
e
and
T
e
c
hnol
ogy
, vol
. 56, pp. 13
–
20, 2016, doi:
10.1016/j
.t
if
s
.2016.07.011.
[
6]
X
. L
i
e
t
al
.
, “
O
r
ig
in
t
r
a
c
e
a
bi
li
ty
a
nd a
dul
te
r
a
ti
on de
te
c
ti
on of
s
oybe
a
n us
in
g ne
a
r
i
nf
r
a
r
e
d hype
r
s
pe
c
tr
a
l
im
a
gi
ng,”
F
ood F
r
ont
ie
r
s
,
vol
. 5, no. 2, pp. 237
–
244, J
a
n. 2024, doi:
10.1002/f
f
t2
.345.
[
7]
F
.
R
.
A
ba
di
,
R
.
E
.
M
a
s
it
hoh,
L
.
S
ut
ia
r
s
o,
a
nd
S
.
R
a
ha
yoe
,
“
E
va
lu
a
ti
on
of
I
ndone
s
ia
n
lo
c
a
l
s
oybe
a
n
ba
s
e
d
on
c
he
mi
c
a
l
c
ha
r
a
c
te
r
is
ti
c
s
a
nd
vi
s
ib
le
-
ne
a
r
in
f
r
a
r
e
d
s
pe
c
tr
a
w
it
h
c
he
mom
e
tr
ic
s
,”
B
io
tr
opi
a
,
vol
.
31,
no.
1,
pp.
63
–
75,
2
024,
doi
:
10.11598/B
T
B
.2024.31.1.2054.
[
8]
H
.
Z
.
A
ma
na
h
e
t
al
.
,
“
N
onde
s
tr
uc
ti
ve
me
a
s
ur
e
me
nt
of
a
nt
hoc
ya
ni
n
in
in
ta
c
t
s
oybe
a
n
s
e
e
d
us
in
g
f
our
ie
r
tr
a
ns
f
or
m
ne
a
r
-
in
f
r
a
r
e
d
(
F
T
-
N
I
R
)
a
nd
F
our
ie
r
tr
a
ns
f
or
m
in
f
r
a
r
e
d
(
F
T
-
I
R
)
s
pe
c
tr
os
c
opy,”
I
nf
r
ar
e
d
P
hy
s
ic
s
and
T
e
c
hnol
ogy
,
vol
.
111,
2020,
doi
:
10.1016/j
.i
nf
r
a
r
e
d.2020.103477.
[
9]
M
.
F
.
R
.
P
a
hl
a
w
a
n,
B
.
M
.
A
.
M
ur
ti
,
a
nd
R
.
E
.
M
a
s
it
hoh,
“
T
he
pot
e
nc
y
of
V
is
/NI
R
s
pe
c
tr
os
c
opy
f
or
c
la
s
s
if
ic
a
ti
on
of
s
oy
be
a
n
ba
s
e
d
of
c
ol
our
,”
I
O
P
C
onf
e
r
e
n
c
e
Se
r
ie
s
:
E
ar
th
and
E
nv
ir
onm
e
nt
al
Sc
ie
nc
e
,
vol
.
1018,
no.
1,
2022,
doi
:
10.1088/1755
-
1315/1018/
1/
01
2015.
[
10]
S
.
K
a
us
ha
l,
D
.
K
.
T
a
mm
in
e
ni
,
P
.
R
a
n
a
,
M
.
S
ha
r
ma
,
K
.
S
r
id
ha
r
,
a
nd
H
.
H
.
C
he
n,
“
C
omput
e
r
vi
s
io
n
a
nd
de
e
p
le
a
r
ni
ng
-
ba
s
e
d
a
ppr
oa
c
he
s
f
or
de
te
c
ti
on
of
f
ood
nut
r
ie
nt
s
/n
ut
r
it
io
n:
ne
w
in
s
ig
ht
s
a
nd
a
dva
nc
e
s
,”
T
r
e
nds
in
F
ood
Sc
ie
n
c
e
and
T
e
c
hnol
ogy
,
vol
.
146, no. Oc
to
be
r
2023, 2024, doi:
10.1016/j
.t
if
s
.2024.1044
08.
[
11]
S
.
J
it
a
na
n
a
nd
P
.
C
hi
ml
e
k,
“
Q
u
a
li
ty
gr
a
di
ng
of
s
oybe
a
n
s
e
e
d
s
us
in
g
im
a
ge
a
n
a
ly
s
is
,
”
I
nt
e
r
nat
io
nal
J
our
nal
of
E
le
c
t
r
ic
al
and
C
om
put
e
r
E
ngi
ne
e
r
in
g
, vol
. 9, no. 5, pp. 3495
–
3503, 2019, doi:
10
.11591/i
je
c
e
.v9i
5.pp3495
-
3503.
[
12]
R
.
de
C
.
M
.
M
ont
e
ir
o,
G
.
I
.
G
a
dot
ti
,
V
.
M
a
ld
a
ne
r
,
A
.
B
.
J
.
C
ur
i,
a
nd
M
.
B
.
N
e
to
,
“
I
ma
ge
pr
oc
e
s
s
in
g
to
id
e
nt
if
y
da
ma
ge
to
s
oybe
a
n s
e
e
ds
,
”
C
ie
nc
ia
R
ur
al
, vol
. 51, no. 2, pp. 1
–
8, 2020, do
i:
10.1590/0103
-
8478c
r
20200107.
[
13
]
S
.
J
a
r
di
m,
J
.
V
a
le
nt
e
,
A
.
A
lm
e
id
a
,
a
nd
C
.
M
or
a
,
“
C
ompa
r
in
g
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
c
la
s
s
if
ic
a
ti
on
mode
ls
to
im
pr
ove
a
n
im
a
ge
c
ompa
r
is
on s
ys
te
m wit
h u
s
e
r
i
nput
s
,”
SN
C
om
put
e
r
S
c
ie
nc
e
, vo
l.
5, no. 1, 2024, doi:
10.1007/s
42979
-
023
-
02375
-
y.
[
14]
S
.
Y
a
ng,
Y
.
L
in
,
Y
.
L
i,
D
.
X
u,
S
.
Z
ha
ng,
a
nd
L
.
P
e
ng,
“
D
e
e
p
ne
ur
a
l
ne
twor
k
-
ba
s
e
d
s
or
ghum
a
dul
te
r
a
ti
on
de
te
c
ti
on
in
b
a
ij
iu
br
e
w
in
g,”
I
E
E
E
O
pe
n
J
our
nal
of
I
ns
t
r
um
e
nt
at
io
n
and
M
e
as
ur
e
m
e
nt
,
vol
.
1,
no.
J
un
e
,
pp.
1
–
8,
2
022,
doi
:
10.1109/OJ
I
M
.2022.3190024.
[
15]
P
.
S
a
r
a
nya
a
nd
R
.
D
ur
ga
,
“
F
ood
s
a
f
e
ty
c
ont
r
ol
us
in
g C
N
N
mo
de
l
in
im
a
ge
pr
oc
e
s
s
in
g
te
c
hni
que
,”
2023
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
N
e
w
F
r
ont
ie
r
s
in
C
om
m
uni
c
at
io
n,
A
ut
om
at
io
n,
M
anage
m
e
nt
and
Se
c
ur
it
y
,
I
C
C
A
M
S
2023
,
vol
.
1,
pp.
1
–
8,
2023,
doi
:
10.1109/I
C
C
A
M
S
6011
3.2023.10526092.
[
16]
J
.
V
.
C
.
C
.
P
a
ix
ã
o
e
t
al
.
,
“
C
la
s
s
if
ic
a
ti
on
of
s
oybe
a
n
c
ul
ti
va
r
s
by
me
a
ns
of
a
r
ti
f
ic
ia
l
ne
ur
a
l
ne
two
r
ks
,”
A
gr
onomy
Sc
ie
nc
e
and
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
Ar
ti
f
I
ntell
,
Vol.
14
,
No.
5
,
Oc
tober
20
25
:
37
34
-
3743
3742
B
io
te
c
hnol
ogy
, vol
. 9, pp. 1
–
11, 2023, doi:
10.33158/as
b.r
186.v9.2023.
[
17]
E
.
M
.
S
a
ndova
l,
M
.
E
.
M
.
R
os
a
s
,
J
.
R
.
M
.
S
a
ndova
l,
M
.
M
.
M
.
V
e
la
s
c
o,
a
nd
H
.
C
.
D
e
Á
vi
la
,
“
C
ol
or
a
na
ly
s
i
s
a
nd
i
ma
ge
pr
oc
e
s
s
in
g
a
ppl
ie
d
in
a
gr
ic
ul
tu
r
e
,”
in
C
ol
or
ime
tr
y
and
I
m
age
P
r
oc
e
s
s
in
g
,
C
.
M
.
T
r
a
vi
e
s
o
-
G
onz
a
le
z
,
E
d.,
2018,
doi
:
10.5772/i
nt
e
c
hope
n.71935.
[
18]
C
.
J
unhua
a
nd
L
.
J
in
g,
“
R
e
s
e
a
r
c
h
on
c
ol
or
im
a
ge
c
la
s
s
if
ic
a
ti
on
ba
s
e
d
on
H
S
V
c
ol
or
s
pa
c
e
,”
P
r
oc
e
e
di
ng
s
of
th
e
2012
2nd
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
I
ns
tr
um
e
nt
at
io
n
and
M
e
as
u
r
e
m
e
nt
,
C
om
put
e
r
,
C
om
m
uni
c
at
io
n
and
C
ont
r
ol
,
I
M
C
C
C
2012
,
no
.
3,
pp. 944
–
947, 2012, doi:
10.1109/I
M
C
C
C
.2012.226.
[
19]
L
.
L
.
P
.
N
guye
n,
L
.
B
a
r
a
nya
i,
D
.
N
a
gy,
P
.
V
.
M
a
ha
ja
n,
V
.
Z
.
-
M
uha
,
a
nd
T
.
Z
s
om,
“
C
ol
or
a
n
a
ly
s
is
of
hor
ti
c
ul
tu
r
a
l
pr
oduc
e
s
us
in
g
hue
s
pe
c
tr
a
f
in
ge
r
pr
in
ti
ng,”
M
e
th
ods
X
, vol
. 8, no. Nove
mbe
r
, 2021, doi:
10.1016/j
.me
x.2021.101594.
[
20]
C
.
Y
.
C
hi
a
nd
S
.
C
.
T
a
i,
“
P
e
r
c
e
pt
ua
l
c
ol
or
c
ont
r
a
s
t
ba
s
e
d
w
a
t
e
r
s
he
d
f
or
c
ol
or
im
a
ge
s
e
gme
nt
a
ti
on,”
C
onf
e
r
e
nc
e
P
r
oc
e
e
di
n
gs
-
I
E
E
E
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
Sy
s
te
m
s
,
M
an
and
C
y
be
r
ne
ti
c
s
,
vol
.
4,
pp.
3548
–
3553,
2
006,
doi
:
10.1109/I
C
S
M
C
.2006.384669.
[
21]
S
.
M
.
M
.
-
M
ogha
dda
m,
M
.
R
.
Y
a
ma
gha
ni
,
a
nd
A
.
B
a
kh
s
hi
pour
,
“
A
ppl
ic
a
ti
on
of
e
le
c
tr
oni
c
nos
e
a
nd
e
y
e
s
ys
t
e
ms
f
or
de
te
c
ti
on
of
a
dul
te
r
a
ti
on
in
ol
iv
e
oi
l
ba
s
e
d
on
c
h
e
mom
e
tr
ic
s
a
nd
opt
im
iz
a
ti
on
a
ppr
oa
c
he
s
,”
J
our
nal
of
U
ni
v
e
r
s
al
C
om
put
e
r
Sc
ie
nc
e
,
vol
.
29,
no. 4, pp. 300
–
325, 2023, doi
:
10.3897/j
uc
s
.90346.
[
22]
A
.
P
a
la
na
nda
a
nd
W
.
K
im
pa
n,
“
T
ur
bi
di
ty
of
c
o
c
onut
oi
l
de
te
r
mi
na
ti
on
us
in
g
th
e
M
A
M
o
H
me
th
od
in
im
a
ge
pr
oc
e
s
s
in
g,
”
I
E
E
E
A
c
c
e
s
s
, vol
. 9, pp. 41494
–
41505, 2021, doi:
10.1109/AC
C
E
S
S
.
2021.3065004.
[
23]
R
.
N
a
pi
a
nt
o,
A
.
T
.
P
r
a
s
to
w
o,
A
.
D
.
W
a
hyudi,
W
a
ji
r
a
n,
Y
.
E
.
P
.
Y
udoutom
o,
a
nd
A
.
A
.
S
upa
r
to
,
“
I
de
nt
if
ic
a
ti
on
of
be
e
f
f
r
e
s
hn
e
s
s
us
in
g
li
ne
a
r
di
s
c
r
im
in
a
nt
a
na
ly
s
i
s
(
L
D
A
)
a
lg
or
it
hm
a
nd
e
xt
r
a
c
ti
on
of
H
S
V
va
lu
e
s
,
”
P
r
oc
e
e
di
ngs
-
2022
2nd
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
E
le
c
tr
oni
c
and
E
l
e
c
t
r
ic
al
E
ngi
ne
e
r
in
g
and
I
nt
e
ll
ig
e
nt
Sy
s
te
m
,
I
C
E
3I
S
2022
,
pp.
114
–
119,
2022,
doi
:
10.1109/I
C
E
3I
S
56585.2022.10010006.
[
24]
X
.
W
u,
X
.
F
u,
Y
.
L
iu
,
E
.
P
.
L
im
,
S
.
C
.
H
.
H
oi
,
a
nd
Q
.
S
un,
“
A
la
r
ge
-
s
c
a
le
be
nc
hma
r
k
f
or
f
ood
im
a
ge
s
e
gme
nt
a
ti
on,”
M
M
20
21
-
P
r
oc
e
e
di
n
gs
of
th
e
29t
h
A
C
M
I
nt
e
r
nat
io
nal
C
onf
e
r
e
n
c
e
on
M
ul
ti
m
e
di
a
,
vol
.
1,
no.
1,
pp.
506
–
515,
20
21,
doi
:
10.1145/3474085.34
75201.
[
25]
T
.
S
a
r
ka
r
e
t
al
.
,
“
A
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
a
id
e
d
a
dul
te
r
a
ti
on
de
t
e
c
ti
on
a
nd
qua
nt
if
ic
a
ti
on
f
or
r
e
d
c
hi
ll
i
pow
de
r
,”
F
ood
A
nal
y
t
i
c
al
M
e
th
ods
, vol
. 16, no. 4, pp. 721
–
748, 2023, doi:
10.1007/s
12161
-
023
-
02445
-
0.
[
26]
M
.
A
.
M
or
s
hi
di
,
M
.
H
.
M
a
r
ha
ba
n,
a
nd
A
.
J
a
nt
a
n,
“
C
ol
or
s
e
gme
nt
a
ti
on
us
in
g
mul
ti
la
ye
r
ne
u
r
a
l
ne
twor
k
a
nd
th
e
H
S
V
c
ol
or
s
pa
c
e
,”
P
r
oc
e
e
di
ngs
of
th
e
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
C
om
put
e
r
and
C
om
m
uni
c
at
io
n
E
ngi
ne
e
r
in
g
2008,
I
C
C
C
E
08:
G
lo
bal
L
in
k
s
f
or
H
um
an D
e
v
e
lo
pm
e
nt
, pp. 1335
–
1339, 2008, doi:
10.1
109/
I
C
C
C
E
.2008.4580822.
[
27]
N
.
S
in
gh,
S
.
S
in
gh,
a
nd
K
.
S
he
vka
ni
,
“
M
a
iz
e
:
c
ompos
it
io
n,
bi
oa
c
ti
ve
c
ons
ti
tu
e
nt
s
,
a
nd
unl
e
a
ve
n
e
d
br
e
a
d,”
in
F
lo
ur
and
B
r
e
ads
and the
ir
F
or
ti
fi
c
at
io
n i
n H
e
al
th
and Dis
e
as
e
P
r
e
v
e
nt
io
n
, V
. R
. P
r
e
e
dy, R
. R
. W
a
ts
on, a
nd V
. B
. B
. T
.
-
F
. a
nd B
. a
nd t
he
i
r
F
.
in
H
.
a
nd D
. P
. P
a
te
l,
E
ds
., S
a
n D
ie
go:
A
c
a
de
mi
c
P
r
e
s
s
, 2011, pp. 89
–
99, doi:
10.1016/B
978
-
0
-
12
-
380886
-
8.10009
-
1.
[
28]
K
.
G
oya
l,
P
.
K
uma
r
,
a
nd
K
.
V
e
r
ma
,
“
F
ood
a
dul
te
r
a
ti
on
de
te
c
ti
on
us
in
g
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
:
a
s
ys
te
ma
ti
c
r
e
vi
e
w
,”
A
r
c
hi
v
e
s
of
C
om
put
at
io
nal
M
e
th
ods
i
n E
ngi
ne
e
r
in
g
, vol
. 29, no. 1, pp. 397
–
426, 2022, doi:
10.1007/s
11831
-
021
-
09600
-
y.
[
29]
L
.
F
u,
Z
.
L
iu
,
Y
.
M
a
je
e
d,
a
nd
Y
.
C
ui
,
“
K
iwi
f
r
ui
t
yi
e
ld
e
s
ti
ma
ti
on
us
in
g
im
a
ge
pr
oc
e
s
s
in
g
by
a
n
a
ndr
oi
d
mobi
le
phone
,”
I
F
A
C
-
P
ape
r
s
O
nL
in
e
, vol
. 51, no. 17, pp. 185
–
190, 2018, doi:
10.1016/j
.i
f
a
c
ol
.2018.08.137.
[
30]
D
.
H
e
ma
a
nd
D
.
S
.
K
a
nn
a
n,
“
I
nt
e
r
a
c
ti
ve
c
ol
or
im
a
ge
s
e
gme
nt
a
ti
on
us
in
g
H
S
V
c
ol
or
s
pa
c
e
,
”
Sc
ie
nc
e
&
T
e
c
hnol
ogy
J
ou
r
nal
,
vol
. 7, no. 1, pp. 37
–
41, 2019, doi:
10.22232/s
tj
.2019.07.01.05.
[
31]
L
. A
lz
uba
id
i
e
t
al
.
, “
R
e
vi
e
w
of
de
e
p l
e
a
r
ni
ng:
c
onc
e
pt
s
, C
N
N
a
r
c
hi
te
c
tu
r
e
s
, c
ha
ll
e
nge
s
, a
ppl
ic
a
ti
ons
, f
ut
ur
e
di
r
e
c
ti
on
s
,”
J
our
na
l
of
B
ig
D
at
a
, vol
. 8, no. 1, 2021, doi:
10.1186/s
40537
-
021
-
00444
-
8.
[
32]
H
.
K
im
,
H
.
K
im
,
S
.
C
ho,
a
nd
E
.
H
w
a
ng,
“
M
a
ni
pul
a
ti
ng
n
e
ur
a
l
ne
twor
k
bl
oc
k
f
or
r
obus
t
im
a
ge
s
e
gme
nt
a
ti
on,”
in
P
r
oc
e
e
di
n
gs
-
2022
I
E
E
E
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
B
ig
D
at
a
and
Sm
ar
t
C
om
put
in
g,
B
ig
C
om
p
2022
,
U
.
H
.,
K
.
Y
.
-
K
.,
H
.
H
.,
C
.
S
.
-
B
.,
P
.
S
.,
K
.
K
.,
H
.
Y
.G
.,
K
.
J
.,
M
.
A
.,
W
.
C
.,
K
.
H
.
-
Y
.,
M
.
Y
.
-
S
.,
a
nd
L
.
C
.,
E
ds
.,
S
c
hool
of
E
le
c
tr
ic
a
l
E
nge
ni
e
e
r
in
g,
K
or
e
a
U
ni
ve
r
s
it
y,
S
e
oul
,
S
out
h
K
or
e
a
:
I
ns
ti
tu
te
o
f
E
le
c
tr
ic
a
l
a
nd
E
le
c
tr
on
ic
s
E
ngi
ne
e
r
s
I
nc
.,
2022,
pp.
246
–
250,
doi
:
10.1109/B
ig
C
omp54360.2022.000
54.
[
33]
B
.
K
nys
h
a
nd
Y
.
K
ul
yk,
“
D
e
ve
lo
pme
nt
of
a
n
im
a
ge
s
e
gme
nt
a
ti
on
mode
l
ba
s
e
d
on
a
c
onvolut
io
na
l
ne
ur
a
l
ne
twor
k,”
E
as
te
r
n
-
E
ur
ope
an J
our
nal
of
E
nt
e
r
pr
i
s
e
T
e
c
hnol
ogi
e
s
, vol
. 2, no.
2
–
110, pp. 6
–
15, 2021, doi:
10.15587/1729
-
4061.2021.228644.
[
34]
M
.
M
.
-
L
in
ow
,
F
.
P
.
-
E
s
pi
nos
a
,
H
.
S
c
ha
r
r
,
a
nd
U
.
R
a
s
c
he
r
,
“
T
h
e
le
a
f
a
ngl
e
di
s
tr
ib
ut
io
n
of
na
tu
r
a
l
p
la
nt
popula
ti
ons
:
a
s
s
e
s
s
in
g
th
e
c
a
nopy with a nove
l
s
of
twa
r
e
t
ool
,”
P
la
nt
M
e
th
ods
, vol
. 11, n
o.
1, pp. 1
–
16, 2015, doi:
10.1186/s
13007
-
015
-
0052
-
z.
[
35]
M
.
H
.
N
a
r
ge
s
i
a
nd
K
.
K
he
ir
a
li
pour
,
“
V
i
s
ib
le
f
e
a
tu
r
e
e
ngi
n
e
e
r
in
g
to
de
te
c
t
f
r
a
ud
in
bl
a
c
k
a
nd
r
e
d
pe
ppe
r
s
,
”
Sc
ie
nt
if
ic
R
e
po
r
ts
,
vol
. 14, no. 1, 2024, doi
:
10.1038/s
41598
-
024
-
76617
-
1.
[
36]
N
.
F
a
ti
ma
,
Q
.
M
.
A
r
e
e
b,
I
.
M
.
K
ha
n,
a
nd
M
.
M
.
K
ha
n,
“
S
ia
me
s
e
ne
twor
k
-
ba
s
e
d
c
omput
e
r
vi
s
io
n
a
ppr
oa
c
h
to
d
e
te
c
t
pa
p
a
ya
s
e
e
d
a
dul
te
r
a
ti
on i
n bl
a
c
k pe
ppe
r
c
or
ns
,”
J
ou
r
nal
of
F
ood P
r
oc
e
s
s
in
g
and P
r
e
s
e
r
v
at
io
n
, vol
. 46, no. 9, 2022, doi
:
10.1111/j
f
pp.1604
3.
[
37]
B
.
S
.
A
na
mi
,
N
.
N
.
M
a
lv
a
de
,
a
nd
S
.
P
a
la
ia
h,
“
A
ut
oma
te
d
r
e
c
ogni
ti
on
a
nd
c
la
s
s
if
ic
a
ti
on
of
a
dul
te
r
a
ti
on
le
ve
l
s
f
r
om
bul
k
pa
ddy
gr
a
in
s
a
mpl
e
s
,”
I
nf
or
m
at
io
n P
r
oc
e
s
s
in
g i
n A
gr
ic
ul
tu
r
e
, vol
. 6,
no. 1, pp. 47
–
60, 2019, doi:
10.1016/j
.i
npa
.2018.09.001.
B
I
OG
RA
P
HI
E
S
OF
AU
T
HO
RS
F
a
ri
d
R
a
k
h
m
a
t
A
b
a
di
o
b
t
ai
n
e
d
h
i
s
D
o
c
t
o
ra
l
D
eg
ree
at
St
u
d
y
Pr
o
g
ram
o
f
A
g
r
i
cu
l
t
u
ral
E
n
g
i
n
eer
i
n
g
Sci
en
ce,
U
n
i
v
er
s
i
t
as
G
ad
j
a
h
Mad
a,
In
d
o
n
es
i
a.
H
e
i
s
cu
rre
n
t
l
y
w
o
r
k
i
n
g
at
t
h
e
Bea
n
Cro
p
A
s
s
em
b
l
y
an
d
T
e
s
t
i
n
g
C
en
t
er
(Ba
l
ai
Pera
k
i
t
an
d
a
n
Pen
g
u
j
i
a
n
T
an
ama
n
A
n
e
k
a
K
acan
g
),
J
l
.
Ray
a
K
e
n
d
a
l
p
a
y
ak
N
o
.
6
6
,
Seg
aran
,
Mal
an
g
,
J
aw
a
T
i
mu
r
6
5
1
6
2
,
In
d
o
n
e
s
ia
.
H
e
can
b
e
co
n
t
act
ed
at
ema
i
l
:
fari
d
_
ra9
8
@
y
a
h
o
o
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
Ar
ti
f
I
ntell
I
S
S
N:
2252
-
8938
C
ompar
is
on
of
HSV
-
c
olor
and
A
N
N
-
HSV
-
c
olor
s
e
gme
ntat
ion
for
de
tec
ti
ng
…
(
F
ar
id
R
ahmat
A
badi
)
3743
R
udi
a
t
i
E
v
i
M
a
s
i
th
o
h
o
b
t
ai
n
ed
h
er
D
o
c
t
o
ra
l
D
eg
ree
at
S
t
u
d
y
Pr
o
g
ram
o
f
A
g
r
i
cu
l
t
u
ral
E
n
g
i
n
e
eri
n
g
Sci
e
n
ce,
U
n
i
v
ers
i
t
a
s
G
ad
j
a
h
Mad
a,
In
d
o
n
e
s
i
a,
an
d
Mas
t
er
D
e
g
ree
a
t
t
h
e
U
n
i
v
ers
i
t
y
o
f
Mel
b
o
u
r
n
e,
A
u
s
t
ra
l
i
a.
Sh
e
i
s
cu
rren
t
l
y
t
h
e
H
ea
d
o
f
D
o
c
t
o
ra
l
p
r
o
g
ram
a
t
A
g
r
i
cu
l
t
u
ral
E
n
g
i
n
eeri
n
g
Sc
i
en
ce,
U
n
i
v
er
s
i
t
as
G
ad
j
ah
Mad
a,
In
d
o
n
e
s
i
a.
H
er
res
earc
h
i
n
t
eres
t
s
i
n
c
l
u
d
e
t
h
e
n
o
n
-
d
e
s
t
r
u
ct
i
v
e
ev
a
l
u
a
t
i
o
n
o
f
fo
o
d
an
d
ag
ri
c
u
l
t
u
ra
l
p
ro
d
u
ct
s
.
Sh
e
can
b
e
co
n
t
ac
t
e
d
at
emai
l
:
ev
i
@
u
g
m.
ac.
i
d
.
Li
l
i
k
Sut
i
a
r
s
o
o
b
t
a
i
n
e
d
h
i
s
D
o
ct
o
ral
D
e
g
ree
at
T
s
u
k
u
b
a
U
n
i
v
er
s
i
t
y
,
J
a
p
an
,
an
d
Mas
t
er
D
eg
ree
at
t
h
e
A
s
i
a
n
In
s
t
i
t
u
t
e
o
f
T
ech
n
o
l
o
g
y
,
T
h
a
i
l
a
n
d
.
H
e
i
s
cu
rren
t
l
y
t
h
e
H
ea
d
o
f
D
ep
ar
t
men
t
o
f
A
g
r
i
cu
l
t
u
ral
a
n
d
Bi
o
s
y
s
t
em
s
E
n
g
i
n
eeri
n
g
,
U
n
i
v
ers
i
t
a
s
G
ad
j
ah
Ma
d
a,
In
d
o
n
e
s
i
a.
H
i
s
res
earch
i
n
t
eres
t
s
i
n
cl
u
d
e
ag
r
i
cu
l
t
u
ral
s
y
s
t
ems
an
d
p
reci
s
i
o
n
farmi
n
g
.
H
e
can
b
e
co
n
t
act
e
d
at
emai
l
:
l
i
l
i
k
-
s
o
e
t
i
ar
s
o
@
u
g
m.
ac.
i
d
.
Sri
R
a
ha
y
o
e
o
b
t
a
i
n
e
d
h
er
D
o
c
t
o
ra
l
D
eg
ree
at
Ch
em
i
cal
E
n
g
i
n
eeri
n
g
,
U
n
i
v
er
s
i
t
as
G
ad
j
ah
Mad
a,
an
d
Mas
t
e
r
D
eg
ree
a
t
A
g
r
i
cu
l
t
u
ral
E
n
g
i
n
eeri
n
g
,
U
n
i
v
er
s
i
t
as
G
ad
j
a
h
Mad
a.
Sh
e
i
s
cu
rre
n
t
l
y
t
h
e
V
i
ce
D
ean
a
t
t
h
e
Facu
l
t
y
o
f
A
g
ri
c
u
l
t
u
ral
T
ec
h
n
o
l
o
g
y
,
U
n
i
v
ers
i
t
a
s
G
ad
j
a
h
Mad
a,
In
d
o
n
e
s
i
a.
H
er
res
earch
i
n
t
eres
t
s
i
n
cl
u
d
e
fo
o
d
an
d
p
o
s
t
h
arv
e
s
t
en
g
i
n
eeri
n
g
.
Sh
e
can
b
e
co
n
t
act
e
d
at
emai
l
:
s
ra
h
ay
o
e@
u
g
m.
ac.
i
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.