I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
,
p
p
.
1
5
2
3
~1
5
3
1
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v
1
5
i
4
.
pp
1
5
2
3
-
1
5
3
1
1523
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
Autom
a
tic
mu
lti
-
cla
ss
clas
sifica
tion o
f
g
eo
g
ra
phica
l
indica
tion
-
tag
g
ed t
riba
l art
using
deep
l
ea
rni
ng
M
a
la
y
S.
B
ha
t
t
1
,
Apurv
a
A.
M
eht
a
2
1
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
,
G
.
H
.
P
a
t
e
l
C
o
l
l
e
g
e
o
f
En
g
i
n
e
e
r
i
n
g
,
C
V
M
U
n
i
v
e
r
si
t
y
,
V
V
N
a
g
a
r
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
En
g
i
n
e
e
r
i
n
g
,
D
h
a
r
ms
i
n
h
D
e
sa
i
U
n
i
v
e
r
si
t
y
,
N
a
d
i
a
d
,
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
Sep
3
0
,
2
0
2
5
R
ev
is
ed
J
u
n
1
8
,
2
0
2
6
Acc
ep
ted
J
u
l 5
,
2
0
2
6
In
th
e
f
ield
o
f
c
o
m
p
u
ter
v
isio
n
,
d
e
e
p
lea
rn
i
n
g
h
a
s
sig
n
ifi
c
a
n
tl
y
imp
ro
v
e
d
a
u
to
m
a
ted
ima
g
e
c
las
sifica
ti
o
n
,
y
e
t
th
e
c
las
sifica
ti
o
n
o
f
I
n
d
i
a
n
tri
b
a
l
a
rtwo
rk
s
re
m
a
in
s
larg
e
ly
u
n
e
x
p
l
o
re
d
.
Th
e
se
a
rt
fo
rm
s
e
x
h
ib
i
t
su
b
sta
n
ti
a
l
sty
li
stic
o
v
e
rlap
,
v
a
riatio
n
s
in
c
o
lo
r
p
a
lettes
,
a
n
d
c
o
m
p
lex
h
a
n
d
-
d
ra
wn
p
a
tt
e
rn
s,
m
a
k
i
n
g
m
a
n
u
a
l
i
d
e
n
ti
fic
a
ti
o
n
d
iff
icu
lt
a
n
d
a
u
t
o
m
a
ted
c
las
sifica
ti
o
n
e
v
e
n
m
o
re
c
h
a
ll
e
n
g
i
n
g
.
E
x
isti
n
g
m
e
th
o
d
s
fo
r
a
rtwo
r
k
re
c
o
g
n
it
i
o
n
a
re
li
m
it
e
d
b
y
sm
a
ll
d
a
tas
e
ts,
d
o
m
a
in
-
s
p
e
c
ifi
c
v
a
riatio
n
s
,
a
n
d
a
lac
k
o
f
c
u
ra
ted
tri
b
a
l
a
rt
c
o
ll
e
c
ti
o
n
s.
T
h
is
st
u
d
y
a
d
d
re
ss
e
s
th
e
s
e
g
a
p
s
b
y
d
e
v
e
lo
p
in
g
a
d
e
e
p
lea
rn
in
g
-
b
a
se
d
fra
m
e
wo
rk
fo
r
t
h
e
m
u
lt
i
-
c
las
s
c
las
sifica
ti
o
n
o
f
th
re
e
c
u
lt
u
ra
ll
y
sig
n
ifi
c
a
n
t
I
n
d
ian
tr
ib
a
l
a
rts:
M
a
n
ju
sh
a
,
G
o
n
d
,
a
n
d
Warli
.
A
m
a
n
u
a
ll
y
c
u
ra
ted
d
a
tas
e
t
o
f
9
0
0
ima
g
e
s
w
a
s
a
u
g
m
e
n
ted
u
si
n
g
g
e
o
m
e
tri
c
a
n
d
c
o
lo
r
-
b
a
se
d
tr
a
n
sfo
rm
a
ti
o
n
s
t
o
i
n
c
re
a
se
ro
b
u
st
n
e
ss
.
Th
e
m
o
d
e
l
is
b
u
il
t
o
n
t
h
e
In
c
e
p
ti
o
n
-
Re
sN
e
t
-
V2
a
rc
h
it
e
c
tu
r
e
u
sin
g
tran
sfe
r
lea
rn
i
n
g
.
E
x
p
e
rime
n
tal
e
v
a
lu
a
ti
o
n
d
e
m
o
n
stra
tes
h
i
g
h
p
e
rfo
rm
a
n
c
e
,
a
c
h
iev
in
g
9
8
.
2
1
%
tes
ti
n
g
a
c
c
u
ra
c
y
a
n
d
sta
b
le
v
a
li
d
a
ti
o
n
t
re
n
d
s.
Th
e
p
ro
p
o
se
d
a
p
p
ro
a
c
h
p
ro
v
i
d
e
s
a
sc
a
lab
le
b
a
se
li
n
e
fo
r
tri
b
a
l
a
rt
c
las
sifica
ti
o
n
,
su
p
p
o
r
ti
n
g
d
i
g
it
a
l
h
e
rit
a
g
e
p
re
se
rv
a
ti
o
n
a
n
d
fu
t
u
re
d
e
v
e
l
o
p
m
e
n
t
o
f
re
tri
e
v
a
l
o
r
e
d
u
c
a
ti
o
n
a
l
sy
st
e
m
s.
K
ey
w
o
r
d
s
:
C
u
ltu
r
al
h
er
itag
e
co
m
p
u
tin
g
Dee
p
lear
n
in
g
I
m
ag
e
p
r
o
ce
s
s
in
g
T
r
an
s
f
er
lear
n
i
n
g
T
r
ib
al
ar
t c
lass
if
icatio
n
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
:
Ap
u
r
v
a
A.
Me
h
ta
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
,
D
h
ar
m
s
in
h
Desai
Un
i
v
er
s
ity
Nad
iad
,
Gu
jar
at
,
I
n
d
ia
E
m
ail:
ap
u
r
v
am
e
h
ta.
ce
@
d
d
u
.
ac
.
in
1.
I
NT
RO
D
UCT
I
O
N
Go
v
er
n
m
e
n
ts
wo
r
ld
wid
e
ar
e
in
cr
ea
s
in
g
ly
d
ig
itizin
g
cu
lt
u
r
al
h
er
itag
e
co
llectio
n
s
to
im
p
r
o
v
e
ac
ce
s
s
ib
ilit
y
an
d
lo
n
g
-
ter
m
p
r
eser
v
atio
n
.
I
n
d
ian
tr
ib
al
ar
t,
a
cr
u
cial
co
m
p
o
n
e
n
t
o
f
th
is
h
er
itag
e,
ap
p
ea
r
s
in
p
ain
tin
g
s
,
tex
tiles
,
cr
a
f
ts
,
an
d
co
n
tem
p
o
r
ar
y
d
esig
n
.
Ho
wev
er
,
d
esp
ite
its
c
u
ltu
r
al
im
p
o
r
tan
ce
,
au
to
m
atica
lly
class
if
y
in
g
tr
ib
al
ar
two
r
k
r
em
ain
s
a
n
o
n
-
tr
iv
ial
task
.
Ma
n
y
tr
ib
al
ar
t
f
o
r
m
s
s
h
ar
e
s
im
ilar
v
is
u
al
s
tr
u
ctu
r
es,
r
ec
u
r
r
in
g
g
eo
m
etr
ic
m
o
tifs,
an
d
o
v
er
la
p
p
in
g
co
lo
r
s
ch
e
m
es,
m
ak
in
g
it
d
if
f
icu
lt
ev
en
f
o
r
e
x
p
er
ts
to
d
is
tin
g
u
is
h
t
h
em
r
eliab
ly
.
R
ec
en
t
ad
v
an
ce
s
in
d
ee
p
lear
n
in
g
h
av
e
s
h
o
wn
s
tr
o
n
g
p
er
f
o
r
m
an
ce
in
im
ag
e
r
ec
o
g
n
itio
n
task
s
s
u
ch
as
f
in
e
-
ar
t
class
if
icatio
n
,
cu
lt
u
r
al
h
er
itag
e
an
aly
s
is
,
an
d
te
x
tu
r
e
-
b
ased
p
atter
n
r
ec
o
g
n
itio
n
.
Sev
er
al
s
tu
d
ies
h
av
e
em
p
l
o
y
ed
C
NNs,
h
y
b
r
id
m
o
d
els,
o
r
tr
an
s
f
er
lear
n
in
g
to
class
if
y
ar
two
r
k
s
o
r
h
er
itag
e
i
m
ag
er
y
,
ac
h
iev
in
g
h
ig
h
ac
cu
r
ac
y
u
n
d
er
c
u
r
ated
d
atasets
.
Ho
wev
er
,
th
ese
ap
p
r
o
ac
h
es
p
r
im
ar
il
y
f
o
c
u
s
o
n
w
ester
n
ar
t
g
en
r
es,
C
h
in
ese
b
r
u
s
h
s
tr
o
k
e
s
ty
les,
o
r
h
er
itag
e
m
o
n
u
m
e
n
ts
,
leav
in
g
I
n
d
ian
tr
i
b
al
ar
t
lar
g
ely
u
n
ad
d
r
ess
ed
.
Mo
r
eo
v
e
r
,
th
e
ab
s
en
ce
o
f
p
u
b
licly
av
aila
b
le
tr
ib
al
ar
t d
atasets
f
u
r
th
e
r
lim
its
p
r
o
g
r
ess
in
th
is
d
o
m
ai
n
.
T
h
is
s
tu
d
y
ad
d
r
ess
es
th
ese
g
ap
s
b
y
in
v
esti
g
atin
g
d
ee
p
lear
n
in
g
tech
n
iq
u
es
f
o
r
th
e
class
i
f
icatio
n
o
f
th
r
ee
m
ajo
r
I
n
d
ian
tr
ib
al
ar
t
f
o
r
m
s
:
Ma
n
ju
s
h
a,
Go
n
d
,
an
d
W
ar
li.
T
h
e
o
b
jectiv
e
is
to
d
es
ig
n
an
ef
f
ec
tiv
e
an
d
r
o
b
u
s
t
m
o
d
el
ca
p
a
b
le
o
f
d
is
tin
g
u
is
h
in
g
th
ese
v
is
u
ally
s
im
ilar
ar
t
s
ty
les,
wh
ile
estab
lis
h
in
g
a
b
aselin
e
d
ataset
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
5
2
3
-
1
5
3
1
1524
an
d
m
et
h
o
d
o
lo
g
y
th
at
ca
n
s
u
p
p
o
r
t
f
u
tu
r
e
r
esear
c
h
.
Usi
n
g
tr
an
s
f
er
lear
n
i
n
g
with
I
n
ce
p
ti
o
n
-
R
esNet
-
V2
an
d
ex
ten
s
iv
e
d
ata
au
g
m
e
n
tatio
n
,
t
h
e
p
r
o
p
o
s
ed
f
r
am
ew
o
r
k
p
r
o
v
i
d
es
a
s
ca
lab
le
s
o
lu
tio
n
f
o
r
tr
ib
al
ar
t
class
if
icatio
n
an
d
co
n
tr
ib
u
tes to
th
e
d
ig
ital p
r
eser
v
atio
n
o
f
in
d
ig
e
n
o
u
s
cu
lt
u
r
al
ass
ets.
Fig
u
r
e
1
illu
s
tr
ates
ten
p
r
o
m
in
en
t
I
n
d
ian
f
o
lk
-
ar
t
tr
a
d
itio
n
s
.
Fig
u
r
e
1
(
a)
s
h
o
ws
W
ar
li
ar
t
ch
ar
ac
ter
ized
b
y
g
eo
m
etr
ic
s
t
ick
f
ig
u
r
es,
wh
ile
Fig
u
r
e
1
(
b
)
p
r
esen
ts
Kh
o
v
ar
a
r
t
k
n
o
wn
f
o
r
its
m
u
r
al
-
s
ty
le
m
o
tifs.
Fig
u
r
e
1
(
c)
d
ep
icts
T
a
n
jo
r
e/T
h
an
ja
v
u
r
a
r
t
with
r
ich
r
elief
an
d
g
o
ld
f
o
ils
.
Fig
u
r
e
1
(
d
)
d
is
p
lay
s
Kaw
ad
s
to
r
y
tellin
g
p
an
els.
Fig
u
r
e
1
(
e
)
s
h
o
ws
Ma
d
h
u
b
an
i
ar
t
with
i
n
tr
icate
m
y
th
o
lo
g
ical
s
ce
n
es.
Fig
u
r
e
1
(
f
)
p
r
esen
ts
Pattach
itra
s
cr
o
ll
p
ain
tin
g
s
.
Fig
u
r
e
1
(
g
)
f
ea
t
u
r
es
Sau
r
a
ar
t
b
a
s
ed
o
n
tr
ib
al
r
itu
als.
Fig
u
r
e
1
(
h
)
s
h
o
ws
Go
n
d
ar
t,
id
en
tifia
b
le
b
y
lin
e
p
atter
n
s
a
n
d
s
h
ad
in
g
.
Fig
u
r
e
1
(
i)
p
r
esen
ts
B
h
il
ar
t
u
s
in
g
d
o
t
p
atter
n
s
,
an
d
Fig
u
r
e
1
(
j)
illu
s
tr
ates M
an
ju
s
h
a
ar
t w
ith
its
d
is
tin
ctiv
e
lin
ea
r
b
o
r
d
e
r
s
an
d
s
y
m
b
o
lic
co
lo
r
s
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
(
g
)
(
h
)
(
i)
(
j)
Fig
u
r
e
1
.
Pro
m
in
en
t I
n
d
ian
f
o
l
k
ar
ts
:
(
a)
W
ar
li
,
(
b
)
K
h
o
v
a
r
,
(
c)
T
an
jo
r
e/T
h
an
jav
u
r
,
(
d
)
Ka
wad
,
(
e)
Ma
d
h
u
b
an
i
,
(
f
)
Pattach
itra
,
(
g
)
Sau
r
a
,
(
h
)
Go
n
d
,
(
i)
B
h
il
,
an
d
(
j)
Ma
n
ju
s
h
a
Go
n
d
ar
t
[
1
]
is
a
f
o
r
m
o
f
th
e
Go
n
d
co
m
m
u
n
ity
o
f
Ma
d
h
y
a
Pra
d
esh
is
f
o
r
p
r
es
er
v
in
g
an
d
co
m
m
u
n
icatin
g
th
eir
cu
ltu
r
e.
Or
ig
in
atin
g
f
r
o
m
th
e
s
tate
o
f
Ma
d
h
y
a
Pra
d
esh
an
d
n
ea
r
b
y
s
tates,
th
i
s
ar
t
i
s
ca
r
v
ed
o
n
th
e
m
u
d
walls.
T
h
ese
p
ain
tin
g
s
ar
e
r
ic
h
in
d
e
tails
,
lin
es,
co
lo
r
s
,
m
y
s
ter
y
,
an
d
h
u
m
o
r
.
T
h
ese
p
ain
tin
g
s
ar
e
also
d
r
awn
o
n
p
ap
er
,
ca
n
v
as,
an
d
clo
t
h
.
I
n
th
e
s
e
p
ain
tin
g
s
,
lin
es,
d
o
ts
,
an
d
d
ash
es
ar
e
im
p
o
r
tan
t
f
ea
tu
r
es.
T
h
e
p
o
wer
f
u
l
r
ep
r
es
en
tatio
n
o
f
n
atu
r
al
elem
en
ts
with
th
e
r
ec
u
r
r
en
t
p
r
esen
ce
o
f
d
esig
n
is
th
e
s
o
u
n
d
f
ea
tu
r
e
o
f
th
is
ar
t
f
o
r
m
.
T
h
is
p
ain
tin
g
is
d
r
aw
n
d
u
r
in
g
m
ajo
r
f
e
s
tiv
als
s
u
ch
as
Ho
li,
Diwali,
an
d
m
a
n
y
o
th
er
s
.
T
h
ese
p
ain
tin
g
s
ar
e
s
o
g
o
o
d
in
q
u
ality
th
at
th
e
y
ca
n
last
f
o
r
a
b
o
u
t 2
0
y
ea
r
s
ev
en
with
o
u
t
an
y
tam
p
er
in
g
.
Ma
n
ju
s
h
a
a
r
t
[
2
]
is
th
e
h
er
itag
e
o
f
An
g
a
Pra
d
esh
.
Acc
o
r
d
in
g
to
Ved
ic
h
is
to
r
ian
s
,
t
h
is
ar
t
is
th
e
a
r
t
o
f
th
e
in
d
u
s
-
v
alley
civ
ilizatio
n
.
I
n
wh
ich
th
e
s
to
r
y
o
f
‘
B
ih
u
la
B
is
h
ar
i’
is
d
ep
icted
.
I
t
is
th
e
f
ir
s
t
f
o
lk
tr
ad
itio
n
al
ar
t in
Asi
a.
W
h
ich
is
a
f
u
ll st
o
r
y
.
T
h
is
ar
t o
r
ig
in
ated
i
n
th
e
s
tate
ca
p
ital,
C
h
am
p
a,
wh
ich
is
cu
r
r
en
tly
l
o
ca
ted
in
B
h
ag
alp
u
r
B
ih
ar
.
Pin
k
,
y
ello
w,
an
d
g
r
ee
n
co
lo
r
s
a
r
e
m
ai
n
l
y
u
s
ed
in
p
ain
tin
g
.
T
h
ese
co
lo
r
s
h
av
e
eso
ter
ic
a
n
d
s
y
m
b
o
lic
m
ea
n
in
g
s
.
T
h
e
p
in
k
an
d
y
ello
w
co
lo
r
s
s
ig
n
if
y
e
x
citem
en
t
an
d
ex
u
b
er
an
ce
,
wh
ile
g
r
ee
n
is
a
s
y
m
b
o
l
o
f
g
lo
o
m
a
n
d
g
r
o
wth
.
B
o
r
d
er
s
in
Ma
n
ju
s
h
a
ar
t
in
clu
d
e
d
esig
n
s
o
f
B
elp
atr
a
,
L
eh
ar
iy
a,
T
r
ia
n
g
le,
Mo
k
h
a
,
an
d
a
s
er
ies
o
f
Sn
ak
es.
T
h
e
tr
a
d
itio
n
al
ar
tis
t
o
f
Ma
n
ju
s
h
a
p
ain
tin
g
,
C
h
ak
r
av
a
r
ti
Dev
i
u
s
ed
n
at
u
r
al
co
lo
r
s
.
B
u
t
n
o
w
p
o
s
ter
s
,
wate
r
,
o
il,
an
d
‘
ac
r
y
lic
co
lo
r
’
a
r
e
b
ein
g
u
s
ed
.
B
elo
n
g
in
g
to
th
e
s
tate
o
f
Ma
h
ar
ash
tr
a,
W
ar
li
tr
ib
al
ar
t
[
1
]
is
f
am
o
u
s
f
o
r
its
ele
m
en
tar
y
wall
p
ain
tin
g
s
.
I
t
is
o
n
e
o
f
th
e
f
in
es
t
ex
am
p
les
o
f
f
o
lk
p
ain
tin
g
s
.
I
n
th
is
,
b
asic
g
eo
m
etr
ical
s
h
ap
es
ar
e
u
s
ed
s
u
ch
as
s
q
u
ar
es,
cir
cles,
an
d
tr
ian
g
les.
T
h
ese
p
ain
tin
g
s
ar
e
ca
r
v
ed
i
n
wh
ite
co
lo
u
r
(
b
am
b
o
o
is
u
s
ed
as
a
b
r
u
s
h
)
o
n
a
d
ar
k
r
ed
c
o
lo
u
r
ed
b
a
ck
g
r
o
u
n
d
.
T
h
e
p
ain
tin
g
p
o
r
tr
a
y
s
h
u
n
tin
g
,
f
esti
v
als,
f
is
h
in
g
,
f
ar
m
in
g
,
d
an
cin
g
,
a
n
d
o
th
e
r
s
ce
n
es
o
f
d
aily
life
.
T
h
e
f
ir
s
t
co
lu
m
n
in
Fig
u
r
e
2
co
n
tain
s
Go
n
d
ar
t
im
ag
es,
Ma
n
ju
s
h
a
a
r
t
s
am
p
les
ar
e
in
th
e
s
ec
o
n
d
co
lu
m
n
an
d
th
e
last
co
lu
m
n
r
ep
r
esen
ts
W
ar
li a
r
t im
a
g
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
u
to
ma
tic
mu
lti
-
cla
s
s
cla
s
s
ific
a
tio
n
o
f
g
eo
g
r
a
p
h
ica
l
in
d
ic
a
tio
n
-
ta
g
g
ed
tr
ib
a
l
a
r
t …
(
Ma
la
y
S
.
B
h
a
tt
)
1525
Fig
u
r
e
2
.
First
co
lu
m
n
-
g
o
n
d
a
r
t,
s
ec
o
n
d
c
o
lu
m
n
-
M
a
n
ju
s
h
a,
an
d
th
ir
d
co
lu
m
n
-
W
ar
li a
r
t
2.
L
I
T
E
R
AT
U
RE
SU
RVE
Y
C
o
m
p
u
tatio
n
al
alg
o
r
ith
m
s
a
n
d
d
ee
p
lea
r
n
in
g
m
o
d
els
h
av
e
b
ee
n
wid
ely
e
x
p
lo
r
e
d
f
o
r
th
e
class
if
icatio
n
o
f
ar
two
r
k
s
,
wh
er
e
th
e
task
g
e
n
er
ally
in
v
o
lv
es
ca
teg
o
r
izin
g
a
p
ain
tin
g
b
y
its
g
en
r
e,
v
is
u
al
s
ty
le,
o
r
cr
ea
to
r
.
E
a
r
ly
wo
r
k
i
n
th
i
s
d
o
m
ain
ac
h
iev
ed
p
r
o
m
is
in
g
a
cc
u
r
ac
y
u
s
in
g
d
ee
p
n
etw
o
r
k
s
f
o
r
tr
a
d
itio
n
al
C
h
in
ese
p
ain
tin
g
(
T
C
P)
class
if
icatio
n
ac
r
o
s
s
s
ix
s
ty
lis
t
ic
th
em
es
[
3
]
.
Oth
er
s
tu
d
ie
s
h
av
e
in
tr
o
d
u
ce
d
h
an
d
cr
a
f
ted
f
ea
tu
r
es
s
u
ch
as
b
r
u
s
h
s
tr
o
k
e
p
atter
n
s
to
d
is
tin
g
u
is
h
C
h
in
ese
f
r
o
m
W
ester
n
p
ain
tin
g
s
,
d
em
o
n
s
tr
atin
g
t
h
at
tex
tu
r
e
-
le
v
el
cu
es
ca
n
s
ig
n
if
ican
tly
en
h
an
ce
b
in
a
r
y
class
if
icatio
n
[
4
]
.
Geo
m
etr
ic
m
o
tif
an
aly
s
is
h
as
also
b
ee
n
a
p
p
lied
to
m
o
s
aic
ar
two
r
k
s
,
wh
e
r
e
C
NNs
s
u
cc
ess
f
u
lly
r
ec
o
g
n
ize
d
s
h
ap
es
an
d
p
atter
n
s
ev
en
u
n
d
er
p
ar
tial
d
ef
o
r
m
atio
n
[
5
]
.
B
ey
o
n
d
th
ese,
s
p
ar
s
e
m
etr
ic
lear
n
in
g
c
o
m
b
in
e
d
with
h
y
b
r
id
SVM
-
ANN
m
o
d
elin
g
h
as
b
ee
n
u
s
ed
to
class
if
y
p
ain
tin
g
s
b
y
ar
tis
t,
g
en
r
e,
an
d
s
ty
le,
th
o
u
g
h
p
e
r
f
o
r
m
an
ce
r
em
ain
s
m
o
d
er
ate
d
u
e
to
th
e
h
ig
h
v
ar
ia
b
ilit
y
in
f
in
e
-
ar
t
d
atasets
[
6
]
.
Mo
r
e
r
ec
en
t
wo
r
k
s
ad
o
p
t
d
ee
p
lear
n
in
g
p
ip
elin
es
th
at
in
te
g
r
ate
b
o
t
h
g
lo
b
al
an
d
lo
ca
l
i
m
ag
e
cu
es.
Du
al
-
p
ath
n
etwo
r
k
s
u
tili
zin
g
R
GB
an
d
s
tr
o
k
e
in
f
o
r
m
atio
n
[
7
]
,
an
d
p
atch
-
b
ased
C
NN
ar
ch
itectu
r
es
th
at
f
ee
d
in
to
s
h
allo
w
n
eu
r
al
n
etwo
r
k
s
[
8
]
,
h
a
v
e
s
h
o
wn
ef
f
ec
tiv
e
n
ess
in
ad
d
r
ess
in
g
in
tr
a
-
s
ty
le
am
b
ig
u
ity
.
L
ar
g
e
-
s
ca
le
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
5
2
3
-
1
5
3
1
1526
d
atasets
s
u
ch
as
W
ik
i
Ar
t
h
a
v
e
en
a
b
led
C
NN
-
b
ased
g
en
r
e
r
ec
o
g
n
itio
n
[
9
]
,
a
n
d
ad
d
iti
o
n
al
r
esear
c
h
h
as
ex
p
lo
r
ed
s
em
an
tic
an
d
s
tr
u
ctu
r
al
cu
es
f
o
r
r
an
k
in
g
s
im
ilar
ity
o
r
h
an
d
lin
g
ar
tis
t
-
lev
el
class
if
icatio
n
u
s
in
g
Ma
r
k
o
v
r
a
n
d
o
m
f
ield
s
an
d
C
NN
h
y
b
r
id
s
[
1
0
]
,
[
1
1
]
.
T
r
an
s
f
er
lear
n
in
g
h
as
also
g
ain
ed
tr
ac
tio
n
in
h
er
itag
e
im
ag
in
g
,
with
m
o
d
els
s
u
cc
ess
f
u
lly
class
if
y
in
g
I
n
d
ian
cu
lt
u
r
al
h
er
itag
e
s
ites
u
s
in
g
lar
g
e
i
m
ag
e
co
r
p
o
r
a
[
1
2
]
,
wh
ile
o
th
er
s
tu
d
ies h
av
e
u
s
ed
C
NN
v
ar
ian
ts
to
class
if
y
Ma
d
h
u
b
an
i a
r
t a
n
d
r
elate
d
cu
ltu
r
al
m
o
tifs
[
1
3
]
.
Par
allel
d
ev
elo
p
m
en
ts
ac
r
o
s
s
d
o
m
ain
s
s
u
ch
as
f
ac
e
-
m
ask
d
etec
tio
n
[
1
4
]
,
s
k
in
lesi
o
n
a
n
aly
s
is
[
1
5
]
,
d
ig
ital
ar
t
p
r
o
ce
s
s
in
g
[
1
6
]
,
m
u
s
ic
m
o
d
elin
g
[
1
7
]
,
ag
r
ic
u
ltu
r
e
[
1
8
]
,
m
e
d
ical
im
ag
in
g
[
1
9
]
,
m
ar
k
etin
g
[
2
0
]
,
a
n
d
d
ef
en
s
e
[
2
1
]
r
eiter
ate
th
e
b
r
o
ad
ef
f
ec
tiv
en
ess
o
f
d
ee
p
C
NNs
f
o
r
co
m
p
lex
v
is
u
al
r
ec
o
g
n
itio
n
task
s
.
W
o
r
k
s
in
v
o
lv
in
g
f
lo
wer
r
ec
o
g
n
itio
n
u
s
in
g
s
eg
m
en
tatio
n
-
ass
is
te
d
C
NNs
[
2
2
]
,
Den
s
eNe
t
-
ML
P
h
y
b
r
i
d
s
[
2
3
]
,
an
d
C
NN
-
b
ased
lo
g
o
r
ec
o
g
n
itio
n
[
2
4
]
f
u
r
th
e
r
r
ein
f
o
r
ce
t
h
e
v
er
s
atility
o
f
d
ee
p
lear
n
in
g
ar
ch
itectu
r
es.
C
o
m
p
lem
en
tin
g
th
ese
co
n
tr
ib
u
tio
n
s
,
s
u
r
v
ey
s
o
n
d
ata
au
g
m
en
tatio
n
h
ig
h
lig
h
t
th
e
cr
itical
n
ee
d
f
o
r
s
y
n
t
h
etic
s
am
p
le
g
en
er
atio
n
,
esp
ec
iall
y
in
s
ce
n
ar
io
s
in
v
o
lv
in
g
s
m
all
o
r
im
b
alan
ce
d
d
atasets
-
co
n
d
itio
n
s
f
r
e
q
u
en
tly
en
c
o
u
n
ter
e
d
in
c
u
ltu
r
al
an
d
ar
t
is
tic
im
ag
e
class
if
icat
io
n
[
2
5
]
.
T
ab
le
1
s
u
m
m
ar
izes
th
e
k
ey
d
atasets
,
class
e
s
,
ar
ch
itectu
r
es,
an
d
r
ep
o
r
ted
p
er
f
o
r
m
a
n
ce
s
f
r
o
m
th
ese
s
tu
d
ies.
W
h
ile
th
e
liter
atu
r
e
d
em
o
n
s
tr
ates
th
at
d
ee
p
C
NNs
an
d
tr
an
s
f
er
lear
n
i
n
g
ap
p
r
o
ac
h
es
ca
n
ac
h
iev
e
s
tr
o
n
g
r
esu
lts
ac
r
o
s
s
v
a
r
io
u
s
ar
t
an
d
p
a
tter
n
d
o
m
ai
n
s
,
m
o
s
t
r
ely
o
n
lar
g
e,
s
tan
d
a
r
d
ized
d
atasets
o
r
v
is
u
ally
d
is
tin
ct
s
ty
les.
I
n
c
o
n
tr
ast,
I
n
d
ian
t
r
ib
al
ar
two
r
k
s
s
u
ch
as
Ma
n
ju
s
h
a,
Go
n
d
,
a
n
d
W
ar
li
s
h
ar
e
o
v
er
lap
p
in
g
m
o
tifs,
h
an
d
-
d
r
awn
v
ar
iatio
n
s
,
an
d
s
im
ilar
co
lo
r
p
alettes,
m
ak
in
g
th
em
h
ar
d
er
to
c
lass
if
y
au
to
m
atica
lly
.
Mo
r
eo
v
er
,
n
o
p
u
b
licly
av
ailab
le
d
ataset
ex
is
t
s
f
o
r
th
ese
ar
t
f
o
r
m
s
,
an
d
n
o
p
r
io
r
wo
r
k
h
as
attem
p
ted
th
eir
m
u
lti
-
class
clas
s
if
icatio
n
.
T
h
is
g
ap
h
ig
h
lig
h
ts
th
e
n
ee
d
f
o
r
a
r
o
b
u
s
t d
ee
p
lear
n
i
n
g
f
r
a
m
ewo
r
k
tailo
r
ed
to
s
m
all,
v
is
u
ally
s
im
il
ar
cu
ltu
r
al
ar
t d
at
asets
-
an
o
b
jectiv
e
ad
d
r
ess
ed
i
n
th
e
p
r
esen
t stu
d
y
.
T
ab
le
1
.
Su
m
m
a
r
ized
d
etails o
f
th
e
r
ec
en
t w
o
r
k
s
W
o
r
k
D
a
t
a
s
e
t
N
o
.
o
f
c
l
a
ss
e
s
A
r
c
h
i
t
e
c
t
u
r
e
/
M
L
a
l
g
o
r
i
t
h
m
P
e
r
f
o
r
ma
n
c
e
[
1
]
1
1
7
0
T
C
P
i
ma
g
e
s
6
N
A
S
N
e
t
La
r
g
e
+
S
V
M
9
9
.
6
6
%
[
2
]
1
0
8
3
C
h
i
n
e
se
i
n
k
p
a
i
n
t
i
n
g
s
a
n
d
1
7
6
6
W
e
st
e
r
n
p
a
i
n
t
i
n
g
s
2
S
V
M
8
9
.
0
4
%
[
3
]
4
0
7
m
o
s
a
i
c
i
m
a
g
e
s
5
C
N
N
9
7
.
0
5
%
[
4
]
8
1
,
4
9
9
i
ma
g
e
s
o
f
f
i
n
e
-
a
r
t
p
a
i
n
t
i
n
g
s
G
e
n
r
e
:
9
A
r
t
i
st
:
9
S
t
y
l
e
:
1
5
H
y
b
r
i
d
S
u
p
p
o
e
r
V
e
c
t
o
r
M
a
c
h
i
n
e
-
A
r
t
i
f
i
c
i
a
l
N
e
u
r
a
l
N
e
t
w
o
r
k
G
e
n
r
e
:
6
4
.
3
4
%
A
r
t
i
st
:
6
8
.
4
5
%
S
t
y
l
e
:
4
9
.
7
8
%
[
5
]
G
a
l
l
e
r
i
x
d
a
t
a
se
t
o
f
1
6
0
,
0
0
0
i
m
a
g
e
s
G
e
n
r
e
:
8
A
r
t
i
st
:
1
8
S
t
y
l
e
:
2
0
1
3
1
-
l
a
y
e
r
D
P
N
G
e
n
r
e
:
5
2
.
2
7
%
A
r
t
i
st
:
3
4
.
4
%
S
t
y
l
e
:
6
3
.
6
9
%
[
6
]
D
a
t
a
s
e
t
1
:
3
0
8
7
0
D
a
t
a
s
e
t
2
:
8
5
0
0
0
D
a
t
a
s
e
t
3
:
1
9
3
2
0
D
a
t
a
s
e
t
1
:
6
D
a
t
a
s
e
t
2
:
2
3
D
a
t
a
s
e
t
3
:
1
9
C
N
N
a
n
d
sh
a
l
l
o
w
n
e
u
r
a
l
n
e
t
w
o
r
k
D
a
t
a
s
e
t
1
:
6
7
.
1
6
%
D
a
t
a
s
e
t
2
:
6
6
.
7
1
%
D
a
t
a
s
e
t
3
:
7
7
.
5
3
%
[
6
]
D
a
t
a
s
e
t
1
:
3
0
8
7
0
D
a
t
a
s
e
t
2
:
8
5
0
0
0
D
a
t
a
s
e
t
3
:
1
9
3
2
0
D
a
t
a
s
e
t
1
:
6
D
a
t
a
s
e
t
2
:
2
3
D
a
t
a
s
e
t
3
:
1
9
C
N
N
a
n
d
sh
a
l
l
o
w
n
e
u
r
a
l
n
e
t
w
o
r
k
D
a
t
a
s
e
t
1
:
6
7
.
1
6
%
D
a
t
a
s
e
t
2
:
6
6
.
7
1
%
D
a
t
a
s
e
t
3
:
7
7
.
5
3
%
[
7
]
W
i
k
i
A
r
t
7
C
N
N
8
1
.
7
7
%
[
8
]
D
u
n
h
u
a
n
g
mu
r
a
l
d
a
t
a
s
e
t
2
ANN
9
8
%
[
9
]
W
i
k
i
A
r
t
:
2
0
4
0
5
P
a
i
n
t
i
n
g
D
b
:
1
3
0
0
W
i
k
i
A
r
t
:
2
3
P
a
i
n
t
i
n
g
D
b
:
1
3
M
R
F
-
C
N
N
W
i
k
i
A
r
t
:
7
2
.
8
9
%
P
a
i
n
t
i
n
g
D
b
:
7
6
.
9
2
%
[
1
0
]
I
n
d
i
a
n
D
i
g
i
t
a
l
H
e
r
i
t
a
g
e
S
p
a
c
e
M
o
n
u
me
n
t
s
d
a
t
a
:
1
0
8
0
8
0
9
Tr
a
n
sf
e
r
l
e
a
r
n
i
n
g
9
8
.
7
5
%
[
1
1
]
O
w
n
d
a
t
a
b
a
se
o
f
M
a
d
h
u
b
a
n
i
a
r
t
f
o
r
m
s
5
Tr
a
n
sf
e
r
l
e
a
r
n
i
n
g
9
8
.
8
2
%
[
1
2
]
K
a
g
g
l
e
:
1
3
7
6
i
ma
g
e
s
o
f
w
i
d
t
h
a
n
d
h
e
i
g
h
t
o
f
2
2
4
×
2
2
4
2
C
N
N
:
M
o
b
i
l
e
N
e
t
V
2
9
9
.
9
9
%
[
1
3
]
I
S
I
C
-
2
0
1
8
d
a
t
a
se
t
o
f
s
k
i
n
l
e
si
o
n
:
1
0
0
1
5
d
e
r
m
o
sc
o
p
i
c
i
m
a
g
e
s
o
f
t
h
e
l
e
si
o
n
7
C
N
N
:
D
e
n
seN
e
t
a
n
d
I
n
c
e
p
t
o
n
-
v3
8
1
%
3.
P
RO
P
O
SE
D
AP
P
RO
ACH
T
r
an
s
f
er
lear
n
in
g
is
u
s
ed
to
r
ed
u
ce
th
e
tr
ain
in
g
tim
e.
W
e
h
a
v
e
u
s
ed
I
n
ce
p
tio
n
R
esNetV2
,
with
o
u
t th
e
to
p
lay
er
,
as
a
b
ase
m
o
d
el.
I
n
ce
p
tio
n
-
R
esNet
-
v2
[
2
6
]
m
er
g
es
b
o
th
th
e
co
n
ce
p
ts
o
f
I
n
ce
p
t
io
n
-
v
3
an
d
R
esNet
[
2
7
]
a
r
ch
itectu
r
es.
I
n
ce
p
tio
n
-
R
esNet
-
v
2
h
as
a
s
tem
as
in
I
n
ce
p
tio
n
-
v
3
a
n
d
R
esid
u
al
b
lo
ck
s
as
in
th
e
R
esNet
m
o
d
el.
B
u
t,
in
s
id
e
ev
e
r
y
r
esi
d
u
al
b
lo
ck
,
f
ilter
co
n
ca
ten
atio
n
is
ca
r
r
ied
o
u
t
a
n
d
th
e
f
ilter
s
izes
v
ar
y
f
o
r
th
ese
r
esid
u
al
b
lo
ck
s
.
I
n
tr
an
s
f
er
lear
n
in
g
,
th
e
r
e
ar
e
m
ain
ly
two
p
h
ases
:
Featu
r
e
ex
tr
ac
tio
n
f
r
o
m
t
h
e
in
p
u
t
im
a
g
e
an
d
th
e
s
ec
o
n
d
p
h
ase
is
f
in
e
-
tu
n
i
n
g
o
f
th
e
b
ase
m
o
d
el.
As
th
e
tr
ai
n
ed
m
o
d
el,
we
ca
n
u
s
e
th
e
p
r
e
-
tr
ain
ed
weig
h
ts
b
ased
o
n
th
e
av
ailab
l
e
d
ataset
s
.
I
n
o
u
r
m
o
d
el,
we
h
av
e
ch
o
s
en
“im
ag
en
et”
weig
h
ts
.
T
h
e
la
s
t
lay
er
o
f
th
e
b
ase
m
o
d
el
in
clu
d
es
th
e
class
if
icati
o
n
s
o
f
tm
ax
lay
er
.
As
we
wan
t
to
h
av
e
o
th
er
la
y
er
s
o
n
t
o
p
o
f
it,
we
n
ee
d
t
o
s
et
in
clu
d
e_
to
p
=
Fals
e
to
r
em
o
v
e
th
e
class
if
icatio
n
lay
er
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
u
to
ma
tic
mu
lti
-
cla
s
s
cla
s
s
ific
a
tio
n
o
f
g
eo
g
r
a
p
h
ica
l
in
d
ic
a
tio
n
-
ta
g
g
ed
tr
ib
a
l
a
r
t …
(
Ma
la
y
S
.
B
h
a
tt
)
1527
W
h
en
th
e
d
ataset
is
eith
er
n
o
t
av
ailab
le
at
all
o
r
it
is
v
e
r
y
less
,
d
ata
au
g
m
e
n
tatio
n
is
es
s
en
tial
to
m
ak
e
th
e
m
o
d
el
r
o
b
u
s
t,
r
elia
b
le,
an
d
ac
c
u
r
ate
[
2
8
]
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
is
d
ep
icted
in
Fig
u
r
e
3
.
I
t
is
clea
r
f
r
o
m
th
e
f
ig
u
r
e
th
at
all
th
r
ee
s
am
p
le
im
ag
es
h
av
e
a
‘
cir
cle’
as
a
p
r
o
m
in
en
t
s
h
ap
e
an
d
th
ey
also
h
av
e
s
im
ilar
ities
in
ter
m
s
o
f
co
lo
r
s
u
s
ed
.
I
t
is
th
e
m
in
u
te
d
etails
in
ea
ch
th
at
ca
n
s
ep
ar
ate
o
n
e
ar
t
f
o
r
m
f
r
o
m
th
e
o
th
er
.
Fig
u
r
e
3
.
Pro
p
o
s
ed
a
p
p
r
o
ac
h
4.
DATAS
E
T
I
n
th
is
wo
r
k
,
we
h
a
v
e
co
n
s
id
e
r
ed
th
r
ee
d
i
f
f
er
en
t
f
o
lk
ar
ts
,
n
am
ely
Ma
n
ju
s
h
a,
W
ar
li,
an
d
Go
n
d
.
T
h
e
ar
t
is
p
o
p
u
lar
o
n
th
e
c
o
v
er
p
a
g
e
o
f
d
iar
ies,
o
n
th
e
p
en
s
tan
d
,
an
d
in
th
e
wall
clo
ck
b
ac
k
g
r
o
u
n
d
.
So
m
e
o
f
th
e
s
am
p
le
im
ag
es
o
f
p
en
s
tan
d
s
ar
e
s
h
o
wn
in
Fig
u
r
e
4
.
Fig
u
r
e
4
(
a)
is
a
Go
n
d
ar
t
p
e
n
s
tan
d
,
Fig
u
r
e
4
(
b
)
is
Ma
n
ju
s
h
a
ar
t p
en
s
tan
d
,
an
d
Fig
u
r
e
4
(
c)
is
W
ar
li a
r
t p
en
s
tan
d
.
(
a)
(
b
)
(
c)
Fig
u
r
e
4
.
Pen
s
tan
d
s
with
a
r
t
wo
r
k
,
(
a
)
Go
n
d
ar
t p
e
n
s
tan
d
,
(
b
)
Ma
n
ju
s
h
a
a
r
t p
en
s
tan
d
,
an
d
(
c)
W
ar
li a
r
t p
e
n
s
tan
d
T
h
er
e
is
n
o
r
ea
d
ily
av
ailab
le
d
ataset,
s
o
we
h
av
e
m
an
u
ally
cr
ea
ted
a
d
ataset.
T
h
e
d
ataset
co
n
s
is
ts
o
f
3
0
0
im
ag
es
f
o
r
ea
ch
ar
two
r
k
an
d
th
ese
im
ag
es
ar
e
d
o
wn
l
o
a
d
ed
f
r
o
m
v
ar
io
u
s
web
s
ites
.
I
n
to
tal,
th
er
e
ar
e
9
0
0
im
ag
es.
Dee
p
C
NNs
n
ee
d
a
lar
g
e
d
ataset
wh
ich
is
n
o
t
p
o
s
s
ib
le
to
cr
ea
te
m
an
u
ally
s
o
we
h
av
e
u
s
ed
d
ata
au
g
m
en
tatio
n
tec
h
n
iq
u
es
to
in
cr
ea
s
e
th
e
s
ize
o
f
th
e
d
ataset.
Data
au
g
m
en
tatio
n
m
ak
es
th
e
m
o
d
el
m
o
r
e
r
o
b
u
s
t
wh
ich
ca
n
ef
f
icien
tly
h
an
d
le
n
ew
im
ag
es
d
u
r
in
g
th
e
test
in
g
.
W
e
h
av
e
ap
p
lied
r
o
tatio
n
,
f
lip
p
in
g
in
x
an
d
y
d
ir
ec
tio
n
s
,
z
o
o
m
in
g
,
s
h
ea
r
tr
an
s
f
o
r
m
atio
n
in
x
an
d
y
d
ir
ec
tio
n
s
,
an
d
cr
o
p
p
i
n
g
a
lg
o
r
ith
m
s
f
o
r
d
ata
au
g
m
en
tatio
n
.
T
h
e
d
ata
au
g
m
en
tatio
n
g
iv
es a
r
eliab
le
n
u
m
b
er
o
f
s
am
p
les f
o
r
ea
ch
tr
i
b
al
ar
t.
T
o
u
s
e
th
e
b
ase
m
o
d
el,
we
h
av
e
u
s
ed
tf
.
k
e
r
as.a
p
p
licati
o
n
s
wh
ich
co
n
tain
m
a
n
y
s
u
cc
ess
f
u
l
m
o
d
els.
T
h
e
p
ap
er
u
s
es
I
n
ce
p
tio
n
-
R
e
s
Net
-
v
2
as
a
b
ase
m
o
d
el
an
d
ap
p
lies
a
tr
an
s
f
er
lear
n
in
g
ap
p
r
o
ac
h
with
d
ata
au
g
m
en
tatio
n
tech
n
iq
u
es f
o
r
c
lass
if
icatio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
5
2
3
-
1
5
3
1
1528
T
h
e
d
ataset
is
d
iv
id
ed
in
to
tr
a
in
in
g
,
v
alid
atio
n
,
an
d
test
in
g
.
Ou
t
o
f
9
0
0
im
ag
es,
1
8
0
im
a
g
es
(
2
0
%)
ar
e
u
s
ed
f
o
r
v
alid
atio
n
.
T
h
er
e
is
a
s
ep
ar
ate
d
ataset
o
f
1
8
0
im
ag
es
(
6
0
im
ag
es
f
o
r
ea
c
h
class
)
f
o
r
t
h
e
test
in
g
o
f
th
e
m
o
d
el.
T
h
e
s
u
g
g
ested
C
NN
ar
ch
itectu
r
e
is
s
h
o
wn
in
Fig
u
r
e
5
.
I
n
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
s
tep
,
th
e
w
h
o
le
d
ataset
is
r
esized
to
3
3
1
×
3
3
1
t
o
m
ak
e
it su
itab
le
f
o
r
th
e
i
n
p
u
t
lay
er
.
Fig
u
r
e
5
.
Pro
p
o
s
ed
m
o
d
el
wit
h
tr
an
s
f
er
lear
n
in
g
5.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
W
e
h
av
e
ac
h
iev
ed
9
9
%
tr
ain
i
n
g
ac
cu
r
ac
y
an
d
9
8
.
2
%
test
in
g
ac
cu
r
ac
y
.
T
h
e
o
p
tim
al
ep
o
c
h
s
u
s
ed
in
th
is
ex
p
er
im
en
t
a
r
e
1
0
.
‘
Ad
a
m
’
o
p
tim
izer
is
u
s
ed
with
ea
r
ly
s
to
p
p
in
g
cr
iter
ia.
T
h
e
g
r
a
p
h
s
h
o
wn
in
Fig
u
r
e
6
r
ep
r
esen
ts
ep
o
ch
s
v
s
tr
ain
in
g
an
d
v
alid
atio
n
lo
s
s
es.
T
h
e
co
n
f
u
s
io
n
m
atr
i
x
was
o
b
tain
ed
f
o
r
1
8
0
test
in
g
im
ag
es.
Ou
t
o
f
1
8
0
im
ag
es,
o
n
e
im
ag
e
o
f
th
e
g
o
n
d
class
an
d
th
r
ee
im
ag
es
o
f
th
e
W
ar
li
cla
s
s
wer
e
m
is
class
if
ied
as sh
o
wn
in
Fig
u
r
e
7
.
Fig
u
r
e
6
.
T
r
ain
in
g
l
o
s
s
f
o
r
1
0
ep
o
ch
s
f
o
r
in
ce
p
tio
n
R
esNet
V2
with
tr
an
s
f
er
lear
n
i
n
g
Fig
u
r
e
8
p
r
esen
ts
th
e
m
is
c
lass
if
ied
im
ag
es
id
en
tifie
d
i
n
th
e
test
in
g
p
h
ase.
E
ac
h
s
u
b
f
ig
u
r
e
co
r
r
esp
o
n
d
s
to
an
ar
two
r
k
in
c
o
r
r
ec
tly
lab
ele
d
b
y
th
e
m
o
d
el,
h
elp
in
g
an
aly
ze
t
h
e
v
is
u
al
att
r
ib
u
tes
co
n
tr
ib
u
tin
g
to
m
is
class
if
icatio
n
.
Fig
u
r
e
s
8
(
a)
-
(
d
)
illu
s
tr
ate
th
ese
ca
s
es a
cr
o
s
s
Go
n
d
,
W
ar
li,
an
d
Ma
n
ju
s
h
a
ca
teg
o
r
ies.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
u
to
ma
tic
mu
lti
-
cla
s
s
cla
s
s
ific
a
tio
n
o
f
g
eo
g
r
a
p
h
ica
l
in
d
ic
a
tio
n
-
ta
g
g
ed
tr
ib
a
l
a
r
t …
(
Ma
la
y
S
.
B
h
a
tt
)
1529
Fig
u
r
e
7
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
test
in
g
d
ataset
Fig
u
r
e
8
(
a)
,
wh
ic
h
b
elo
n
g
s
to
class
‘
Go
n
d
’
,
is
m
is
clas
s
if
ied
as
‘
W
ar
li’.
Fig
u
r
e
s
8
(
b
)
-
8
(
d
)
b
elo
n
g
to
class
‘
W
ar
li’.
Fig
u
r
e
8
(
b
)
is
m
is
class
if
ied
as
‘
Go
n
d
’
wh
ile
th
e
r
em
ain
in
g
two
im
ag
es
ar
e
m
is
clas
s
if
ied
as
‘
Ma
n
ju
s
h
a’
.
I
n
Fig
u
r
e
8
(
a)
,
as
we
zo
o
m
in
,
ev
en
th
o
u
g
h
it
co
n
tain
s
lo
ts
o
f
s
m
all
*
’
a
n
d
s
tr
aig
h
t
lin
es,
it
cr
ea
tes
an
illu
s
io
n
o
f
g
eo
m
etr
ic
s
h
ap
es
s
u
ch
as
tr
ian
g
les
an
d
cir
cles.
T
h
at
m
ay
h
av
e
tr
ig
g
er
e
d
th
e
m
is
class
if
icatio
n
as ‘
W
ar
li’.
I
n
Fig
u
r
e
8
(
b
)
th
e
W
ar
li c
h
ar
ac
ter
s
ar
e
in
th
e
ce
n
ter
wh
ile
th
e
p
r
o
m
in
en
t a
r
ea
h
as
n
o
d
etails
o
f
th
e
W
ar
li
ar
t
f
o
r
m
.
T
h
e
co
lo
r
s
ar
e
u
s
ed
h
ea
v
il
y
f
o
r
th
e
d
em
o
n
s
tr
atio
n
o
f
d
if
f
er
en
t
f
esti
v
als
in
Go
n
d
p
ain
tin
g
s
.
T
h
e
s
am
e
is
p
r
esen
t
in
Fig
u
r
e
8
(
b
)
lea
d
in
g
to
th
e
m
is
class
if
icatio
n
as
Go
n
d
in
s
tead
o
f
W
ar
li
.
Fig
u
r
e
8
(
c
)
h
as
s
h
ad
es
o
f
g
r
e
en
co
lo
u
r
in
th
e
m
aj
o
r
ity
o
f
t
h
e
im
ag
e
a
r
ea
.
I
t
is
a
h
ig
h
lig
h
tin
g
f
ea
tu
r
e
o
f
th
e
Ma
n
ju
s
h
a
ar
t
f
o
r
m
.
T
h
is
co
u
l
d
b
e
th
e
r
ea
s
o
n
f
o
r
th
e
m
is
class
if
icatio
n
.
Sam
e
f
o
r
Fig
u
r
e
8
(
d
)
it
is
ab
u
n
d
an
tl
y
y
ello
w,
o
n
e
o
f
th
e
h
ea
v
ily
u
s
ed
co
lo
u
r
s
in
th
e
Ma
n
ju
s
h
a
ar
t f
o
r
m
.
A
c
t
u
a
l
:
D
o
n
d
P
r
e
d
i
c
t
e
d
:
W
a
r
l
i
A
c
t
u
a
l
:
W
a
r
l
i
P
r
e
d
i
c
t
e
d
:
G
o
n
d
A
c
t
u
a
l
:
D
o
n
d
P
r
e
d
i
c
t
e
d
:
M
a
n
j
u
s
h
a
A
c
t
u
a
l
:
W
a
r
l
i
P
r
e
d
i
c
t
e
d
:
M
a
n
j
u
s
h
a
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
8
.
Misclass
if
ied
i
m
ag
es
;
(
a)
g
o
n
d
,
(
b)
W
ar
li,
(
c)
W
ar
li
,
an
d
(
d
)
W
ar
li
6.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
p
r
o
p
o
s
ed
a
d
ee
p
lear
n
in
g
–
b
ased
f
r
am
ewo
r
k
f
o
r
th
e
au
to
m
atic
class
if
icat
io
n
o
f
th
r
ee
I
n
d
ian
tr
ib
al
a
r
t
f
o
r
m
s
u
s
in
g
t
r
an
s
f
er
lear
n
in
g
o
n
t
h
e
I
n
ce
p
tio
n
-
R
esNet
-
V2
ar
ch
itectu
r
e.
A
m
an
u
ally
cu
r
ate
d
d
ataset
o
f
9
0
0
im
ag
es
was
ex
p
an
d
ed
u
s
in
g
d
ata
au
g
m
en
tatio
n
,
en
ab
lin
g
th
e
m
o
d
el
to
ac
h
iev
e
a
h
ig
h
test
in
g
ac
cu
r
ac
y
o
f
9
8
.
2
1
%.
T
h
ese
r
e
s
u
lts
d
em
o
n
s
tr
ate
th
e
ef
f
ec
tiv
en
ess
o
f
d
ee
p
lear
n
in
g
f
o
r
d
is
tin
g
u
is
h
in
g
v
is
u
ally
s
im
ilar
tr
ad
itio
n
al
ar
t f
o
r
m
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
1
5
,
No
.
4
,
D
e
c
em
b
er
20
2
6
:
1
5
2
3
-
1
5
3
1
1530
B
ey
o
n
d
p
er
f
o
r
m
a
n
ce
,
th
e
f
in
d
in
g
s
co
n
tr
ib
u
te
to
c
u
ltu
r
al
in
f
o
r
m
atics
b
y
p
r
o
v
id
in
g
a
b
aselin
e
d
ataset
an
d
m
et
h
o
d
o
lo
g
y
f
o
r
tr
ib
al
a
r
t
class
if
icatio
n
.
Ho
wev
er
,
th
e
s
tu
d
y
is
lim
ited
b
y
d
a
taset
s
iz
e,
class
im
b
alan
ce
r
is
k
s
,
an
d
r
elian
ce
o
n
2
D
im
a
g
e
v
ar
iatio
n
s
.
Fu
tu
r
e
r
esear
ch
m
ay
e
x
ten
d
th
is
wo
r
k
b
y
e
x
p
an
d
in
g
th
e
d
ataset,
ex
p
lo
r
in
g
g
en
er
ativ
e
au
g
m
e
n
tatio
n
(
GANs)
,
i
n
teg
r
atin
g
ex
p
lain
a
b
le
AI
m
eth
o
d
s
to
h
ig
h
lig
h
t
d
ec
is
io
n
r
eg
io
n
s
,
an
d
d
ev
elo
p
in
g
m
o
b
ile
ap
p
licatio
n
s
to
s
u
p
p
o
r
t r
ea
l
-
t
im
e
ar
t r
ec
o
g
n
itio
n
an
d
ed
u
ca
tio
n
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
e
au
th
o
r
s
s
tate
th
at
n
o
f
u
n
d
i
n
g
was
in
v
o
lv
e
d
.
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
Ma
lay
S.
B
h
att
✓
✓
✓
✓
✓
✓
✓
✓
✓
Ap
u
r
v
a
A.
Me
h
ta
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
ter
est.
DATA AV
AI
L
AB
I
L
I
T
Y
Data
av
ailab
ilit
y
d
o
es n
o
t a
p
p
l
y
to
th
is
p
ap
e
r
as n
o
n
ew
d
ata
wer
e
cr
ea
ted
o
r
an
aly
ze
d
in
th
is
s
tu
d
y
.
RE
F
E
R
E
NC
E
S
[
1
]
S
.
S
o
n
e
r
,
R
.
Li
t
o
r
i
y
a
,
R
.
K
h
a
t
r
i
,
A
.
A
.
H
u
ss
a
i
n
,
S
.
P
a
g
a
r
e
,
a
n
d
S
.
K
.
K
u
s
h
w
a
h
a
,
“
R
e
a
l
-
t
i
m
e
f
a
c
e
m
a
sk
d
e
t
e
c
t
i
o
n
i
n
m
a
ss g
a
t
h
e
r
i
n
g
t
o
r
e
d
u
c
e
C
O
V
I
D
-
1
9
s
p
r
e
a
d
,
”
J
o
u
rn
a
l
o
f
A
u
t
o
m
a
t
i
o
n
,
M
o
b
i
l
e
Ro
b
o
t
i
c
s
a
n
d
I
n
t
e
l
l
i
g
e
n
t
S
y
st
e
m
s
,
p
p
.
5
1
–
5
8
,
2
0
2
3
,
d
o
i
:
1
0
.
1
4
3
1
3
/
j
a
mr
i
s/
1
-
2
0
2
3
/
7
.
[
2
]
R
.
C
h
a
n
d
r
a
a
n
d
M
.
H
a
j
i
a
r
b
a
b
i
,
“
S
k
i
n
l
e
s
i
o
n
d
e
t
e
c
t
i
o
n
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
”
J
o
u
r
n
a
l
o
f
A
u
t
o
m
a
t
i
o
n
,
Mo
b
i
l
e
Ro
b
o
t
i
c
s
a
n
d
In
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
,
v
o
l
.
1
6
,
n
o
.
3
,
p
p
.
5
6
–
6
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
4
3
1
3
/
j
a
mr
i
s/
3
-
2
0
2
2
/
2
4
.
[
3
]
S
.
Li
o
n
g
,
Y
.
H
u
a
n
g
,
S
.
Li
,
Z.
H
u
a
n
g
,
J.
M
a
,
a
n
d
Y
.
S
.
G
a
n
,
“
A
u
t
o
mat
i
c
t
r
a
d
i
t
i
o
n
a
l
C
h
i
n
e
s
e
p
a
i
n
t
i
n
g
c
l
a
ss
i
f
i
c
a
t
i
o
n
:
A
b
e
n
c
h
mar
k
i
n
g
a
n
a
l
y
si
s,
”
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
6
,
n
o
.
3
,
p
p
.
1
1
8
3
–
1
1
9
9
,
A
u
g
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
1
1
/
c
o
i
n
.
1
2
3
2
8
.
[
4
]
L.
Q
i
a
o
,
X
.
G
u
o
,
a
n
d
W
.
Li
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
C
h
i
n
e
s
e
a
n
d
w
e
s
t
e
r
n
p
a
i
n
t
i
n
g
i
ma
g
e
s
b
a
se
d
o
n
b
r
u
sh
s
t
r
o
k
e
s
f
e
a
t
u
r
e
,
”
L
e
c
t
u
re
N
o
t
e
s
i
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
2
5
2
3
LN
C
S
,
p
p
.
3
2
5
–
3
3
7
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
0
-
6
5
7
3
6
-
9
_
3
0
.
[
5
]
M
.
G
h
o
s
h
,
S
.
M
.
O
b
a
i
d
u
l
l
a
h
,
F
.
G
h
e
r
a
r
d
i
n
i
,
a
n
d
M
.
Zd
i
ma
l
o
v
a
,
“
C
l
a
ssi
f
i
c
a
t
i
o
n
o
f
g
e
o
met
r
i
c
f
o
r
ms
i
n
mo
s
a
i
c
s
u
si
n
g
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
J
o
u
r
n
a
l
o
f
I
m
a
g
i
n
g
,
v
o
l
.
7
,
n
o
.
8
,
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
j
i
m
a
g
i
n
g
7
0
8
0
1
4
9
.
[
6
]
F
.
X
u
e
t
a
l
.
,
“
Ex
t
e
n
si
v
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
v
i
s
u
a
l
a
r
t
p
a
i
n
t
i
n
g
s
f
o
r
e
n
h
a
n
c
i
n
g
e
d
u
c
a
t
i
o
n
s
y
st
e
m
u
s
i
n
g
h
y
b
r
i
d
S
V
M
-
A
N
N
w
i
t
h
s
p
a
r
se
met
r
i
c
l
e
a
r
n
i
n
g
b
a
se
d
o
n
k
e
r
n
e
l
r
e
g
r
e
s
si
o
n
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
I
n
t
e
r
a
c
t
i
v
e
M
u
l
t
i
m
e
d
i
a
a
n
d
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
7
,
n
o
.
2
,
p
p
.
2
2
4
–
2
3
1
,
2
0
2
1
,
d
o
i
:
1
0
.
9
7
8
1
/
i
j
i
ma
i
.
2
0
2
1
.
1
0
.
0
0
1
.
[
7
]
S
.
h
u
a
Z
h
o
n
g
,
X
.
H
u
a
n
g
,
a
n
d
Z.
X
i
a
o
,
“
F
i
n
e
-
a
r
t
p
a
i
n
t
i
n
g
c
l
a
ss
i
f
i
c
a
t
i
o
n
v
i
a
t
w
o
-
c
h
a
n
n
e
l
d
u
a
l
p
a
t
h
n
e
t
w
o
r
k
s,”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
a
n
d
C
y
b
e
r
n
e
t
i
c
s
,
v
o
l
.
1
1
,
n
o
.
1
,
p
p
.
1
3
7
–
1
5
2
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
3
0
4
2
-
0
1
9
-
0
0
9
6
3
-
0.
[8
]
C
.
S
a
n
d
o
v
a
l
,
E
.
P
i
r
o
g
o
v
a
,
a
n
d
M
.
L
e
c
h
,
“
Tw
o
-
st
a
g
e
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
t
o
t
h
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
f
i
n
e
-
a
r
t
p
a
i
n
t
i
n
g
s,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
7
,
p
p
.
4
1
7
7
0
–
4
1
7
8
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
1
9
.
2
9
0
7
9
8
6
.
[
9
]
J.
D
u
B
o
i
s,
“
U
si
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
t
o
c
l
a
ssi
f
y
a
r
t
g
e
n
r
e
,
”
S
t
a
n
f
o
r
d
U
n
i
v
.
,
2
0
1
5
.
[
1
0
]
X
.
L
i
u
,
“
A
r
t
p
a
i
n
t
i
n
g
i
ma
g
e
c
l
a
s
si
f
i
c
a
t
i
o
n
b
a
se
d
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
C
o
m
p
u
t
a
t
i
o
n
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
N
e
u
r
o
s
c
i
e
n
c
e
,
v
o
l
.
2
0
2
2
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
2
/
3
1
1
9
6
0
4
.
[
1
1
]
K
.
L.
H
u
a
,
T.
T
.
H
o
,
K
.
A
.
J
a
n
g
t
j
i
k
,
Y
.
J.
C
h
e
n
,
a
n
d
M
.
C
.
Y
e
h
,
“
A
r
t
i
s
t
-
b
a
sed
p
a
i
n
t
i
n
g
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
M
a
r
k
o
v
r
a
n
d
o
m
f
i
e
l
d
s
w
i
t
h
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
Mu
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
7
9
,
n
o
.
1
7
–
1
8
,
p
p
.
1
2
6
3
5
–
1
2
6
5
8
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
0
4
2
-
0
1
9
-
0
8
5
4
7
-
4.
[
1
2
]
U
.
K
u
l
k
a
r
n
i
,
S
.
M
.
M
e
e
n
a
,
S
.
V
.
G
u
r
l
a
h
o
su
r
,
a
n
d
U
.
M
u
d
e
n
g
u
d
i
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
c
u
l
t
u
r
a
l
h
e
r
i
t
a
g
e
si
t
e
s u
s
i
n
g
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
,
”
Pro
c
e
e
d
i
n
g
s
-
2
0
1
9
I
E
EE
5
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
M
u
l
t
i
m
e
d
i
a
Bi
g
D
a
t
a
,
Bi
g
M
M
2
0
1
9
,
p
p
.
3
9
1
–
3
9
7
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
B
i
g
M
M
.
2
0
1
9
.
0
0
0
2
0
.
[
1
3
]
S
.
V
a
r
sh
n
e
y
,
C
.
V
.
La
k
s
h
mi
,
a
n
d
C
.
P
a
t
v
a
r
d
h
a
n
,
“
M
a
d
h
u
b
a
n
i
A
r
t
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
t
r
a
n
sf
e
r
l
e
a
r
n
i
n
g
w
i
t
h
d
e
e
p
f
e
a
t
u
r
e
f
u
s
i
o
n
a
n
d
d
e
c
i
si
o
n
f
u
s
i
o
n
-
b
a
se
d
t
e
c
h
n
i
q
u
e
s,
”
E
n
g
i
n
e
e
r
i
n
g
A
p
p
l
i
c
a
t
i
o
n
s
o
f
A
rt
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
1
9
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
n
g
a
p
p
a
i
.
2
0
2
2
.
1
0
5
7
3
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
A
u
to
ma
tic
mu
lti
-
cla
s
s
cla
s
s
ific
a
tio
n
o
f
g
eo
g
r
a
p
h
ica
l
in
d
ic
a
tio
n
-
ta
g
g
ed
tr
ib
a
l
a
r
t …
(
Ma
la
y
S
.
B
h
a
tt
)
1531
[
1
4
]
J.
S
h
a
r
m
a
,
“
Tr
i
b
a
l
a
r
t
s
o
f
I
n
d
i
a
,
”
H
o
l
i
d
i
f
y
,
2
0
2
6
.
h
t
t
p
s
:
/
/
w
w
w
.
h
o
l
i
d
i
f
y
.
c
o
m
/
p
a
g
e
s
/
t
r
i
b
a
l
-
a
r
t
-
of
-
i
n
d
i
a
-
4
4
9
7
.
h
t
m
l
(
a
c
c
e
ss
e
d
A
u
g
.
1
8
,
2
0
2
6
)
.
[
1
5
]
U
p
e
n
d
r
a
M
a
h
a
r
a
t
h
i
S
h
i
l
p
A
n
u
sa
n
d
h
a
n
S
a
n
st
h
a
n
,
“
M
a
n
j
u
sh
a
A
r
t
,
”
2
0
2
6
.
h
t
t
p
:
/
/
u
ms
a
s.
o
r
g
.
i
n
/
e
n
/
c
r
a
f
t
s/
m
a
n
j
u
sh
a
-
a
r
t
(
a
c
c
e
sse
d
A
u
g
.
1
8
,
2
0
2
6
)
.
[
1
6
]
I
.
S
a
n
t
o
s,
L
.
C
a
s
t
r
o
,
N
.
R
o
d
r
i
g
u
e
z
-
F
e
r
n
a
n
d
e
z
,
Á
.
T
o
r
r
e
n
t
e
-
P
a
t
i
ñ
o
,
a
n
d
A
.
C
a
r
b
a
l
l
a
l
,
“
A
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
i
n
t
h
e
v
i
su
a
l
a
r
t
s
:
a
r
e
v
i
e
w
,
”
N
e
u
r
a
l
C
o
m
p
u
t
i
n
g
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
3
3
,
n
o
.
1
,
p
p
.
1
2
1
–
1
5
7
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s0
0
5
2
1
-
020
-
0
5
5
6
5
-
4.
[
1
7
]
C
.
H
e
r
n
a
n
d
e
z
-
O
l
i
v
a
n
a
n
d
J.
R
.
B
e
l
t
r
á
n
,
“
M
u
s
i
c
c
o
mp
o
s
i
t
i
o
n
w
i
t
h
d
e
e
p
l
e
a
r
n
i
n
g
:
a
r
e
v
i
e
w
,
”
S
i
g
n
a
l
s
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
p
p
.
2
5
–
5
0
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
1
8
4
4
4
-
4
_
2
.
[
1
8
]
M
.
H
.
S
a
l
e
e
m,
J
.
P
o
t
g
i
e
t
e
r
,
a
n
d
K
.
M
.
A
r
i
f
,
“
A
u
t
o
m
a
t
i
o
n
i
n
a
g
r
i
c
u
l
t
u
r
e
b
y
ma
c
h
i
n
e
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s:
a
r
e
v
i
e
w
o
f
r
e
c
e
n
t
d
e
v
e
l
o
p
me
n
t
s
,
”
Pre
c
i
si
o
n
A
g
r
i
c
u
l
t
u
re
,
v
o
l
.
2
2
,
n
o
.
6
,
p
p
.
2
0
5
3
–
2
0
9
1
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
1
1
9
-
021
-
0
9
8
0
6
-
x.
[
1
9
]
C
.
B
h
a
t
t
,
I
.
K
u
mar,
V
.
V
i
j
a
y
a
k
u
m
a
r
,
K
.
U
.
S
i
n
g
h
,
a
n
d
A
.
K
u
mar,
“
T
h
e
s
t
a
t
e
o
f
t
h
e
a
r
t
o
f
d
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
s i
n
me
d
i
c
a
l
sc
i
e
n
c
e
a
n
d
t
h
e
i
r
c
h
a
l
l
e
n
g
e
s,
”
M
u
l
t
i
m
e
d
i
a
S
y
s
t
e
m
s
,
v
o
l
.
2
7
,
n
o
.
4
,
p
p
.
5
9
9
–
6
1
3
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s0
0
5
3
0
-
0
2
0
-
0
0
6
9
4
-
1.
[
2
0
]
X
.
L
i
u
,
“
D
e
e
p
l
e
a
r
n
i
n
g
i
n
m
a
r
k
e
t
i
n
g
:
a
r
e
v
i
e
w
a
n
d
r
e
s
e
a
r
c
h
a
g
e
n
d
a
,
”
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
i
n
M
a
rk
e
t
i
n
g
,
p
p
.
2
3
9
–
2
7
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
8
/
s
1
5
4
8
-
6
4
3
5
2
0
2
3
0
0
0
0
0
2
0
0
1
4
.
[
2
1
]
S
.
P
r
a
d
e
e
p
a
n
d
Y
.
K
.
S
h
a
r
ma,
“
D
e
e
p
l
e
a
r
n
i
n
g
b
a
s
e
d
r
e
a
l
t
i
me
o
b
j
e
c
t
r
e
c
o
g
n
i
t
i
o
n
f
o
r
s
e
c
u
r
i
t
y
i
n
a
i
r
d
e
f
e
n
se
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
1
9
6
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
i
n
g
f
o
r
S
u
s
t
a
i
n
a
b
l
e
G
l
o
b
a
l
D
e
v
e
l
o
p
m
e
n
t
,
I
N
D
I
A
C
o
m
2
0
1
9
,
2
0
1
9
,
p
p
.
2
9
5
–
2
9
8
.
[
2
2
]
H
.
H
i
a
r
y
,
H
.
S
a
a
d
e
h
,
M
.
S
a
a
d
e
h
,
a
n
d
M
.
Y
a
q
u
b
,
“
F
l
o
w
e
r
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
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,
”
I
ET
C
o
m
p
u
t
e
r
Vi
si
o
n
,
v
o
l
.
1
2
,
n
o
.
6
,
p
p
.
8
5
5
–
8
6
2
,
2
0
1
8
,
d
o
i
:
1
0
.
1
0
4
9
/
i
e
t
-
c
v
i
.
2
0
1
7
.
0
1
5
5
.
[
2
3
]
J.
X
.
S
h
e
e
,
K
.
M
.
Li
m,
C
.
P
.
Le
e
,
a
n
d
J.
Y
.
Li
m
,
“
F
l
o
w
e
r
sp
e
c
i
e
s
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
D
e
n
s
e
N
e
t
2
0
1
a
n
d
mu
l
t
i
l
a
y
e
r
p
e
r
c
e
p
t
r
o
n
,
”
2
0
2
3
1
1
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
I
n
f
o
rm
a
t
i
o
n
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
I
C
o
I
C
T
2
0
2
3
,
v
o
l
.
2
0
2
3
-
A
u
g
u
s,
p
p
.
3
0
7
–
3
1
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
o
I
C
T
5
8
2
0
2
.
2
0
2
3
.
1
0
2
6
2
5
9
3
.
[
2
4
]
Z.
N
.
A
b
d
u
l
l
a
h
,
Z.
A
.
F
f
i
g
A
b
u
t
i
h
e
e
n
,
A
.
A
.
A
b
d
u
l
m
u
n
e
m
,
a
n
d
Z
.
A
.
H
a
r
j
a
n
,
“
O
f
f
i
c
i
a
l
l
o
g
o
r
e
c
o
g
n
i
t
i
o
n
b
a
s
e
d
o
n
m
u
l
t
i
l
a
y
e
r
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
m
o
d
e
l
,
”
T
e
l
k
o
m
n
i
k
a
(
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
C
o
m
p
u
t
i
n
g
E
l
e
c
t
r
o
n
i
c
s
a
n
d
C
o
n
t
r
o
l
)
,
v
o
l
.
2
0
,
n
o
.
5
,
p
p
.
1
0
8
3
–
1
0
9
0
,
2
0
2
2
,
d
o
i
:
1
0
.
1
2
9
2
8
/
T
EL
K
O
M
N
I
K
A
.
v
2
0
i
5
.
2
3
4
6
4
.
[
2
5
]
C
.
S
h
o
r
t
e
n
a
n
d
T.
M
.
K
h
o
s
h
g
o
f
t
a
a
r
,
“
A
su
r
v
e
y
o
n
i
ma
g
e
d
a
t
a
a
u
g
m
e
n
t
a
t
i
o
n
f
o
r
d
e
e
p
l
e
a
r
n
i
n
g
,
”
J
o
u
r
n
a
l
o
f
Bi
g
D
a
t
a
,
v
o
l
.
6
,
n
o
.
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
8
6
/
s4
0
5
3
7
-
0
1
9
-
0
1
9
7
-
0.
[
2
6
]
C
.
S
z
e
g
e
d
y
,
S
.
I
o
f
f
e
,
V
.
V
a
n
h
o
u
c
k
e
,
a
n
d
A
.
A
.
A
l
e
m
i
,
“
I
n
c
e
p
t
i
o
n
-
v
4
,
i
n
c
e
p
t
i
o
n
-
R
e
sN
e
t
a
n
d
t
h
e
i
mp
a
c
t
o
f
r
e
si
d
u
a
l
c
o
n
n
e
c
t
i
o
n
s
o
n
l
e
a
r
n
i
n
g
,
”
3
1
st
AA
AI
C
o
n
f
e
re
n
c
e
o
n
A
rt
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
AAAI
2
0
1
7
,
p
p
.
4
2
7
8
–
4
2
8
4
,
2
0
1
7
,
d
o
i
:
1
0
.
1
6
0
9
/
a
a
a
i
.
v
3
1
i
1
.
1
1
2
3
1
.
[
2
7
]
K
.
H
e
,
X
.
Z
h
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
,
”
P
r
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
s
i
o
n
a
n
d
Pa
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
v
o
l
.
2
0
1
6
-
D
e
c
e
m,
p
p
.
7
7
0
–
7
7
8
,
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
9
0
.
[
2
8
]
S
.
A
n
u
b
h
a
P
e
a
r
l
i
n
e
,
V
.
S
a
t
h
i
e
sh
K
u
mar,
a
n
d
S
.
H
a
r
i
n
i
,
“
A
st
u
d
y
o
n
p
l
a
n
t
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
c
o
n
v
e
n
t
i
o
n
a
l
i
ma
g
e
p
r
o
c
e
ss
i
n
g
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s,
”
J
o
u
r
n
a
l
o
f
I
n
t
e
l
l
i
g
e
n
t
a
n
d
F
u
zz
y
S
y
st
e
m
s
,
v
o
l
.
3
6
,
n
o
.
3
,
p
p
.
1
9
9
7
–
2
0
0
4
,
2
0
1
9
,
d
o
i
:
1
0
.
3
2
3
3
/
JI
F
S
-
1
6
9
9
1
1
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Pro
f.
M
a
la
y
S
.
B
h
a
tt
wa
s
b
o
r
n
a
t
B
h
u
j
in
G
u
jara
t,
In
d
ia
o
n
2
0
t
h
F
e
b
r
u
a
ry
,
1
9
8
1
.
P
ro
f.
B
h
a
tt
c
o
m
p
lete
d
Ba
c
h
e
lo
r
o
f
En
g
in
e
e
rin
g
wit
h
sp
e
c
ializa
ti
o
n
in
C
o
m
p
u
ter
E
n
g
i
n
e
e
rin
g
fro
m
C.
U.
S
h
a
h
C
o
ll
e
g
e
o
f
En
g
in
e
e
rin
g
a
n
d
Tec
h
n
o
l
o
g
y
,
S
u
re
n
d
ra
n
a
g
a
r,
G
u
jara
t,
In
d
ia
in
th
e
y
e
a
r
2
0
0
2
.
P
r
o
f.
B
h
a
tt
c
o
m
p
lete
d
M
a
ste
rs
in
En
g
in
e
e
ri
n
g
wi
th
c
o
m
p
u
ter
e
n
g
in
e
e
ri
n
g
a
s
a
sp
e
c
ializa
ti
o
n
fro
m
Dh
a
rm
sin
h
D
e
sa
i
Un
iv
e
rsity
,
Na
d
ia
d
,
G
u
jara
t,
In
d
ia i
n
th
e
y
e
a
r
2
0
0
4
.
P
ro
f
.
Bh
a
tt
h
a
s
c
o
m
p
lete
d
P
h
.
D.
in
Co
m
p
u
ter
En
g
in
e
e
rin
g
fr
o
m
Ra
i
Un
i
v
e
rsity
,
A
h
m
e
d
a
b
a
d
,
G
u
jara
t,
in
th
e
y
e
a
r
2
0
1
7
.
Th
e
a
u
th
o
r’s
m
a
jo
r
field
o
f
stu
d
y
c
o
v
e
rs
Im
a
g
e
P
ro
c
e
ss
in
g
a
n
d
M
u
lt
ime
d
ia
In
f
o
rm
a
ti
o
n
Re
tri
e
v
a
l.
He
h
a
s
b
e
e
n
wo
rk
in
g
a
s
As
so
c
iate
P
ro
fe
ss
o
r
in
th
e
Co
m
p
u
ter
E
n
g
i
n
e
e
rin
g
De
p
a
rtm
e
n
t
a
t
Dh
a
rm
sin
h
De
sa
i
Un
iv
e
rs
it
y
,
Na
d
iad
,
G
u
jara
t,
In
d
ia
sin
c
e
2
0
0
8
.
He
d
e
li
v
e
rs
e
x
p
e
rt
ta
lk
s
a
n
d
se
rv
e
s
a
s
a
re
v
iew
e
r
in
in
tern
a
ti
o
n
a
l
c
o
n
fe
re
n
c
e
s
a
n
d
jo
u
r
n
a
ls.
He
h
a
s
a
lso
p
re
se
n
ted
p
a
p
e
rs
in
in
tern
a
ti
o
n
a
l
Jo
u
r
n
a
ls,
IEE
E
c
o
n
fe
re
n
c
e
s,
Na
ti
o
n
a
l
Jo
u
rn
a
ls
,
a
n
d
Na
ti
o
n
a
l
C
o
n
fe
re
n
c
e
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
sb
c
e
_
s
p
c
e
@s
p
u
.
a
c
.
in
.
Pro
f.
Apu
r
v
a
A.
Mehta
wa
s
b
o
r
n
a
t
Bh
a
v
n
a
g
a
r
i
n
G
u
jara
t,
In
d
ia
o
n
1
0
t
h
Ju
n
e
,
1
9
9
0
.
He
c
o
m
p
lete
d
B
.
E.
,
M
.
Tec
h
.
a
n
d
P
h
.
D.
with
a
sp
e
c
ializa
ti
o
n
i
n
Co
m
p
u
ter
E
n
g
in
e
e
rin
g
fro
m
Dh
a
rm
sin
h
De
sa
i
Un
i
v
e
rsit
y
,
Na
d
ia
d
,
G
u
jara
t,
I
n
d
ia
in
t
h
e
y
e
a
rs
2
0
1
1
,
2
0
1
4
,
a
n
d
2
0
2
2
re
sp
e
c
ti
v
e
ly
.
T
h
e
a
u
t
h
o
r’s
m
a
jo
r
field
o
f
stu
d
y
c
o
v
e
rs
Co
m
p
u
tatio
n
a
l
Bi
o
l
o
g
y
,
M
a
c
h
i
n
e
Lea
rn
in
g
,
a
n
d
Im
a
g
e
P
ro
c
e
ss
in
g
.
He
h
a
s
b
e
e
n
wo
rk
in
g
a
s
As
sista
n
t
P
ro
fe
ss
o
r
in
t
h
e
Co
m
p
u
ter
En
g
i
n
e
e
rin
g
De
p
a
rtme
n
t
a
t
D
h
a
rm
sin
h
De
sa
i
U
n
iv
e
r
sit
y
,
Na
d
iad
,
G
u
jara
t,
In
d
ia
si
n
c
e
2
0
1
1
.
He
d
e
li
v
e
rs
e
x
p
e
rt
talk
s
a
n
d
se
rv
e
s
a
s
a
re
v
iew
e
r
fo
r
in
tern
a
ti
o
n
a
l
c
o
n
fe
re
n
c
e
s
a
n
d
jo
u
r
n
a
ls
.
He
h
a
s
a
lso
p
re
se
n
ted
p
a
p
e
rs
i
n
i
n
tern
a
ti
o
n
a
l
jo
u
rn
a
ls
a
n
d
IE
EE
c
o
n
fe
re
n
c
e
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
a
p
u
rv
a
m
e
h
ta.
c
e
@d
d
u
.
a
c
.
i
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.