I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
14
,
No
.
3
,
J
u
n
e
2
0
2
5
,
p
p
.
1960
~
1
9
6
7
I
SS
N:
2
2
5
2
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
14
.i
3
.
p
p
1
9
6
0
-
1
9
6
7
1960
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Co
nv
o
lutiona
l
neura
l
netw
o
rk
ba
s
ed
enco
der
-
deco
d
er
for
eff
icie
nt
real
-
ti
m
e
o
bject
det
ection
M
o
t
hira
m
Ra
j
a
s
e
k
a
ra
n
1
,
Ch
it
ra
Sa
ba
pa
t
hy
Ra
ng
a
na
t
ha
n
2
,
Na
g
a
ra
j
a
n
M
o
ha
nk
u
m
a
r
3
,
Ra
j
esh
k
u
m
a
r
Sa
m
pa
t
hra
j
a
n
4
,
T
ha
y
a
la
g
a
ra
n
M
er
lin
I
nb
a
m
a
la
r
5
,
Na
g
esh
v
a
ra
n
Na
nd
h
ini
6
,
Sh
a
n
m
ug
a
m
S
uja
t
ha
7
1
S
e
n
i
o
r
S
o
l
u
t
i
o
n
C
o
n
su
l
t
a
n
t
,
P
i
n
e
C
a
n
d
l
e
W
a
y
,
S
a
i
n
t
A
u
g
u
st
i
n
e
,
U
n
i
t
e
d
S
t
a
t
e
s
2
A
sso
c
i
a
t
e
V
i
c
e
P
r
e
si
d
e
n
t
,
M
p
h
a
si
s
C
o
r
p
o
r
a
t
i
o
n
,
C
h
a
n
d
l
e
r
,
U
n
i
t
e
d
S
t
a
t
e
s
3
S
y
mb
i
o
si
s
I
n
st
i
t
u
t
e
of
T
e
c
h
n
o
l
o
g
y
,
S
y
mb
i
o
si
s
I
n
t
e
r
n
a
t
i
o
n
a
l
(
D
e
e
me
d
U
n
i
v
e
r
si
t
y
)
,
N
a
g
p
u
r
C
a
mp
u
s
,
P
u
n
e
,
I
n
d
i
a
4
P
r
i
n
c
i
p
a
l
C
l
o
u
d
A
r
c
h
i
t
e
c
t
,
M
c
K
i
n
se
y
&
C
o
mp
a
n
y
,
C
h
a
n
d
l
e
r
,
U
n
i
t
e
d
S
t
a
t
e
s
5
D
e
p
a
r
t
me
n
t
of
El
e
c
t
r
o
n
i
c
s
a
n
d
I
n
st
r
u
me
n
t
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
S
a
v
e
e
t
h
a
E
n
g
i
n
e
e
r
i
n
g
C
o
l
l
e
g
e,
C
h
e
n
n
a
i
,
I
n
d
i
a
6
D
e
p
a
r
t
me
n
t
of
I
n
f
o
r
mat
i
o
n
T
e
c
h
n
o
l
o
g
y
,
P.
S.
V
C
o
l
l
e
g
e
of
En
g
i
n
e
e
r
i
n
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
K
r
i
sh
n
a
g
i
r
i
,
I
n
d
i
a
7
D
e
p
a
r
t
me
n
t
of
B
i
o
me
d
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
S
a
v
e
e
t
h
a
S
c
h
o
o
l
of
En
g
i
n
e
e
r
i
n
g
,
S
a
v
e
e
t
h
a
I
n
st
i
t
u
t
e
of
M
e
d
i
c
a
l
a
n
d
T
e
c
h
n
i
c
a
l
S
c
i
e
n
c
e
s,
S
a
v
e
e
t
h
a
U
n
i
v
e
r
si
t
y
,
C
h
e
n
n
a
i
,
I
n
d
i
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Feb
2
7
,
2
0
2
4
R
ev
i
s
ed
No
v
2
6
,
2
0
2
4
A
cc
ep
ted
J
an
2
7
,
2
0
2
5
Co
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
s
(CNN
)
a
re
a
p
p
li
e
d
to
a
v
a
riet
y
of
c
o
m
p
u
ter
v
isio
n
p
ro
b
lem
s,
su
c
h
as
o
b
jec
t
re
c
o
g
n
it
io
n
,
im
a
g
e
c
las
si
f
ica
ti
o
n
,
se
m
a
n
ti
c
se
g
m
e
n
tatio
n
,
a
n
d
m
a
n
y
o
th
e
rs.
On
e
of
th
e
m
o
st
i
m
p
o
rtan
t
a
n
d
d
if
f
icu
lt
issu
e
s
in
c
o
m
p
u
ter
v
isio
n
,
o
b
jec
t
d
e
tec
ti
o
n
,
h
a
s
a
tt
ra
c
ted
a
lo
t
of
a
tt
e
n
ti
o
n
late
l
y
.
Ob
jec
t
d
e
tec
ti
o
n
v
a
li
d
a
ti
n
g
th
e
o
c
c
u
rre
n
c
e
of
th
e
o
b
jec
t
in
th
e
p
ict
u
re
or
v
id
e
o
a
n
d
th
e
n
p
r
o
p
e
rly
lo
c
a
ti
n
g
it
f
o
r
re
c
o
g
n
it
io
n
.
Ho
w
e
v
e
r,
u
n
d
e
r
c
e
rtain
c
ircu
m
sta
n
c
e
s,
su
c
h
as
w
h
e
n
an
it
e
m
h
a
s
issu
e
s
li
k
e
o
c
c
lu
sio
n
,
d
isto
rti
o
n
,
or
sm
a
ll
siz
e
,
th
e
re
m
a
y
stil
l
be
su
b
p
a
r
d
e
tec
ti
o
n
p
e
rf
o
r
m
a
n
c
e
.
T
h
is
w
o
rk
a
i
m
s
to
p
ro
p
o
se
an
e
f
f
icie
n
t
d
e
e
p
lea
rn
in
g
m
o
d
e
l
w
it
h
CNN
a
n
d
e
n
c
o
d
e
r
d
e
c
o
d
e
r
f
o
r
e
ff
icie
n
t
o
b
jec
t
d
e
tec
ti
o
n
.
T
h
e
p
ro
p
o
se
d
m
o
d
e
l
is
e
x
p
e
ri
m
e
n
ted
on
M
icro
so
f
t
Co
m
m
o
n
Ob
jec
ts
in
Co
n
tex
t
(M
S
-
COCO
)
d
a
tas
e
t
a
n
d
a
c
h
iev
e
d
m
e
a
n
a
v
e
r
a
g
e
p
re
c
isio
n
(m
A
P
)
of
a
b
o
u
t
5
4
.
1
%
a
n
d
a
c
c
u
ra
c
y
of
9
9
%
.
T
h
e
in
v
e
stig
a
ti
o
n
a
l
o
u
tco
m
e
s
a
m
p
l
y
sh
o
w
e
d
th
a
t
th
e
su
g
g
e
ste
d
m
e
c
h
a
n
is
m
c
o
u
ld
a
c
h
i
e
v
e
a
h
ig
h
d
e
tec
ti
o
n
e
f
f
icie
n
c
y
c
o
m
p
a
re
d
w
it
h
th
e
e
x
isti
n
g
tec
h
n
iq
u
e
s
a
n
d
n
e
e
d
e
d
little
c
o
m
p
u
tati
o
n
a
l
re
so
u
rc
e
s.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
e
u
r
al
n
e
t
w
o
r
k
s
Dee
p
lear
n
in
g
E
n
co
d
er
-
d
ec
o
d
er
Me
an
av
er
a
g
e
p
r
ec
is
io
n
MS
-
C
OC
O
d
a
t
as
e
t
Ob
j
ec
t d
etec
tio
n
T
h
is
is
an
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r
th
e
CC
BY
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Nag
ar
aj
an
Mo
h
an
k
u
m
ar
S
y
m
b
io
s
is
I
n
s
ti
tu
te
of
T
ec
h
n
o
lo
g
y
,
S
y
m
b
io
s
is
I
n
ter
n
atio
n
al
(
Dee
m
ed
Un
i
v
er
s
it
y
)
,
Nag
p
u
r
C
a
m
p
u
s
P
u
n
e,
I
n
d
ia
E
m
ail:
n
m
k
p
r
o
f
es
s
o
r
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Mo
r
e
th
an
90%
of
h
u
m
a
n
u
n
d
er
s
tan
d
in
g
is
v
is
u
al,
a
n
d
v
ar
io
u
s
i
m
a
g
i
n
g
eq
u
ip
m
en
t
ar
e
f
r
eq
u
en
tl
y
u
s
ed
in
f
ield
s
th
at
ar
e
d
ir
ec
tly
r
elate
d
to
h
u
m
a
n
ac
ti
v
it
y
a
n
d
li
v
in
g
[
1
]
.
T
h
e
p
r
o
ce
s
s
in
g
of
p
h
o
to
s
a
n
d
o
th
er
in
f
o
r
m
atio
n
h
a
s
b
ee
n
s
u
cc
e
s
s
f
u
ll
y
ad
o
p
ted
in
v
ar
io
u
s
i
n
d
u
s
tr
ies
to
t
h
e
o
n
g
o
in
g
g
r
o
w
th
of
m
ac
h
i
n
e
lear
n
in
g
alg
o
r
ith
m
s
.
T
h
e
p
r
im
ar
y
r
e
s
ea
r
ch
ch
alle
n
g
e
in
co
m
p
u
ter
v
is
io
n
,
o
b
j
ec
t
d
etec
tio
n
,
h
a
s
d
r
a
w
n
i
n
cr
ea
s
i
n
g
atten
tio
n
f
r
o
m
ac
ad
e
m
ics.
T
h
e
o
b
j
ec
t
d
is
co
v
er
y
t
y
p
icall
y
co
n
tai
n
s
t
w
o
s
ta
g
es:
f
ir
s
t,
lo
o
k
i
n
g
f
o
r
th
e
ite
m
in
th
e
i
m
a
g
e;
s
ec
o
n
d
,
e
m
p
lo
y
in
g
b
o
u
n
d
i
n
g
b
o
x
e
s
to
f
i
n
d
t
h
e
o
b
j
ec
t.
C
o
n
v
o
lu
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
s
(
C
NN
)
h
as
b
ec
o
m
e
h
i
g
h
l
y
e
f
f
ec
ti
v
e
at
o
b
j
ec
t
d
etec
tio
n
in
r
ec
en
t
y
ea
r
s
[
2
]
–
[
5
]
,
r
eg
io
n
b
ased
c
o
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
(
R
-
C
NN)
[
6
]
,
YOL
O
[
7
]
,
th
e
s
p
atial
p
y
r
a
m
id
p
o
o
lin
g
n
et
w
o
r
k
(
SP
P
)
[
8
]
,
an
d
Fas
t
R
-
C
NN
[
9
]
o
b
j
ec
t
d
etec
tio
n
tech
n
iq
u
e
s
th
a
t
ar
e
u
s
ed
in
t
h
is
f
ield
of
s
tu
d
y
.
Du
e
to
co
m
p
u
tatio
n
al
h
ar
d
w
ar
e
an
d
d
ata
av
ailab
ilit
y
,
tr
ad
itio
n
al
o
b
j
ec
t
d
etec
tio
n
alg
o
r
ith
m
s
h
av
e
s
i
g
n
i
f
ica
n
t
d
r
a
wb
ac
k
s
[
1
0
]
.
C
o
n
v
er
s
el
y
,
w
it
h
t
h
e
d
ev
elo
p
m
e
n
t
of
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
b
a
s
ed
en
c
o
d
er
-
d
ec
o
d
er fo
r
efficien
t rea
l
-
time
…
(
Mo
th
ir
a
m
R
a
ja
s
ek
a
r
a
n
)
1961
ar
tif
icial
in
telli
g
en
ce
(
A
I
)
an
d
p
r
o
ce
s
s
in
g
p
o
w
er
in
r
ec
en
t
y
ea
r
s
,
th
e
en
tire
p
r
o
ce
s
s
ca
n
n
o
w
be
a
u
to
m
ated
w
it
h
litt
le
to
no
h
u
m
an
i
n
v
o
lv
e
m
e
n
t.
T
h
e
p
r
im
ar
y
d
i
s
ti
n
cti
o
n
is
th
at
tr
ad
itio
n
al
o
b
j
ec
t
d
etec
tio
n
tech
n
iq
u
e
s
r
el
y
on
h
u
m
a
n
ex
p
er
ien
ce
s
tan
d
ar
d
s
an
d
ex
p
er
t
j
u
d
g
e
m
en
t
to
ex
tr
ac
t
f
ea
t
u
r
es,
w
h
er
ea
s
AI
u
s
es
a
s
o
p
h
i
s
ticate
d
n
eu
r
al
n
et
w
o
r
k
t
h
at
can
be
tr
ai
n
ed
to
r
o
u
tin
el
y
id
en
ti
f
y
p
o
w
e
r
f
u
l
a
n
d
j
u
d
icial
f
ea
t
u
r
es
.
In
p
ar
ticu
lar
,
e
n
co
d
er
-
d
ec
o
d
er
m
o
d
els
b
a
s
ed
on
f
u
ll
y
co
n
v
o
lu
tio
n
al
n
e
t
w
o
r
k
s
(
FC
Ns)
h
av
e
s
ig
n
i
f
ica
n
tl
y
e
n
h
a
n
ce
d
p
er
f
o
r
m
an
ce
,
s
u
ch
as
s
e
m
an
t
ic
s
e
g
m
en
tatio
n
[
1
1
]
,
[
1
2
]
,
ed
g
e
r
ec
o
g
n
itio
n
[
1
3
]
o
b
j
ec
t
ex
p
o
s
u
r
e
[
1
4
]
,
an
d
cr
o
w
d
co
u
n
ti
n
g
[
1
5
]
.
E
s
s
en
tiall
y
,
th
e
tr
e
n
d
of
p
o
p
u
lar
o
b
j
ec
t
id
en
tif
icatio
n
tech
n
iq
u
e
s
ar
e
o
p
er
ate
w
it
h
i
n
th
e
en
co
d
er
-
d
e
c
o
d
er
f
r
am
e
w
o
r
k
.
Fo
r
th
e
d
etec
tio
n
task
,
s
o
m
e
r
esear
ch
er
s
cr
ea
ted
s
tr
u
ctu
r
es
b
ased
on
th
e
en
co
d
er
-
d
ec
o
d
er
p
ar
ad
ig
m
a
n
d
attai
n
ed
c
u
tti
n
g
-
ed
g
e
p
er
f
o
r
m
a
n
ce
[
1
6
]
.
W
ith
r
eg
ar
d
to
b
en
ch
m
ar
k
d
atase
ts
,
C
NN
-
b
as
ed
en
co
d
er
-
d
ec
o
d
er
m
o
d
els
ar
e
p
ar
ticu
lar
l
y
cr
u
cial
f
o
r
co
n
ti
n
u
o
u
s
l
y
i
m
p
r
o
v
i
n
g
d
etec
tio
n
p
er
f
o
r
m
an
ce
[
1
7
]
.
C
o
n
v
o
lu
tio
n
is
d
o
n
e
by
t
h
e
e
n
co
d
er
,
w
h
er
ea
s
d
ec
o
n
v
o
lu
tio
n
,
un
-
p
o
o
lin
g
,
an
d
up
-
s
a
m
p
li
n
g
ar
e
d
o
n
e
by
t
h
e
d
ec
o
d
er
to
f
o
r
ec
ast
p
ix
el
-
w
i
s
e
clas
s
lab
els.
T
h
e
up
-
s
a
m
p
lin
g
d
ec
o
d
e
th
at
co
r
r
esp
o
n
d
s
th
e
lo
w
-
r
eso
lu
t
io
n
en
co
d
er
f
ea
tu
r
e
m
ap
s
,
is
th
e
i
m
p
o
r
tan
t
f
ea
t
u
r
e.
T
h
is
ar
ch
it
ec
tu
r
e
e
m
p
lo
y
s
th
e
en
co
d
er
'
s
p
o
o
lin
g
i
n
d
icato
r
s
to
up
-
s
a
m
p
le
to
m
ap
p
ix
el
-
w
is
e
ca
teg
o
r
izatio
n
w
h
ile
al
s
o
s
i
g
n
i
f
ica
n
tl
y
r
ed
u
ci
n
g
th
e
n
u
m
b
er
of
tr
ai
n
ab
le
p
ar
a
m
eter
s
.
T
h
is
p
ap
er
is
s
tr
u
ctu
r
ed
as
f
o
llo
w
s
.
T
h
e
g
o
al
of
o
b
j
ec
t
d
etec
tio
n
,
w
h
ic
h
is
t
y
p
ical
l
y
d
o
n
e
w
it
h
p
h
o
to
s
or
v
id
eo
s
,
is
to
f
in
d
b
o
r
d
er
s
as
w
ell
as
to
s
h
o
w
t
h
e
o
b
j
ec
t'
s
r
a
n
g
e
a
n
d
lo
ca
tio
n
.
T
h
e
n
e
x
t
s
tep
is
to
class
i
f
y
t
h
e
o
b
j
ec
t'
s
ca
te
g
o
r
y
an
d
to
p
r
o
v
id
e
th
e
ca
teg
o
r
izatio
n
li
k
eli
h
o
o
d
.
T
h
i
s
tas
k
is
m
o
r
e
d
if
f
icu
lt
th
a
n
s
i
m
p
le
p
ictu
r
e
clas
s
i
f
icatio
n
b
ec
au
s
e
t
h
e
p
o
s
itio
n
s
of
m
an
y
ite
m
s
m
u
s
t
be
d
eter
m
i
n
ed
f
r
o
m
th
e
i
m
ag
e
or
v
id
eo
.
C
NNs
h
av
e
b
ee
n
u
s
ed
f
o
r
th
e
d
etec
tio
n
an
d
class
i
f
icatio
n
of
o
b
j
ec
ts
w
ith
s
u
cc
e
s
s
[
1
8
]
.
C
u
r
r
en
t
m
o
d
el
s
i
n
clu
d
e
w
a
y
s
to
ca
te
g
o
r
is
e
eith
er
a
f
u
l
l
i
n
p
u
t
w
i
n
d
o
w
f
o
r
ea
c
h
s
ce
n
e
f
o
r
a
b
o
u
n
d
i
n
g
box
of
s
ev
er
al
o
b
j
ec
ts
.
Se
m
an
t
ic
s
e
g
m
en
ta
tio
n
h
as
h
ad
a
b
r
ea
k
t
h
r
o
u
g
h
th
an
k
s
to
FC
N.
It
h
as
p
r
o
v
id
ed
a
p
o
ten
t
m
et
h
o
d
f
o
r
b
o
o
s
tin
g
t
h
e
e
f
f
ec
ti
v
e
n
ess
of
C
NNs
by
p
r
o
v
id
in
g
in
p
u
ts
of
an
y
s
ize
[
1
9
]
.
T
h
e
en
co
d
er
-
d
ec
o
d
er
-
b
ased
c
o
n
ce
p
t
th
at
p
r
esen
ted
by
[
2
0
]
.
It
s
u
g
g
e
s
ted
f
o
r
f
ea
tu
r
e
lear
n
in
g
th
at
is
u
n
s
u
p
er
v
i
s
ed
;
t
h
en
,
n
eu
r
al
n
e
t
w
o
r
k
s
b
ac
k
ed
by
en
co
d
er
-
d
ec
o
d
er
s
h
av
e
e
m
er
g
ed
as
a
p
o
ten
t
ial
r
ep
lace
m
e
n
t
f
o
r
f
u
r
t
h
er
aid
s
.
An
i
n
tr
i
g
u
i
n
g
p
ed
estrian
co
lli
s
io
n
aler
t
s
y
s
te
m
f
o
r
ad
v
an
ce
d
d
r
iv
er
ass
is
tan
ce
s
y
s
te
m
s
w
as
s
u
g
g
ested
in
[
2
1
]
.
Ho
w
ev
er
,
it
is
o
n
l
y
ca
p
ab
le
of
d
etec
ti
n
g
a
n
d
w
ar
n
in
g
p
ed
estrian
s
.
F
ac
ia
l
f
ea
t
u
r
e
lo
ca
lizatio
n
[
2
2
]
ex
tr
a
cted
in
f
o
r
m
atio
n
f
r
o
m
in
p
u
t
s
t
r
in
g
s
t
h
at
co
u
ld
o
n
l
y
b
e
o
n
e
d
i
m
e
n
s
io
n
u
s
i
n
g
th
e
Viter
b
i
d
ec
o
d
in
g
tec
h
n
iq
u
e
.
Su
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SV
M
)
-
b
ased
p
r
ed
ictiv
e
m
o
d
eli
n
g
[
2
3
]
u
til
is
ed
t
h
e
s
i
m
ilar
co
n
ce
p
t to
ex
p
an
d
SV
M
o
u
tco
m
es
u
s
i
n
g
t
w
o
-
d
i
m
e
n
s
io
n
al
m
ap
s
.
As
an
atte
n
tio
n
g
e
n
er
ati
n
g
m
o
d
u
le
th
at
lear
n
s
to
s
p
ec
i
f
icall
y
atten
d
to
s
i
g
n
i
f
ica
n
t
lo
ca
tio
n
s
f
o
r
ev
er
y
p
ix
el
by
e
m
p
lo
y
in
g
b
id
ir
ec
ti
o
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
(
Bi
-
L
ST
M
)
m
o
d
u
le
w
it
h
i
n
th
e
f
ea
t
u
r
e
m
ap
s
,
p
ae
d
iatr
ic
in
ten
s
i
v
e
ca
r
e
a
u
d
it
n
et
w
o
r
k
(
P
iC
A
Net
)
w
a
s
p
r
o
p
o
s
ed
in
[
2
4
]
.
Fo
r
C
-
ele
g
an
s
tis
s
u
es
w
i
th
F
C
N
in
f
er
en
ce
,
co
ar
s
e
m
u
lti
-
clas
s
s
eg
m
e
n
tat
io
n
C
NN
w
it
h
FC
N
ar
ch
itect
u
r
e.
In
o
r
d
er
to
f
o
r
ec
ast
p
ix
el
-
le
v
el
lab
el
s
an
d
to
i
m
p
r
o
v
e
t
h
e
lab
el
m
a
p
u
s
i
n
g
co
n
d
itio
n
al
r
a
n
d
o
m
f
ield
(
C
R
F),
n
et
w
o
r
k
ac
h
ie
v
es
d
en
s
er
s
co
r
e
m
ap
s
u
s
i
n
g
F
C
N
ar
ch
i
tectu
r
e.
O
n
e
of
th
e
c
u
r
r
en
t
m
aj
o
r
tr
en
d
s
in
C
NN
ar
c
h
itect
u
r
e
d
esig
n
is
t
h
e
i
n
co
r
p
o
r
atio
n
of
en
co
d
er
an
d
d
ec
o
d
er
to
im
p
r
o
v
e
p
er
f
o
r
m
a
n
ce
.
A
p
ar
t
f
r
o
m
th
ese
o
b
j
ec
t
d
etec
tio
n
m
o
d
els;
s
e
v
er
al
d
etec
tio
n
alg
o
r
ith
m
s
ar
e
i
m
p
le
m
e
n
ted
on
h
ar
d
w
ar
e
p
latf
o
r
m
s
to
i
m
p
r
o
v
e
th
e
d
etec
tio
n
p
er
f
o
r
m
an
ce
.
P
y
r
a
m
id
s
ce
n
e
a
n
al
y
s
i
s
n
et
wo
r
k
(
P
SP
Net
)
is
y
e
t
a
n
o
th
er
e
f
f
ec
tiv
e
C
NN
ar
c
h
itect
u
r
e
t
h
a
t
w
as
j
u
s
t
r
elea
s
ed
.
It
is
in
ten
d
ed
f
o
r
p
r
ed
ictio
n
j
o
b
s
at
th
e
p
ix
el
lev
el.
T
h
e
g
lo
b
al
p
y
r
a
m
id
p
o
o
lin
g
s
tr
u
ct
u
r
e
th
a
t
co
m
b
i
n
ed
g
lo
b
al
an
d
lo
ca
l
h
in
ts
t
h
at
p
r
o
d
u
ce
th
e
r
esu
l
ts
b
u
ild
s
t
h
e
p
ix
el
-
lev
e
l
f
ea
tu
r
es
f
o
r
ef
f
ec
ti
v
e
s
eg
m
e
n
tatio
n
.
D
u
e
to
t
h
e
P
SP
Net
ar
ch
itec
tu
r
e
'
s
e
x
tr
e
m
e
co
m
p
lex
i
t
y
,
tr
ain
i
n
g
a
n
d
te
s
ti
n
g
p
r
o
ce
s
s
es
n
ee
d
f
o
r
a
s
izab
le
a
m
o
u
n
t
of
p
r
o
ce
s
s
in
g
p
o
w
er
an
d
g
r
ap
h
ics
p
r
o
ce
s
s
i
n
g
u
n
its
(
GP
U
)
ca
p
ab
ilit
ies.
T
h
e
co
n
ce
p
t
of
p
an
o
p
tic
s
eg
m
e
n
tatio
n
(
P
S)
was
r
ec
en
tl
y
in
tr
o
d
u
ce
d
in
a
s
t
u
d
y
ab
o
u
t
p
i
x
el
-
w
is
e
s
e
g
m
en
ta
t
io
n
.
To
co
m
p
lete
a
b
r
o
ad
s
eg
m
en
tatio
n
ta
s
k
,
PS
co
m
b
in
e
s
s
e
g
m
en
tin
g
i
n
s
tan
ce
s
a
n
d
s
e
g
m
en
ta
tio
n
b
a
s
ed
on
s
e
m
an
tics
.
C
o
m
p
ar
ativ
e
l
y
s
p
ea
k
i
n
g
,
it
p
er
f
o
r
m
s
w
e
ll
w
h
en
co
m
p
ar
ed
to
p
r
ev
io
u
s
v
is
u
al
g
eo
m
etr
y
g
r
o
u
p
(
VGG
)
b
ased
n
et
w
o
r
k
s
,
alth
o
u
g
h
s
ize
is
th
e
d
esig
n
'
s
m
ai
n
f
la
w
.
T
h
e
p
r
o
p
h
et
alg
o
r
ith
m
,
K
-
m
e
an
s
cl
u
s
ter
i
n
g
,
an
d
s
ea
s
o
n
al
a
u
to
r
eg
r
es
s
iv
e
i
n
te
g
r
ated
m
o
v
i
n
g
-
a
v
er
ag
e
m
et
h
o
d
s
act
a
tas
k
in
e
n
h
an
ci
n
g
th
e
clo
u
d
i
n
f
r
astr
u
ctu
r
e
s
.
A
l
s
o
,
it
g
r
o
u
p
i
n
g
s
er
v
er
s
i
n
to
clu
s
ter
s
w
it
h
s
i
m
ilar
u
tili
za
t
io
n
p
atter
n
s
.
K
-
m
ea
n
s
clu
s
ter
i
n
g
en
h
an
ce
s
th
e
r
eso
u
r
ce
allo
ca
tio
n
ef
f
icie
n
c
y
[
2
5
]
.
I
n
ter
n
et
of
th
i
n
g
s
(
I
o
T
)
-
d
r
iv
en
i
m
ag
e
r
ec
o
g
n
i
tio
n
s
y
s
te
m
u
t
ilizi
n
g
C
N
Ns
to
n
o
tice
an
d
q
u
a
n
ti
f
y
m
icr
o
p
last
ics
[
2
6
]
.
T
h
e
d
ata
co
llected
by
s
e
n
s
o
r
s
is
f
o
r
w
a
r
d
to
a
ce
n
tr
alize
d
m
o
n
ito
r
i
n
g
s
y
s
te
m
t
h
a
t
d
ec
id
es
w
h
et
h
er
or
n
o
t
an
alar
m
ac
tiv
ated
in
t
h
e
e
v
en
t
if
t
h
e
s
it
u
atio
n
d
iv
er
g
e
f
r
o
m
t
h
eir
id
ea
l
s
tate
[
2
7
]
.
K
-
n
ea
r
est
n
ei
g
h
b
o
r
(
KNN)
an
d
SV
M
alg
o
r
ith
m
f
o
r
m
s
a
p
r
ec
is
e
a
r
r
an
g
e
m
en
t
m
o
d
el
t
o
u
til
ize
th
e
i
m
p
o
r
ta
n
t
d
ata
ex
p
ec
tat
io
n
ex
ac
t
n
es
s
[
2
8
]
.
SVM
w
i
th
r
ec
u
r
r
en
t
n
eu
r
al
n
et
w
o
r
k
s
ar
e
p
o
w
er
f
u
l
clas
s
i
f
i
ca
tio
n
t
h
at
m
a
k
es
it
f
ea
s
ib
le
t
o
class
i
f
y
p
atie
n
ts
’
r
is
k
s
a
n
d
p
r
ed
ict
h
o
w
t
h
e
y
w
i
l
l r
ea
ct
to
t
h
er
ap
y
[
2
9
]
.
C
lo
u
d
co
m
p
u
ti
n
g
g
r
an
t
s
t
h
e
s
eizu
r
e
p
r
ed
ictio
n
s
y
s
te
m
to
i
m
p
r
o
v
e
ac
ce
s
s
ib
le
a
n
d
s
ca
la
b
le
[
3
0
]
an
d
it
ex
a
m
i
n
es
t
h
e
f
ea
tu
r
e
s
elec
tio
n
d
ev
elo
p
ed
in
f
o
r
i
m
p
r
o
v
i
n
g
ac
cu
r
ac
y
[
3
1
]
.
Hy
b
r
id
m
ac
h
i
n
e
lear
n
in
g
tec
h
n
iq
u
es li
k
e
SV
M
w
it
h
C
NN
al
g
o
r
ith
m
to
an
ti
cip
ates A
lzh
ei
m
er
’
s
s
ic
k
n
es
s
[
3
2
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
14
,
No
.
3
,
J
u
n
e
2
0
2
5
:
1
9
6
0
-
1967
1962
2.
P
RO
P
O
SE
D
DE
E
P
L
E
ARN
I
NG
M
O
DE
L
T
h
e
p
r
o
p
o
s
ed
ar
ch
itectu
r
e
is
a
p
ix
el
-
w
is
e
m
o
d
el
th
at
b
u
ilt
on
t
w
o
d
ec
o
u
p
led
F
C
Ns
f
o
r
e
n
co
d
in
g
as
w
ell
as
d
ec
o
d
in
g
as
ex
p
lai
n
e
d
in
Fi
g
u
r
e
1.
T
h
e
p
r
ev
io
u
s
l
y
d
escr
ib
ed
en
co
d
er
is
b
u
ilt
u
s
i
n
g
t
h
e
f
ir
s
t
16
co
n
v
o
lu
tio
n
al
la
y
er
s
of
VGG
-
19
n
et
w
o
r
k
,
t
h
e
n
a
b
atch
n
o
r
m
ali
s
ed
(
B
N)
lay
er
,
f
u
n
ctio
n
of
ac
tiv
a
tio
n
,
p
o
o
lin
g
la
y
er
,
as
w
ell
as
d
r
o
p
o
u
t
u
n
it
s
.
T
h
e
d
ec
o
d
er
n
et
w
o
r
k
is
co
m
p
o
s
ed
of
la
y
er
s
f
o
r
u
p
s
a
m
p
l
in
g
,
d
ec
o
n
v
o
lu
t
io
n
,
ac
tiv
atio
n
,
b
atch
n
o
r
m
alis
a
tio
n
,
d
r
o
p
o
u
t,
an
d
a
m
u
l
ti
-
c
lass
class
if
icatio
n
.
E
v
er
y
d
ec
o
d
er
is
m
atc
h
ed
to
a
p
o
o
lin
g
u
n
it
of
an
e
n
co
d
er
in
t
h
e
s
y
s
te
m
's
o
v
er
all
e
n
co
d
er
-
d
ec
o
d
er
in
ter
f
ac
e.
C
o
n
s
eq
u
e
n
tl
y
,
t
h
e
d
ec
o
d
er
C
NN
h
as
16
de
-
co
n
v
o
lu
tio
n
la
y
er
s
.
To
p
r
o
b
ab
ilit
ies
of
o
u
tp
u
t
class
m
ea
n
t
to
each
in
d
iv
id
u
al
p
ix
el
in
d
iv
id
u
all
y
,
t
h
e
d
ec
o
d
er
s
en
d
s
its
co
m
p
u
tatio
n
s
to
s
o
f
t
m
ax
cla
s
s
i
f
ier
.
Fig
u
r
e
1
.
P
r
o
p
o
s
ed
ar
ch
itectu
r
e
f
o
r
r
ea
l
-
ti
m
e
o
b
j
ec
t d
etec
tio
n
T
h
e
k
e
y
b
e
n
ef
i
ts
of
o
u
r
s
u
g
g
ested
d
ec
o
u
p
led
ar
ch
itec
tu
r
e
ar
e
its
s
i
m
p
le
tr
ai
n
i
n
g
w
it
h
v
ar
io
u
s
en
v
ir
o
n
m
e
n
tal
f
ac
to
r
s
an
d
ea
s
e
of
cu
s
to
m
izatio
n
.
Fo
r
p
ix
el
-
w
i
s
e
class
if
ica
tio
n
,
t
h
e
en
co
d
er
cr
ea
tes
lo
w
-
r
eso
lu
tio
n
f
ea
t
u
r
e
m
ap
s
,
w
h
ic
h
th
e
d
ec
o
d
er
up
-
s
a
m
p
le
s
th
r
o
u
g
h
co
n
v
o
l
u
tio
n
in
g
t
h
e
tr
ain
ab
le
f
il
ter
s
to
y
ield
i
n
te
n
s
e
f
ea
t
u
r
e
m
ap
s
[
3
3
]
.
T
h
e
f
u
n
d
a
m
en
tal
co
m
p
o
n
en
t
of
t
h
e
s
u
g
g
ested
m
et
h
od
is
th
e
d
ec
o
d
in
g
p
r
o
ce
d
u
r
e,
w
h
ich
p
r
o
v
id
es
s
ev
er
al
u
s
e
f
u
l
ad
v
an
tag
e
s
in
ter
m
s
of
i
m
p
r
o
v
in
g
b
o
u
n
d
ar
y
d
eli
n
ea
tio
n
an
d
r
ed
u
ctio
n
.
A
l
s
o
m
u
ch
i
m
p
r
o
v
ed
is
th
e
ab
ilit
y
to
p
r
o
v
id
e
tr
ain
in
g
by
lo
w
er
i
n
g
t
h
e
am
o
u
n
t
of
tr
ain
ab
le
attr
ib
u
tes.
It
o
f
f
er
s
a
s
i
m
p
le
tr
ain
i
n
g
,
w
h
ic
h
tr
ain
s
b
o
th
th
e
e
n
co
d
er
as
w
ell
as
d
ec
o
d
er
at
th
e
s
a
m
e
ti
m
e
.
W
ith
an
i
n
p
u
t
i
m
ag
e,
th
e
n
e
t
w
o
r
k
b
eg
i
n
s
tr
ain
i
n
g
a
n
d
ac
t
s
d
u
r
i
n
g
t
h
e
n
et
w
o
r
k
to
th
e
t
o
p
la
y
er
s
.
A
d
o
p
tin
g
co
n
v
o
l
u
tio
n
w
it
h
a
p
r
ea
r
r
an
g
ed
s
et
of
f
ilter
b
an
k
s
to
f
ab
r
icate
f
ea
tu
r
e
m
ap
s
,
th
e
b
atch
n
o
r
m
ali
s
atio
n
p
r
o
ce
s
s
is
f
u
lf
illed
by
th
e
e
n
co
d
er
.
Af
ter
w
ar
d
s
,
ac
ti
v
atio
n
s
ar
e
ac
co
m
p
lis
h
ed
by
r
ec
tif
ied
li
n
ea
r
u
n
its
(
R
eL
U
s
)
.
T
h
e
m
a
x
-
p
o
o
lin
g
f
u
n
ct
io
n
is
t
h
en
f
u
l
f
illed
w
it
h
a
w
i
n
d
o
w
s
ize
of
2
x
2
a
n
d
a
tr
ea
d
of
1.
T
h
is
o
u
tco
m
es
in
a
t
w
o
-
f
o
ld
s
u
b
s
a
m
p
lin
g
of
t
h
e
last
i
m
ag
e.
M
u
ltip
le
p
o
o
lin
g
la
y
er
s
ab
le
to
i
n
cr
ea
s
e
tr
a
n
s
latio
n
in
v
ar
ia
n
ce
f
o
r
ef
f
ec
tiv
e
ca
te
g
o
r
izatio
n
j
o
b
s
,
b
u
t
th
e
f
ea
t
u
r
e
m
ap
s
'
s
p
atial
r
e
s
o
lu
tio
n
is
u
n
n
ec
ess
ar
il
y
r
ed
u
ce
d
.
T
h
er
ef
o
r
e,
p
r
io
r
to
th
e
s
u
b
-
s
a
m
p
lin
g
f
u
n
ctio
n
,
th
e
b
o
u
n
d
ar
y
i
n
f
o
r
m
atio
n
n
ee
d
s
to
be
r
ec
o
r
d
ed
an
d
s
to
r
ed
in
th
e
en
co
d
er
f
ea
tu
r
e
m
ap
s
.
Ho
w
ev
er
,
it
is
n
o
t
p
r
ac
tical
to
s
av
e
t
h
e
e
n
tire
e
n
c
o
d
er
f
ea
tu
r
e
m
ap
s
b
ec
au
s
e
to
m
e
m
o
r
y
li
m
itatio
n
s
.
T
h
e
b
est
o
p
tio
n
is
k
ee
p
th
e
m
ax
-
p
o
o
lin
g
in
d
icato
r
s
in
s
to
r
ag
e.
Fo
r
each
2
x
2
p
o
o
lin
g
w
in
d
o
w
,
t
w
o
b
its
ar
e
u
s
ed
to
m
e
m
o
r
i
s
e
th
e
p
o
s
itio
n
s
of
each
m
ax
-
p
o
o
lin
g
f
ea
t
u
r
e
-
m
ap
.
Hav
in
g
a
lo
t
of
f
ea
t
u
r
e
m
ap
s
on
h
a
n
d
is
a
r
ea
ll
y
ef
f
ec
ti
v
e
s
o
lu
t
io
n
.
W
ith
th
is
ap
p
r
o
ac
h
,
th
e
en
co
d
er
c
an
s
to
r
e
d
ata
m
u
ch
m
o
r
e
e
f
f
icie
n
tl
y
a
n
d
f
u
ll
y
co
n
n
ec
ted
la
y
er
s
ca
n
be
d
r
o
p
p
e
d
.
3.
RE
SU
L
T
S
AND
D
I
SCU
SS
I
O
N
T
h
e
MS
-
C
O
C
O
d
ataset
t
h
at
c
o
n
s
is
t
of
91
ite
m
w
it
h
2
.
5
m
il
l
io
n
lab
elled
ex
a
m
p
le
s
in
328k
i
m
ag
e
s
,
is
u
s
ed
to
tr
ai
n
t
h
e
p
r
o
p
o
s
ed
o
b
j
ec
t
d
etec
tio
n
alg
o
r
ith
m
.
On
a
s
in
g
le
12
GB
NVI
DI
A
T
es
la
K4
0
c
GP
U,
th
e
s
u
g
g
e
s
ted
n
et
w
o
r
k
w
a
s
tr
ain
e
d
.
T
h
e
n
et
w
o
r
k
is
tr
ain
ed
u
n
ti
l
th
e
ac
cu
r
ac
y
as
w
el
l
as
lo
s
s
do
n
o
t
s
ig
n
i
f
ica
n
tl
y
g
r
o
w
or
d
ec
r
ea
s
e
an
d
th
e
lo
s
s
h
as
co
n
v
er
g
ed
.
T
h
e
w
h
o
le
n
e
t
w
o
r
k
is
estab
li
s
h
ed
an
d
tr
ai
n
e
d
u
tili
zi
n
g
t
h
e
C
a
f
f
e
B
er
k
ele
y
Vi
s
io
n
L
ib
r
ar
y
.
C
a
f
f
e
p
r
o
v
id
es
a
f
le
x
ib
ilit
y
w
h
i
le
it
r
elate
s
cr
ea
tin
g
n
et
w
o
r
k
la
y
er
s
as
w
el
l
as
tr
ain
i
n
g
t
h
e
n
et
w
o
r
k
to
m
ee
t
th
e
s
u
g
g
e
s
ted
s
p
ec
i
f
icatio
n
s
.
T
h
u
s
,
o
n
ce
co
n
v
er
g
es,
it
is
tr
ain
ed
,
an
d
no
co
n
s
id
er
ab
le
r
ed
u
ce
in
tr
ain
in
g
lo
s
s
is
s
ee
n
.
T
h
e
en
tire
r
e
s
u
l
ts
ar
e
t
h
en
e
v
al
u
ated
,
ex
a
m
i
n
e
d
,
an
d
s
u
b
s
eq
u
en
tl
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
b
a
s
ed
en
c
o
d
er
-
d
ec
o
d
er fo
r
efficien
t rea
l
-
time
…
(
Mo
th
ir
a
m
R
a
ja
s
ek
a
r
a
n
)
1963
co
n
tr
asted
w
it
h
t
h
e
s
p
ec
if
ied
b
en
ch
m
ar
k
r
e
s
u
l
ts
.
T
h
e
d
atase
t
is
d
i
v
id
e
by
tr
ai
n
i
n
g
an
d
te
s
tin
g
.
Her
e,
90%
is
allo
tted
f
o
r
tr
ain
in
g
an
d
10%
is
allo
tted
f
o
r
test
i
n
g
.
Ma
n
y
w
eig
h
ts
ar
e
0
b
e
ca
u
s
e
tr
a
in
in
g
m
o
d
els
f
r
e
q
u
en
tly
u
s
e
th
e
R
eL
U
ac
tiv
at
io
n
f
u
n
cti
o
n
.
In
th
i
s
w
o
r
k
,
it
w
as
f
o
u
n
d
th
at
af
te
r
cr
ea
tin
g
th
e
s
p
ar
s
ity
m
o
d
el,
th
e
g
r
a
d
ien
t
v
an
is
h
e
d
d
u
r
in
g
tr
a
in
in
g
w
ith
R
eL
U6
.
T
h
is
is
as
a
r
esu
lt
of
th
e
m
ask
ex
clu
d
in
g
50%
of
th
e
w
eig
h
ts
f
r
o
m
th
e
g
r
ad
i
en
t
u
p
d
ate
.
As
in
d
ic
ate
d
in
T
a
b
l
e
1
,
th
e
p
u
b
lic
d
ataset
MS
-
C
O
C
O
ev
alu
a
ted
an
d
co
n
tr
asted
w
i
th
th
e
ea
r
lier
tech
n
iq
u
es.
In
t
h
is
w
o
r
k
,
t
h
er
e
ar
e
5
k
an
d
118
k
p
h
o
to
s
ar
e
u
tili
s
ed
f
o
r
test
in
g
an
d
tr
ain
i
n
g
t
h
e
m
o
d
el,
r
esp
ec
ti
v
el
y
.
To
en
s
u
r
e
th
at
th
e
s
u
g
g
e
s
ted
m
et
h
o
d
w
o
r
k
s
,
t
h
e
o
u
tco
m
e
s
of
each
tr
ial
w
er
e
ex
a
m
i
n
ed
.
Fo
r
all
class
es,
av
er
ag
e
p
r
ec
is
io
n
(
AP
)
is
t
y
p
icall
y
d
eter
m
in
ed
,
a
n
d
its
m
id
d
li
n
g
is
k
n
o
w
n
as
th
e
m
ea
n
a
v
e
r
ag
e
p
r
ec
is
io
n
(
m
A
P
)
.
A
d
d
it
io
n
all
y
,
f
o
r
A
P
7
5
ca
n
d
id
ate
i
m
a
g
es,
r
eg
io
n
s
w
i
th
ab
o
v
e
75%
ac
cu
r
ac
y
ar
e
co
u
n
ted
,
an
d
th
e
A
P
5
0
d
esig
n
ates
th
e
50%
ar
ea
p
r
o
p
er
ly
.
Fig
u
r
e
2
s
h
o
w
s
t
h
e
m
u
lti
-
o
b
j
ec
t
d
etec
tio
n
r
esu
lts
r
ec
eiv
ed
v
ia
tr
ai
n
in
g
m
o
d
el
on
MS
-
C
O
C
O
d
ataset
.
Fig
u
r
e
2.
Scr
ee
n
s
h
o
t
f
o
r
m
u
lti
-
o
b
j
ec
t
d
etec
tio
n
of
co
m
p
le
x
s
c
en
es
u
s
i
n
g
p
r
o
p
o
s
ed
m
o
d
el
tr
ain
ed
on
MS
-
C
O
C
O
d
ataset
Fo
r
co
m
p
le
x
s
ce
n
es,
t
h
e
p
r
o
p
o
s
ed
C
NN
b
ased
en
co
d
er
d
ec
o
d
er
m
o
d
el
ac
h
ie
v
ed
b
etter
d
etec
tio
n
p
er
f
o
r
m
a
n
ce
.
T
h
e
d
etec
tio
n
r
esu
lt
s
in
c
lu
d
e
v
ar
io
u
s
o
b
j
ec
ts
s
u
ch
as
h
o
r
s
e,
p
o
tted
p
lan
t
an
d
p
er
s
o
n
as
s
h
o
w
n
in
Fig
u
r
e
3
.
Fo
r
th
is
d
etec
tio
n
,
f
lo
atin
g
p
o
in
t
o
p
er
atio
n
s
p
er
s
ec
o
n
d
(
FL
OP
s
)
is
ab
o
u
t
1
2
8
.
4
6
w
it
h
m
o
d
el
s
ize
is
1
3
4
.
2
2
MB.
Fig
u
r
e
3
illu
s
tr
a
tes
th
e
d
etec
tio
n
o
f
m
u
ltip
le
o
b
j
ec
ts
o
n
MS
-
C
O
C
O
d
atas
et
u
s
i
n
g
p
r
o
p
o
s
ed
m
o
d
el.
T
h
er
e
ar
e
v
ar
io
u
s
o
b
j
e
cts ar
e
d
etec
ted
f
r
o
m
s
a
m
p
le
c
o
m
p
le
x
i
m
ag
e
s
i
n
MS
-
C
O
C
O
d
ataset
.
Fig
u
r
e
3.
R
esu
lts
f
o
r
o
b
j
ec
t
d
e
tectio
n
of
co
m
p
le
x
s
ce
n
es
u
s
i
n
g
p
r
o
p
o
s
ed
m
o
d
el
tr
ain
ed
on
MS
-
C
O
C
O
d
ataset
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
ac
h
ie
v
ed
m
A
P
of
5
4
.
1
%
at
327
FP
S
as
s
h
o
w
n
in
T
ab
le
1.
W
ith
th
e
h
elp
of
th
i
s
in
v
e
s
ti
g
atio
n
,
th
e
m
o
d
el
'
s
p
er
f
o
r
m
a
n
ce
in
r
ea
l
-
t
i
m
e
w
as
g
u
ar
an
teed
.
MS
-
C
OC
O
d
atase
t
co
n
tai
n
s
t
h
e
FP
S
v
alu
e
is
327,
th
e
p
er
ce
n
tag
e
of
m
AP
v
al
u
e
is
5
4
.
1
%,
A
P
5
0
v
alu
e
is
7
7
.
2
%
an
d
AP
75
v
alu
e
is
6
9
.
3
%.
T
ab
le
2
d
em
o
n
s
tr
ate
s
t
h
e
co
m
p
ar
ativ
e
r
esu
l
ts
of
p
r
o
p
o
s
ed
m
o
d
el
w
it
h
ex
i
s
ti
n
g
ap
p
r
o
ac
h
es.
F
i
g
u
r
e
4
ex
p
lai
n
s
t
h
e
ex
ec
u
t
io
n
an
al
y
s
is
of
s
i
n
g
le
-
s
h
o
t
d
etec
to
r
(
SSD
)
,
YOL
O
v
3
,
E
f
f
icie
n
tDe
t,
YOL
O
v
4
ti
n
y
,
R
et
in
aNe
t
,
an
d
p
r
o
p
o
s
ed
C
NN
-
b
ased
en
co
d
er
d
ec
o
d
er
m
o
d
el
f
o
r
o
b
j
ec
t
d
ete
ctio
n
.
C
o
m
p
ar
e
to
all
o
th
er
m
o
d
els,
th
e
p
r
o
p
o
s
ed
m
o
d
el
h
as
p
r
o
v
id
ed
b
etter
r
esu
lts
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
14
,
No
.
3
,
J
u
n
e
2
0
2
5
:
1
9
6
0
-
1967
1964
T
ab
le
1.
R
esu
lts
f
o
r
p
r
o
p
o
s
ed
o
b
j
ec
t
d
etec
tio
n
m
o
d
el
D
a
t
a
se
t
F
P
S
mA
P
(
%)
A
P
5
0
(
%)
A
P
7
5
(
%)
MS
-
C
O
C
O
3
2
7
5
4
.
1
7
7
.
2
6
9
.
3
T
ab
le
2.
C
o
m
p
ar
is
o
n
of
p
r
o
p
o
s
ed
C
NN
-
b
ased
o
b
j
ec
t
d
etec
tio
n
m
o
d
el
w
ith
e
x
i
s
ti
n
g
al
g
o
r
it
h
m
s
M
o
d
e
l
A
r
c
h
i
t
e
c
t
u
r
e
A
P
7
5
(
%)
A
P
5
0
(
%)
mA
P
(
%)
F
P
S
SSD
VGG
3
0
.
3
4
8
.
5
2
8
.
8
36
Y
O
L
O
v
3
D
a
r
k
n
e
t
-
53
3
4
.
3
58
33
66
Ef
f
i
c
i
e
n
t
D
e
t
Ef
f
i
c
i
e
n
t
N
e
t
3
5
.
8
5
2
.
2
3
3
.
8
16
Y
O
L
O
v
4
t
i
n
y
C
S
P
N
e
t
-
15
20
40
22
3
3
0
R
e
t
i
n
a
N
e
t
R
e
sN
e
t
1
0
1
3
6
.
8
5
3
.
1
3
4
.
4
11
P
r
o
p
o
se
d
VGG
-
19
6
9
.
3
7
7
.
2
5
4
.
1
3
2
7
Fig
u
r
e
4.
P
er
f
o
r
m
a
n
ce
an
al
y
s
i
s
of
ex
i
s
ti
n
g
ap
p
r
o
ac
h
es
w
it
h
p
r
o
p
o
s
ed
d
etec
tio
n
m
o
d
el
T
h
e
o
u
tco
m
e
s
of
ev
er
y
e
x
p
e
r
i
m
en
tatio
n
ar
e
ex
a
m
i
n
ed
to
co
n
f
ir
m
th
e
e
f
f
icien
c
y
of
t
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
.
Fo
r
ass
ess
m
e
n
t,
AP
is
u
tili
ze
d
,
th
at
co
n
ce
r
n
s
to
th
e
r
eg
io
n
u
n
d
er
th
e
p
r
ec
is
io
n
-
r
e
ca
ll
cu
r
v
e.
Us
u
all
y
,
AP
is
co
m
p
u
ted
f
o
r
all
cla
s
s
e
s
,
an
d
its
a
v
er
ag
e
is
d
eter
m
i
n
ed
as
th
e
m
A
P
.
In
ad
d
itio
n
,
t
h
e
A
P
5
0
d
en
o
tes
to
th
e
50%
r
eg
io
n
co
r
r
ec
tl
y
d
etec
ted
in
co
m
p
ar
is
o
n
to
th
e
g
r
o
u
n
d
tr
u
t
h
,
an
d
f
o
r
A
P
7
5
ca
n
d
id
ate
i
m
ag
e
s
o
v
er
75%
p
ar
ts
ar
e
co
n
s
id
er
ed
.
T
h
is
s
tu
d
y
as
s
u
r
ed
th
e
o
p
er
atio
n
of
t
h
e
m
o
d
el
f
o
r
r
ea
l
-
t
i
m
e
ap
p
licatio
n
s
w
i
th
a
g
o
o
d
r
ec
o
g
n
itio
n
ac
c
u
r
aten
e
s
s
.
4.
CO
NCLU
SI
O
N
We
h
av
e
n
o
ticed
th
at
r
ec
en
t
ef
f
o
r
ts
on
o
b
j
ec
t
d
etec
tio
n
u
s
i
n
g
C
NN
-
b
ased
en
co
d
er
-
d
ec
o
d
er
m
o
d
el
s
h
av
e
ad
d
r
ess
ed
s
alie
n
t
o
b
j
ec
t
d
etec
tio
n
(
SOD
)
as
a
clas
s
i
f
ic
atio
n
tas
k
at
t
h
e
p
ix
el
le
v
el.
T
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
w
a
s
d
e
m
o
n
s
tr
ated
th
r
o
u
g
h
e
x
p
er
im
e
n
tal
f
i
n
d
in
g
s
on
t
h
e
o
p
en
-
s
o
u
r
ce
MS
-
C
OC
O
2017
d
ataset
to
be
ca
p
ab
le
of
g
o
o
d
d
etec
tio
n
ac
c
u
r
ac
y
a
n
d
q
u
ick
e
x
ec
u
tio
n
.
T
h
e
o
b
j
ec
tiv
e
of
th
is
w
o
r
k
g
o
i
n
g
f
o
r
w
ar
d
is
to
s
i
g
n
i
f
ican
t
l
y
en
h
a
n
ce
m
u
ltip
le
o
b
j
ec
t
d
etec
t
io
n
f
o
r
h
i
g
h
q
u
al
it
y
i
m
ag
e
s
w
i
th
o
u
t
s
ac
r
i
f
ici
n
g
p
r
ed
ictio
n
s
p
ee
d
.
It
e
m
p
lo
y
s
t
h
e
u
n
iq
u
e
tec
h
n
iq
u
e
of
p
o
o
lin
g
in
d
ices
as
w
el
l,
w
h
ich
u
s
e
s
f
e
w
er
p
r
o
ce
s
s
in
g
p
ar
a
m
ete
r
s
an
d
s
p
ee
d
s
up
in
f
er
en
ce
.
W
ith
a
m
A
P
of
5
4
.
1
an
d
327
FP
S
,
th
e
s
u
g
g
ested
n
et
w
o
r
k
m
o
d
el
is
h
ig
h
l
y
s
u
i
ted
f
o
r
m
u
ltip
le
o
b
j
ec
t
id
en
ti
f
icatio
n
.
To
s
u
m
u
p
,
t
h
e
m
o
d
el
's
ea
s
e
of
tr
ai
n
i
n
g
an
d
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
's
lo
w
co
m
p
u
tatio
n
al
r
eso
u
r
c
e
r
eq
u
ir
e
m
en
ts
ar
e
its
k
e
y
f
ea
tu
r
es.
As
a
r
e
s
u
l
t,
th
e
s
u
g
g
ested
ap
p
r
o
ac
h
is
p
r
ac
tical
f
o
r
m
a
n
y
r
ea
l
-
ti
m
e
ap
p
licatio
n
s
a
n
d
o
f
f
er
s
a
m
o
r
e
ec
o
n
o
m
ical
alter
n
ati
v
e.
O
v
e
r
all,
th
e
s
u
g
g
ested
m
et
h
o
d
r
es
u
lts
in
a
s
y
s
te
m
f
o
r
cu
tti
n
g
-
ed
g
e
au
to
d
r
iv
i
n
g
s
y
s
t
e
m
s
t
h
at
is
m
o
r
e
af
f
o
r
d
ab
le
an
d
m
o
r
e
ef
f
ec
ti
v
e
.
F
UNDIN
G
I
NF
O
RM
AT
I
O
N
Fu
n
d
in
g
i
n
f
o
r
m
at
io
n
i
s
n
o
t a
v
ailab
le.
AUTHO
R
CO
NT
RIB
UT
I
O
NS ST
A
T
E
M
E
NT
T
h
is
j
o
u
r
n
al
u
s
e
s
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT
)
to
r
ec
o
g
n
ize
in
d
i
v
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
t
h
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
lla
b
o
r
atio
n
.
0
20
40
60
80
100
S
S
D
Y
O
LO
v
3
E
f
f
i
c
i
e
nt
D
e
t
Y
O
LO
v
4 t
i
ny
R
e
t
i
na
N
e
t
P
r
o
po
se
d
P
e
r
ce
nt
age
(%
)
M
ode
l
s
P
e
r
fo
r
m
ance
C
om
pari
s
on
A
P
75 (
%)
A
P
50 (
%)
m
A
P
(
%
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
b
a
s
ed
en
c
o
d
er
-
d
ec
o
d
er fo
r
efficien
t rea
l
-
time
…
(
Mo
th
ir
a
m
R
a
ja
s
ek
a
r
a
n
)
1965
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Mo
th
ir
a
m
R
aj
asek
ar
an
C
h
i
tr
a
Sab
ap
ath
y
R
an
g
a
n
at
h
a
n
Nag
ar
aj
an
Mo
h
an
k
u
m
ar
R
aj
esh
k
u
m
ar
Sa
m
p
a
th
r
aj
an
T
h
ay
ala
g
ar
an
Me
r
li
n
I
n
b
a
m
al
ar
Nag
es
h
v
ar
an
Na
n
d
h
i
n
i
Sh
a
n
m
u
g
a
m
S
u
j
ath
a
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
si
s
I
:
I
n
v
e
st
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
si
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
h
a
v
e
n
o
co
n
f
lict o
f
in
ter
est r
elev
a
n
t to
th
is
p
ap
er
.
DA
T
A
AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
t
h
i
s
s
t
u
d
y
ar
e
a
v
aila
b
le
o
n
r
eq
u
est
f
r
o
m
t
h
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
,
[
NM
]
.
T
h
e
d
ata,
w
h
ich
co
n
tai
n
i
n
f
o
r
m
atio
n
t
h
at
co
u
ld
co
m
p
r
o
m
i
s
e
t
h
e
p
r
iv
ac
y
o
f
r
esear
c
h
p
ar
ticip
an
ts
,
ar
e
n
o
t p
u
b
licl
y
a
v
ailab
le
d
u
e
to
ce
r
tain
r
estrict
i
o
n
s.
RE
F
E
R
E
NC
E
S
[
1
]
Z
.
Z
o
u
,
K
.
C
h
e
n
,
Z
.
S
h
i
,
Y
.
G
u
o
,
a
n
d
J.
Y
e
,
“
O
b
j
e
c
t
d
e
t
e
c
t
i
o
n
i
n
2
0
y
e
a
r
s:
a
su
r
v
e
y
,
”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
E
E
,
v
o
l
.
1
1
1
,
n
o
.
3
,
p
p
.
2
5
7
–
2
7
6
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
JP
R
O
C
.
2
0
2
3
.
3
2
3
8
5
2
4
.
[
2
]
Z
.
L
i
e
t
a
l
.
,
“
D
e
e
p
l
e
a
r
n
i
n
g
-
b
a
se
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
t
e
c
h
n
i
q
u
e
s
f
o
r
r
e
mo
t
e
se
n
si
n
g
i
mag
e
s:
a
s
u
r
v
e
y
,
”
Re
m
o
t
e
S
e
n
si
n
g
,
v
o
l
.
1
4
,
n
o
.
1
0
,
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
r
s1
4
1
0
2
3
8
5
.
[
3
]
J.
Je
g
a
n
,
M
.
R
.
S
u
g
u
n
a
,
M
.
S
h
o
b
a
n
a
,
H
.
A
z
a
t
h
,
S
.
M
u
r
u
g
a
n
,
a
n
d
M
.
R
a
j
mo
h
a
n
,
“
I
o
T
-
e
n
a
b
l
e
d
b
l
a
c
k
b
o
x
f
o
r
d
r
i
v
e
r
b
e
h
a
v
i
o
r
a
n
a
l
y
si
s
u
s
i
n
g
c
l
o
u
d
c
o
m
p
u
t
i
n
g
,
”
i
n
2
0
2
4
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
A
d
v
a
n
c
e
s
i
n
D
a
t
a
E
n
g
i
n
e
e
ri
n
g
a
n
d
I
n
t
e
l
l
i
g
e
n
t
C
o
m
p
u
t
i
n
g
S
y
s
t
e
m
s (AD
I
C
S
)
,
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
A
D
I
C
S
5
8
4
4
8
.
2
0
2
4
.
1
0
5
3
3
4
7
1
.
[
4
]
K
.
M
u
h
a
mm
a
d
,
J.
A
h
mad
,
Z
.
L
v
,
P
.
B
e
l
l
a
v
i
st
a
,
P
.
Y
a
n
g
,
a
n
d
S
.
W
.
B
a
i
k
,
“
Ef
f
i
c
i
e
n
t
d
e
e
p
C
N
N
-
b
a
se
d
f
i
r
e
d
e
t
e
c
t
i
o
n
a
n
d
l
o
c
a
l
i
z
a
t
i
o
n
i
n
v
i
d
e
o
su
r
v
e
i
l
l
a
n
c
e
a
p
p
l
i
c
a
t
i
o
n
s,
”
I
EEE
T
r
a
n
sa
c
t
i
o
n
s
o
n
S
y
s
t
e
m
s,
M
a
n
,
a
n
d
C
y
b
e
r
n
e
t
i
c
s:
S
y
s
t
e
m
s
,
v
o
l
.
4
9
,
n
o
.
7
,
p
p
.
1
4
1
9
–
1
4
3
4
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
T
S
M
C
.
2
0
1
8
.
2
8
3
0
0
9
9
.
[
5
]
J.
-
M
.
G
u
o
,
J.
-
S
.
Y
a
n
g
,
S
.
S
e
sh
a
t
h
i
r
i
,
a
n
d
H
.
-
W
.
W
u
,
“
A
l
i
g
h
t
-
w
e
i
g
h
t
C
N
N
f
o
r
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
sp
a
r
se
mo
d
e
l
a
n
d
k
n
o
w
l
e
d
g
e
d
i
s
t
i
l
l
a
t
i
o
n
,
”
E
l
e
c
t
r
o
n
i
c
s
,
v
o
l
.
1
1
,
n
o
.
4
,
F
e
b
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s
1
1
0
4
0
5
7
5
.
[
6
]
S
.
S
r
i
n
i
v
a
s
a
n
,
R
.
R
a
j
a
,
C
.
Je
h
a
n
,
S
.
M
u
r
u
g
a
n
,
C
.
S
r
i
n
i
v
a
sa
n
,
a
n
d
M
.
M
u
t
h
u
l
e
k
sh
m
i
,
“
I
o
T
-
e
n
a
b
l
e
d
f
a
c
i
a
l
r
e
c
o
g
n
i
t
i
o
n
f
o
r
sm
a
r
t
h
o
sp
i
t
a
l
i
t
y
f
o
r
c
o
n
t
a
c
t
l
e
ss
g
u
e
st
se
r
v
i
c
e
s
a
n
d
i
d
e
n
t
i
t
y
v
e
r
i
f
i
c
a
t
i
o
n
,
”
i
n
2
0
2
4
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
R
e
l
i
a
b
i
l
i
t
y
,
I
n
f
o
c
o
m
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
O
p
t
i
m
i
z
a
t
i
o
n
(
T
ren
d
s
a
n
d
F
u
t
u
re
D
i
r
e
c
t
i
o
n
s)
(
I
C
RI
T
O
)
,
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
I
T
O
6
1
5
2
3
.
2
0
2
4
.
1
0
5
2
2
3
6
3
.
[
7
]
J.
R
e
d
mo
n
a
n
d
A
.
F
a
r
h
a
d
i
,
“
Y
o
l
o
v
3
:
A
n
i
n
c
r
e
me
n
t
a
l
i
mp
r
o
v
e
me
n
t
,
”
a
rX
i
v
-
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
p
p
.
1
–
6
,
2
0
1
8
,
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
1
8
0
4
.
0
2
7
6
7
.
[
8
]
Y
.
H
.
W
u
,
Y
.
L
i
u
,
X
.
Z
h
a
n
,
a
n
d
M
.
M
.
C
h
e
n
g
,
“
P
2
T
:
p
y
r
a
mi
d
p
o
o
l
i
n
g
t
r
a
n
sf
o
r
me
r
f
o
r
sce
n
e
u
n
d
e
r
st
a
n
d
i
n
g
,
”
I
EEE
T
r
a
n
sa
c
t
i
o
n
s
o
n
P
a
t
t
e
r
n
A
n
a
l
y
si
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
5
,
n
o
.
1
1
,
p
p
.
1
2
7
6
0
–
1
2
7
7
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
T
P
A
M
I
.
2
0
2
2
.
3
2
0
2
7
6
5
.
[
9
]
H
.
Ji
a
n
g
a
n
d
E.
L
e
a
r
n
e
d
-
M
i
l
l
e
r
,
“
F
a
c
e
d
e
t
e
c
t
i
o
n
w
i
t
h
t
h
e
f
a
st
e
r
R
-
C
N
N
,
”
i
n
2
0
1
7
1
2
t
h
I
EE
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Au
t
o
m
a
t
i
c
Fa
c
e
&
G
e
st
u
r
e
R
e
c
o
g
n
i
t
i
o
n
(
FG
2
0
1
7
)
,
M
a
y
2
0
1
7
,
p
p
.
6
5
0
–
6
5
7
,
d
o
i
:
1
0
.
1
1
0
9
/
F
G
.
2
0
1
7
.
8
2
.
[
1
0
]
Z
.
W
a
n
g
,
J.
Z
h
u
,
S
.
F
u
,
S
.
M
a
o
,
a
n
d
Y
.
Y
e
,
“
R
F
P
N
e
t
:
R
e
o
r
g
a
n
i
z
i
n
g
f
e
a
t
u
r
e
p
y
r
a
mi
d
n
e
t
w
o
r
k
s
f
o
r
m
e
d
i
c
a
l
i
mag
e
s
e
g
me
n
t
a
t
i
o
n
,
”
C
o
m
p
u
t
e
rs
i
n
B
i
o
l
o
g
y
a
n
d
Me
d
i
c
i
n
e
,
v
o
l
.
1
6
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
b
i
o
me
d
.
2
0
2
3
.
1
0
7
1
0
8
.
[
1
1
]
A
.
T
r
a
g
a
k
i
s,
C
.
K
a
u
l
,
R
.
M
u
r
r
a
y
-
S
mi
t
h
,
a
n
d
D
.
H
u
sme
i
e
r
,
“
T
h
e
f
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
t
r
a
n
sf
o
r
me
r
f
o
r
m
e
d
i
c
a
l
i
m
a
g
e
seg
me
n
t
a
t
i
o
n
,
”
i
n
2
0
2
3
I
EE
E/
C
VF
W
i
n
t
e
r
C
o
n
f
e
re
n
c
e
o
n
A
p
p
l
i
c
a
t
i
o
n
s
o
f
C
o
m
p
u
t
e
r
Vi
s
i
o
n
(
WA
C
V)
,
2
0
2
3
,
p
p
.
3
6
4
9
–
3
6
5
8
,
d
o
i
:
1
0
.
1
1
0
9
/
W
A
C
V
5
6
6
8
8
.
2
0
2
3
.
0
0
3
6
5
.
[
1
2
]
J.
R
a
m
a
samy
,
E.
S
r
i
v
i
d
h
y
a
,
V
.
V
a
i
d
e
h
i
,
S
.
V
i
mal
a
d
e
v
i
,
N
.
M
o
h
a
n
k
u
mar
,
a
n
d
S
.
M
u
r
u
g
a
n
,
“
C
l
o
u
d
-
e
n
a
b
l
e
d
i
s
o
l
a
t
i
o
n
f
o
r
e
st
f
o
r
a
n
o
mal
y
d
e
t
e
c
t
i
o
n
i
n
U
A
V
-
b
a
se
d
p
o
w
e
r
l
i
n
e
i
n
sp
e
c
t
i
o
n
,
”
i
n
2
0
2
4
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
N
e
t
w
o
rki
n
g
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
s (I
C
N
WC
)
,
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
N
W
C
6
0
7
7
1
.
2
0
2
4
.
1
0
5
3
7
4
0
7
.
[
1
3
]
D
.
B
a
i
,
X
.
Z
h
e
n
g
,
T
.
L
i
u
,
K
.
L
i
,
a
n
d
J.
Y
a
n
g
,
“
F
i
n
g
e
r
d
i
sab
i
l
i
t
y
r
e
c
o
g
n
i
t
i
o
n
b
a
se
d
o
n
h
o
l
i
st
i
c
a
l
l
y
-
n
e
st
e
d
e
d
g
e
d
e
t
e
c
t
i
o
n
,
”
i
n
I
n
t
e
l
l
i
g
e
n
t
Ro
b
o
t
i
c
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
2
0
2
2
,
p
p
.
1
4
6
–
1
5
4
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
1
3
8
4
4
-
7
_
1
5
.
[
1
4
]
M
.
R
.
S
u
d
h
a
e
t
a
l
.
,
“
P
r
e
d
i
c
t
i
v
e
mo
d
e
l
i
n
g
f
o
r
h
e
a
l
t
h
c
a
r
e
w
o
r
k
e
r
w
e
l
l
-
b
e
i
n
g
w
i
t
h
c
l
o
u
d
c
o
mp
u
t
i
n
g
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
f
o
r
st
r
e
ss
man
a
g
e
me
n
t
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
El
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
5
,
n
o
.
1
,
p
p
.
1
2
1
8
–
1
2
2
8
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
c
e
.
v
1
5
i
1
.
p
p
1
2
1
8
-
1
2
2
8
.
[
1
5
]
Y
.
X
i
e
,
Y
.
L
u
,
a
n
d
S
.
W
a
n
g
,
“
R
S
A
N
e
t
:
d
e
e
p
r
e
c
u
r
r
e
n
t
sca
l
e
-
a
w
a
r
e
n
e
t
w
o
r
k
f
o
r
c
r
o
w
d
c
o
u
n
t
i
n
g
,
”
Pr
o
c
e
e
d
i
n
g
s
-
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
m
a
g
e
Pro
c
e
ss
i
n
g
,
I
C
I
P
,
p
p
.
1
5
3
1
–
1
5
3
5
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
P
4
0
7
7
8
.
2
0
2
0
.
9
1
9
1
0
8
6
.
[
1
6
]
I
.
F
i
l
a
l
i
,
M
.
S
.
A
l
l
i
l
i
,
a
n
d
N
.
B
e
n
b
l
i
d
i
a
,
“
M
u
l
t
i
-
sc
a
l
e
sal
i
e
n
t
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
u
si
n
g
g
r
a
p
h
r
a
n
k
i
n
g
a
n
d
g
l
o
b
a
l
–
l
o
c
a
l
sal
i
e
n
c
y
r
e
f
i
n
e
me
n
t
,
”
S
i
g
n
a
l
Pro
c
e
ss
i
n
g
:
I
m
a
g
e
C
o
m
m
u
n
i
c
a
t
i
o
n
,
v
o
l
.
4
7
,
p
p
.
3
8
0
–
4
0
1
,
2
0
1
6
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
m
a
g
e
.
2
0
1
6
.
0
7
.
0
0
7
.
[
1
7
]
Z
.
W
u
,
G
.
A
l
l
i
b
e
r
t
,
F
.
M
e
r
i
a
u
d
e
a
u
,
C
.
M
a
,
a
n
d
C
.
D
e
mo
n
c
e
a
u
x
,
“
H
i
D
A
n
e
t
:
R
G
B
-
D
sal
i
e
n
t
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
v
i
a
h
i
e
r
a
r
c
h
i
c
a
l
d
e
p
t
h
a
w
a
r
e
n
e
ss,”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
I
m
a
g
e
Pr
o
c
e
ssi
n
g
,
v
o
l
.
3
2
,
p
p
.
2
1
6
0
–
2
1
7
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
P
.
2
0
2
3
.
3
2
6
3
1
1
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
I
n
t J
A
r
ti
f
I
n
tell
,
Vo
l.
14
,
No
.
3
,
J
u
n
e
2
0
2
5
:
1
9
6
0
-
1967
1966
[
1
8
]
P
.
M
a
h
e
s
w
a
r
i
,
S
.
G
o
w
r
i
sw
a
r
i
,
S
.
B
a
l
a
s
u
b
r
a
m
a
n
i
,
A
.
R
.
B
a
b
u
,
N
.
K
.
J
i
j
i
t
h
,
a
n
d
S
.
M
u
r
u
g
a
n
,
“
I
n
t
e
l
l
i
g
e
n
t
h
e
a
d
l
i
g
h
t
s
f
o
r
a
d
a
p
t
i
n
g
b
e
a
m
p
a
t
t
e
r
n
s
w
i
t
h
r
a
s
p
b
e
r
r
y
p
i
a
n
d
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
,
”
i
n
2
0
2
4
2
n
d
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
D
e
v
i
c
e
I
n
t
e
l
l
i
g
e
n
c
e
,
C
o
m
p
u
t
i
n
g
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
i
e
s
(
D
I
C
C
T
)
,
2
0
2
4
,
p
p
.
1
8
2
–
1
8
7
,
d
o
i
:
1
0
.
1
1
0
9
/
D
I
C
C
T
6
1
0
3
8
.
2
0
2
4
.
1
0
5
3
3
1
5
9
.
[
1
9
]
J.
H
a
i
,
Y
.
H
a
o
,
F
.
Z
o
u
,
F
.
L
i
n
,
a
n
d
S
.
H
a
n
,
“
A
d
v
a
n
c
e
d
R
e
t
i
n
e
x
N
e
t
:
A
f
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
f
o
r
l
o
w
-
l
i
g
h
t
i
mag
e
e
n
h
a
n
c
e
me
n
t
,
”
S
i
g
n
a
l
Pro
c
e
ss
i
n
g
:
I
m
a
g
e
C
o
m
m
u
n
i
c
a
t
i
o
n
,
v
o
l
.
1
1
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
mag
e
.
2
0
2
2
.
1
1
6
9
1
6
.
[
2
0
]
D
.
S
t
a
v
e
n
s
a
n
d
S
.
T
h
r
u
n
,
“
U
n
su
p
e
r
v
i
se
d
l
e
a
r
n
i
n
g
o
f
i
n
v
a
r
i
a
n
t
f
e
a
t
u
r
e
s
u
s
i
n
g
v
i
d
e
o
,
”
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
Pa
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
p
p
.
1
6
4
9
–
1
6
5
6
,
2
0
1
0
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
0
.
5
5
3
9
7
7
3
.
[
2
1
]
C
.
C
.
S
e
k
h
a
r
,
K
.
V
i
j
a
y
a
l
a
k
sh
m
i
,
A
.
S
.
R
a
o
,
V
.
V
e
d
a
n
a
r
a
y
a
n
a
n
,
M
.
B
.
S
a
h
a
a
i
,
a
n
d
S
.
M
u
r
u
g
a
n
,
“
C
l
o
u
d
-
b
a
se
d
w
a
t
e
r
t
a
n
k
man
a
g
e
me
n
t
a
n
d
c
o
n
t
r
o
l
sy
st
e
m,”
i
n
2
0
2
3
2
n
d
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
S
m
a
r
t
T
e
c
h
n
o
l
o
g
i
e
s
f
o
r S
m
a
r
t
N
a
t
i
o
n
,
S
m
a
r
t
T
e
c
h
C
o
n
2
0
2
3
,
2
0
2
3
,
p
p
.
6
4
1
–
6
4
6
,
d
o
i
:
1
0
.
1
1
0
9
/
S
m
a
r
t
T
e
c
h
C
o
n
5
7
5
2
6
.
2
0
2
3
.
1
0
3
9
1
7
3
0
.
[
2
2
]
S
.
M
.
H
a
n
i
f
,
L
.
P
r
e
v
o
st
,
R
.
B
e
l
a
r
o
u
ss
i
,
a
n
d
M
.
M
i
l
g
r
a
m,
“
R
e
a
l
-
t
i
me
f
a
c
i
a
l
f
e
a
t
u
r
e
l
o
c
a
l
i
z
a
t
i
o
n
b
y
c
o
mb
i
n
i
n
g
sp
a
c
e
d
i
s
p
l
a
c
e
me
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
L
e
t
t
e
rs
,
v
o
l
.
2
9
,
n
o
.
8
,
p
p
.
1
0
9
4
–
1
1
0
4
,
2
0
0
8
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
a
t
r
e
c
.
2
0
0
7
.
0
9
.
0
1
6
.
[
2
3
]
B
.
J
.
G
a
n
e
sh
,
P
.
V
i
j
a
y
a
n
,
V
.
V
a
i
d
e
h
i
,
S
.
M
u
r
u
g
a
n
,
R
.
M
e
e
n
a
k
s
h
i
,
a
n
d
M
.
R
a
j
mo
h
a
n
,
“
S
V
M
-
b
a
se
d
p
r
e
d
i
c
t
i
v
e
mo
d
e
l
i
n
g
o
f
d
r
o
w
si
n
e
ss
i
n
h
o
sp
i
t
a
l
st
a
f
f
f
o
r
o
c
c
u
p
a
t
i
o
n
a
l
safe
t
y
s
o
l
u
t
i
o
n
v
i
a
I
o
T
i
n
f
r
a
st
r
u
c
t
u
r
e
,
”
i
n
2
0
2
4
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
C
o
n
t
r
o
l
(
I
C
4
)
,
2
0
2
4
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
4
5
7
4
3
4
.
2
0
2
4
.
1
0
4
8
6
4
2
9
.
[
2
4
]
N
.
L
i
u
,
J.
H
a
n
,
a
n
d
M
.
H
.
Y
a
n
g
,
“
P
i
C
A
N
e
t
:
p
i
x
e
l
-
w
i
se
c
o
n
t
e
x
t
u
a
l
a
t
t
e
n
t
i
o
n
l
e
a
r
n
i
n
g
f
o
r
a
c
c
u
r
a
t
e
sal
i
e
n
c
y
d
e
t
e
c
t
i
o
n
,
”
I
EE
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
m
a
g
e
Pro
c
e
ss
i
n
g
,
v
o
l
.
2
9
,
p
p
.
6
4
3
8
–
6
4
5
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
T
I
P
.
2
0
2
0
.
2
9
8
8
5
6
8
.
[
2
5
]
A
.
R
.
R
a
t
h
i
n
a
m,
B
.
S
.
V
a
t
h
a
n
i
,
A
.
K
o
mat
h
i
,
J
.
L
e
n
i
n
,
B
.
B
h
a
r
a
t
h
i
,
a
n
d
S
.
M
.
U
r
u
g
a
n
,
“
A
d
v
a
n
c
e
s
a
n
d
p
r
e
d
i
c
t
i
o
n
s
i
n
p
r
e
d
i
c
t
i
v
e
a
u
t
o
-
sca
l
i
n
g
a
n
d
ma
i
n
t
e
n
a
n
c
e
a
l
g
o
r
i
t
h
ms
f
o
r
c
l
o
u
d
c
o
mp
u
t
i
n
g
,
”
2
n
d
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
A
u
t
o
m
a
t
i
o
n
,
C
o
m
p
u
t
i
n
g
a
n
d
Re
n
e
w
a
b
l
e
S
y
st
e
m
s,
I
C
A
C
RS
2
0
2
3
-
Pro
c
e
e
d
i
n
g
s
,
p
p
.
3
9
5
–
4
0
0
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
C
R
S
5
8
5
7
9
.
2
0
2
3
.
1
0
4
0
4
1
8
6
.
[
2
6
]
M
.
D
.
A
.
H
a
sa
n
,
K
.
B
a
l
a
s
u
b
a
d
r
a
,
G
.
V
a
d
i
v
e
l
,
N
.
A
r
u
n
f
r
e
d
,
M
.
V
.
I
sh
w
a
r
y
a
,
a
n
d
S
.
M
u
r
u
g
a
n
,
“
I
o
T
-
d
r
i
v
e
n
i
mag
e
r
e
c
o
g
n
i
t
i
o
n
f
o
r
mi
c
r
o
p
l
a
s
t
i
c
a
n
a
l
y
si
s
i
n
w
a
t
e
r
s
y
st
e
ms
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,”
i
n
2
0
2
4
2
n
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r,
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
C
o
n
t
r
o
l
(
I
C
4
)
,
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
4
5
7
4
3
4
.
2
0
2
4
.
1
0
4
8
6
4
9
0
.
[
2
7
]
S
.
S
e
l
v
a
r
a
su
,
K
.
B
a
sh
k
a
r
a
n
,
K
.
R
a
d
h
i
k
a
,
S
.
V
a
l
a
r
m
a
t
h
y
,
a
n
d
S
.
M
u
r
u
g
a
n
,
“
I
o
T
-
e
n
a
b
l
e
d
me
d
i
c
a
t
i
o
n
safe
t
y
:
r
e
a
l
-
t
i
me
t
e
mp
e
r
a
t
u
r
e
a
n
d
s
t
o
r
a
g
e
mo
n
i
t
o
r
i
n
g
f
o
r
e
n
h
a
n
c
e
d
me
d
i
c
a
t
i
o
n
q
u
a
l
i
t
y
i
n
h
o
s
p
i
t
a
l
s,”
2
n
d
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
Au
t
o
m
a
t
i
o
n
,
C
o
m
p
u
t
i
n
g
a
n
d
Re
n
e
w
a
b
l
e
S
y
s
t
e
m
s,
I
C
A
C
R
S
2
0
2
3
-
Pr
o
c
e
e
d
i
n
g
s
,
p
p
.
2
5
6
–
2
6
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
C
R
S
5
8
5
7
9
.
2
0
2
3
.
1
0
4
0
5
2
1
2
.
[
2
8
]
K
.
P
a
d
ma
n
a
b
a
n
,
A
.
M
.
S
.
K
u
mar,
H
.
A
z
a
t
h
,
A
.
K
.
V
e
l
mu
r
u
g
a
n
,
a
n
d
M
.
S
u
b
b
i
a
h
,
“
H
y
b
r
i
d
d
a
t
a
m
i
n
i
n
g
t
e
c
h
n
i
q
u
e
b
a
se
d
b
r
e
a
s
t
c
a
n
c
e
r
p
r
e
d
i
c
t
i
o
n
,
”
AI
P
C
o
n
f
e
re
n
c
e
P
ro
c
e
e
d
i
n
g
s
,
v
o
l
.
2
5
2
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
6
3
/
5
.
0
1
1
0
2
1
6
.
[
2
9
]
N
.
M
o
h
a
n
k
u
m
a
r
e
t
a
l
.
,
“
A
d
v
a
n
c
i
n
g
c
h
r
o
n
i
c
p
a
i
n
r
e
l
i
e
f
c
l
o
u
d
-
b
a
se
d
r
e
mo
t
e
man
a
g
e
me
n
t
w
i
t
h
mac
h
i
n
e
l
e
a
r
n
i
n
g
i
n
h
e
a
l
t
h
c
a
r
e
,
”
I
n
d
o
n
e
si
a
n
J
o
u
rn
a
l
o
f
E
l
e
c
t
r
i
c
a
l
En
g
i
n
e
e
ri
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
3
7
,
n
o
.
2
,
p
p
.
1
0
4
2
–
1
0
5
2
,
2
0
2
5
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
e
c
s.v
3
7
.
i
2
.
p
p
1
0
4
2
-
1
0
5
2
.
[
3
0
]
M
.
V
a
d
i
v
e
l
,
V
.
B
.
M
a
r
i
n
,
S
.
B
a
l
a
su
b
r
a
ma
n
i
,
S
.
H
e
mal
a
t
h
a
,
S
.
M
u
r
u
g
a
n
,
a
n
d
S
.
V
e
l
mu
r
u
g
a
n
,
“
C
l
o
u
d
-
b
a
se
d
p
a
ss
e
n
g
e
r
e
x
p
e
r
i
e
n
c
e
man
a
g
e
me
n
t
i
n
b
u
s
f
a
r
e
t
i
c
k
e
t
i
n
g
sy
s
t
e
ms
u
si
n
g
r
a
n
d
o
m
f
o
r
e
st
a
l
g
o
r
i
t
h
m,”
i
n
2
0
2
4
1
1
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
R
e
l
i
a
b
i
l
i
t
y
,
I
n
f
o
c
o
m
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
O
p
t
i
m
i
z
a
t
i
o
n
(
T
ren
d
s
a
n
d
Fu
t
u
r
e
D
i
rec
t
i
o
n
s)
(
I
C
RI
T
O
)
,
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
I
T
O
6
1
5
2
3
.
2
0
2
4
.
1
0
5
2
2
2
2
6
.
[
3
1
]
M
.
P
.
A
a
r
t
h
i
,
C
.
M
.
R
e
d
d
y
,
A
.
A
n
b
a
r
a
si
,
N
.
M
o
h
a
n
k
u
mar
,
M
.
V
.
I
sh
w
a
r
y
a
,
a
n
d
S
.
M
u
r
u
g
a
n
,
“
C
l
o
u
d
-
b
a
se
d
r
o
a
d
safe
t
y
f
o
r
r
e
a
l
-
t
i
me
v
e
h
i
c
l
e
r
a
sh
d
r
i
v
i
n
g
a
l
e
r
t
s
w
i
t
h
r
a
n
d
o
m
f
o
r
e
st
a
l
g
o
r
i
t
h
m,”
i
n
2
0
2
4
3
r
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
f
o
r
I
n
n
o
v
a
t
i
o
n
i
n
T
e
c
h
n
o
l
o
g
y
(
I
N
O
C
O
N
)
,
2
0
2
4
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
N
O
C
O
N
6
0
7
5
4
.
2
0
2
4
.
1
0
5
1
1
3
1
6
.
[
3
2
]
M
.
S
.
K
u
mar,
H
.
A
z
a
t
h
,
A
.
K
.
V
e
l
m
u
r
u
g
a
n
,
K
.
P
a
d
ma
n
a
b
a
n
,
a
n
d
M
.
S
u
b
b
i
a
h
,
“
P
r
e
d
i
c
t
i
o
n
o
f
A
l
z
h
e
i
me
r
’
s
d
i
se
a
se
u
si
n
g
h
y
b
r
i
d
mac
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
,
”
AI
P
C
o
n
f
e
re
n
c
e
Pr
o
c
e
e
d
i
n
g
s
,
v
o
l
.
2
5
2
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
6
3
/
5
.
0
1
1
0
2
8
3
.
[
3
3
]
E.
P
.
K
a
n
n
a
n
a
n
d
T
.
V
.
C
h
i
t
h
r
a
,
“
L
a
g
r
a
n
g
e
i
n
t
e
r
p
o
l
a
t
i
o
n
f
o
r
n
a
t
u
r
a
l
c
o
l
o
u
r
i
mag
e
d
e
m
o
sai
c
i
n
g
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ad
v
a
n
c
e
s i
n
S
i
g
n
a
l
a
n
d
I
m
a
g
e
S
c
i
e
n
c
e
s
,
v
o
l
.
7
,
n
o
.
2
,
p
p
.
2
1
–
3
0
,
2
0
2
1
,
d
o
i
:
1
0
.
2
9
2
8
4
/
i
j
a
s
i
s.
7
.
2
.
2
0
2
1
.
2
1
-
3
0
.
B
I
O
G
RAP
H
I
E
S
OF
AUTH
O
RS
M
o
th
ira
m
Ra
j
a
se
k
a
r
a
n
is
an
a
c
c
o
m
p
li
sh
e
d
IT
l
e
a
d
e
r
w
it
h
o
v
e
r
13
y
e
a
r
s
of
e
x
p
e
rien
c
e
in
b
ig
d
a
ta
a
n
d
c
lo
u
d
so
lu
ti
o
n
s.
He
is
re
c
o
g
n
ize
d
f
o
r
h
is
e
x
p
e
rti
se
in
A
p
a
c
h
e
S
p
a
rk
,
Ha
d
o
o
p
,
Hiv
e
,
Im
p
a
la,
a
n
d
o
t
h
e
r
c
u
tt
i
n
g
-
e
d
g
e
tec
h
n
o
lo
g
i
e
s.
He
is
p
ro
f
ici
e
n
c
y
in
AI
a
n
d
m
a
c
h
in
e
lea
rn
in
g
e
n
a
b
les
h
i
m
to
d
e
li
v
e
r
d
a
ta
-
d
riv
e
n
in
sig
h
ts
a
n
d
i
n
n
o
v
a
ti
v
e
so
lu
ti
o
n
s.
He
e
x
c
e
ls
in
c
o
ll
a
b
o
ra
ti
n
g
d
irec
tl
y
w
it
h
c
li
e
n
ts,
lea
d
in
g
th
e
d
e
sig
n
a
n
d
im
p
le
m
e
n
tatio
n
of
d
a
ta
a
n
d
a
p
p
li
c
a
ti
o
n
m
ig
ra
ti
o
n
s
to
p
riv
a
te
a
n
d
p
u
b
li
c
c
lo
u
d
s.
He
is
lea
d
e
rsh
ip
a
c
c
e
l
e
ra
tes
th
e
a
d
o
p
t
io
n
of
e
m
e
rg
in
g
f
e
a
tu
re
s
a
n
d
e
n
s
u
re
s
th
e
d
e
li
v
e
ry
of
a
d
v
a
n
c
e
d
d
a
ta
so
l
u
ti
o
n
s
f
o
r
o
rg
a
n
iza
ti
o
n
s.
He
can
be
co
n
tac
ted
at
e
m
a
il
:
m
o
th
iram
ra
ja
se
k
a
ra
n
s
e
k
a
r@
g
m
a
il
.
c
o
m
.
Ch
itra
S
a
b
a
p
a
t
h
y
Ra
n
g
a
n
a
th
a
n
is
(Cli
e
n
t
P
a
rt
n
e
r
|
A
c
c
o
u
n
t
M
a
n
a
g
e
m
e
n
t
|
IT
T
ra
n
s
f
o
r
m
a
ti
o
n
S
trate
g
y
|
Dig
i
tal
En
g
in
e
e
rin
g
S
o
l
u
ti
o
n
s
&
A
d
v
iso
ry
|
Ag
il
e
De
li
v
e
r
y
A
d
o
p
ti
o
n
|
S
a
les
|
Co
E
&
Co
P
S
e
tu
p
.
Re
su
lt
s
-
o
rie
n
ted
,
a
c
c
o
m
p
li
sh
e
d
b
u
sin
e
ss
tec
h
n
o
l
o
g
y
lea
d
e
r
w
it
h
23+
y
e
a
rs
of
e
x
p
e
rien
c
e
in
so
f
t
w
a
r
e
e
n
g
in
e
e
rin
g
a
n
d
d
e
sig
n
.
P
r
o
v
e
n
trac
k
r
e
c
o
rd
of
c
o
n
c
e
p
tu
a
li
z
in
g
,
a
rc
h
it
e
c
ti
n
g
,
a
n
d
d
e
li
v
e
rin
g
re
li
a
b
le
a
n
d
sc
a
la
b
le
s
y
ste
m
s
in
a
v
a
riet
y
of
a
re
a
s
c
o
m
p
risin
g
m
u
lt
i
-
tec
h
n
o
lo
g
ies
in
c
lu
d
in
g
c
lo
u
d
,
b
ig
d
a
ta
,
A
I,
M
L
,
a
d
v
a
n
c
e
a
n
a
l
y
ti
c
s
,
b
lo
c
k
c
h
a
in
,
m
a
in
f
r
a
m
e
,
a
n
d
b
u
sin
e
ss
in
telli
g
e
n
c
e
.
Ex
e
c
u
ted
c
o
m
p
lex
e
n
g
a
g
e
m
e
n
ts
a
c
ro
ss
m
u
lt
ip
le
v
e
rti
c
a
ls,
m
a
n
a
g
e
sa
les
,
IT
De
li
v
e
r
y
a
n
d
Op
e
ra
ti
o
n
s,
e
sta
b
li
sh
e
d
v
isio
n
,
stra
teg
y
,
a
n
d
jo
u
r
n
e
y
m
a
p
s
th
a
t
a
li
g
n
w
it
h
b
u
sin
e
ss
p
rio
ri
ti
e
s.
En
terp
r
ise
lea
d
e
r
in
d
ig
it
a
l
e
n
g
in
e
e
rin
g
so
lu
ti
o
n
s
&
a
d
v
iso
ry
,
a
g
il
e
d
e
li
v
e
ry
a
d
o
p
ti
o
n
&
m
a
n
a
g
e
m
e
n
t,
p
re
-
sa
les
,
Co
E
&
Co
P
se
tu
p
,
IT
tran
sf
o
r
m
a
ti
o
n
stra
teg
y
,
e
n
terp
ris
e
q
u
a
li
ty
&
d
ig
it
a
l
a
ss
u
ra
n
c
e
.
He
can
be
c
o
n
tac
ted
at
e
m
a
il
:
c
h
it
ra
sa
b
a
p
a
th
y
ra
n
g
a
n
a
th
a
n
@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
A
r
ti
f
I
n
tell
I
SS
N:
2252
-
8938
C
o
n
vo
lu
tio
n
a
l
n
eu
r
a
l n
etw
o
r
k
b
a
s
ed
en
c
o
d
er
-
d
ec
o
d
er fo
r
efficien
t rea
l
-
time
…
(
Mo
th
ir
a
m
R
a
ja
s
ek
a
r
a
n
)
1967
Dr
.
Na
g
a
r
a
j
a
n
M
o
h
a
n
k
u
m
a
r
w
a
s
b
o
rn
in
In
d
ia
in
1
9
7
8
.
He
re
c
e
iv
e
d
h
is
B.
E
.
d
e
g
re
e
f
ro
m
Bh
a
ra
th
iy
a
r
Un
iv
e
r
sity
,
T
a
m
il
n
a
d
u
,
In
d
ia
in
2
0
0
0
a
n
d
M
.
E.
a
n
d
P
h
.
D
d
e
g
re
e
f
ro
m
Ja
d
a
v
p
u
r
Un
iv
e
rsit
y
,
Ko
lk
a
ta
in
2
0
0
4
&
2
0
1
0
.
He
j
o
in
e
d
th
e
Na
n
o
De
v
ice
S
i
m
u
latio
n
L
a
b
o
ra
to
ry
in
2007
a
n
d
w
o
rk
e
d
as
a
S
e
n
io
r
Re
se
a
r
c
h
F
e
ll
o
w
u
n
d
e
r
CS
IR
d
irec
t
S
c
h
e
m
e
till
S
e
p
tem
b
e
r
2009.
L
a
ter
he
jo
in
e
d
S
KP
E
n
g
in
e
e
rin
g
Co
ll
e
g
e
as
a
P
r
o
f
e
ss
o
r
to
d
e
v
e
lo
p
re
se
a
rc
h
a
c
ti
v
it
ies
in
th
e
f
ield
of
V
L
S
I
a
n
d
NA
NO
tec
h
n
o
lo
g
y
.
He
is
c
u
rre
n
tl
y
w
o
rk
in
g
as
a
Re
se
a
rc
h
P
ro
f
e
ss
o
r
at
S
y
m
b
io
si
s
In
stit
u
te
of
Tec
h
n
o
lo
g
y
,
Na
g
p
u
r
Ca
m
p
u
s,
S
y
m
b
io
sis
(In
tern
a
ti
o
n
a
l)
De
e
m
e
d
Un
iv
e
rsity
,
P
u
n
e
,
In
d
ia.
He
is
a
se
n
i
o
r
m
e
m
b
e
r
of
IEE
E.
He
h
a
s
a
b
o
u
t
85
in
tern
a
ti
o
n
a
l
jo
u
rn
a
l
p
u
b
l
ica
ti
o
n
s
i
n
re
p
u
ted
jo
u
rn
a
ls
a
n
d
a
b
o
u
t
50
in
tern
a
ti
o
n
a
l
c
o
n
f
e
re
n
c
e
p
ro
c
e
e
d
in
g
s.
He
re
c
e
iv
e
d
th
e
c
a
r
rier
a
w
a
rd
f
o
r
y
o
u
n
g
tea
c
h
e
rs
(C
A
Y
T
)
f
ro
m
A
IC
T
E,
Ne
w
De
lh
i
in
t
h
e
y
e
a
r
of
2
0
1
2
-
2
0
1
4
.
His
re
se
a
rc
h
in
tere
st
in
c
lu
d
e
s
m
o
d
e
li
n
g
a
n
d
sim
u
latio
n
stu
d
y
of
HEM
Ts,
o
p
ti
m
iza
ti
o
n
of
d
e
v
ice
s
f
o
r
RF
a
p
p
li
c
a
ti
o
n
s
a
n
d
c
h
a
ra
c
teriz
a
ti
o
n
of
a
d
v
a
n
c
e
d
HEM
T
a
rc
h
it
e
c
tu
re
,
tera
h
e
rtz
e
lec
tro
n
ics
,
h
ig
h
f
re
q
u
e
n
c
y
im
a
g
in
g
,
se
n
so
rs
a
n
d
c
o
m
m
u
n
ica
ti
o
n
.
H
e
can
be
c
o
n
tac
ted
at
e
m
a
il
:
n
m
k
p
ro
f
e
ss
o
r@g
m
a
i
l.
c
o
m
.
Ra
je
s
h
k
u
m
a
r
S
a
m
p
a
t
h
r
a
ja
n
is
a
P
r
in
c
ip
a
l
Clo
u
d
A
rc
h
it
e
c
t
at
M
c
Kin
se
y
,
w
h
e
r
e
he
lea
d
s
a
tea
m
of
e
n
g
in
e
e
rs
a
n
d
a
rc
h
it
e
c
ts
in
d
e
sig
n
in
g
a
n
d
b
u
il
d
in
g
h
ig
h
ly
s
c
a
lab
le,
re
sili
e
n
t,
a
n
d
d
istri
b
u
ted
sy
ste
m
s
u
sin
g
th
e
late
st
c
lo
u
d
n
a
ti
v
e
tec
h
n
o
lo
g
y
in
G
o
o
g
le
Clo
u
d
P
latf
o
rm
(
G
CP
).
He
h
a
s
o
v
e
r
17
y
e
a
rs
of
e
x
p
e
rien
c
e
in
th
e
IT
in
d
u
stry
,
sp
a
n
n
in
g
v
a
rio
u
s
d
o
m
a
in
s
su
c
h
as
b
a
n
k
in
g
,
re
tail,
h
e
a
lt
h
c
a
re
,
a
n
d
c
o
n
s
u
lt
i
n
g
.
W
it
h
m
u
lt
ip
le
G
CP
c
e
rti
f
i
c
a
ti
o
n
s,
as
w
e
ll
as
c
r
e
d
e
n
ti
a
ls
in
A
z
u
re
,
S
n
o
w
f
la
k
e
,
H
a
sh
iCo
rp
,
T
e
ra
d
a
ta,
Clo
u
d
e
ra
,
a
n
d
IT
IL
.
S
h
e
is
an
e
x
p
e
rt
in
c
lo
u
d
c
o
m
p
u
ti
n
g
,
b
i
g
d
a
ta,
m
a
c
h
in
e
lea
rn
in
g
,
a
n
d
se
c
u
rit
y
.
He
h
a
s
su
c
c
e
ss
f
u
ll
y
d
e
li
v
e
re
d
so
lu
ti
o
n
s
f
o
r
c
o
m
p
lex
a
n
d
larg
e
-
sc
a
le
d
a
ta
a
n
a
l
y
ti
c
s,
d
a
ta
e
n
g
in
e
e
rin
g
,
a
n
d
d
a
ta
sc
ien
c
e
p
ro
jec
ts,
lev
e
ra
g
in
g
G
C
P
B
ig
Qu
e
ry
,
V
e
rtex
A
I,
Da
tai
k
u
,
a
n
d
o
t
h
e
r
to
o
ls.
He
is
p
a
ss
io
n
a
te
a
b
o
u
t
h
e
lp
i
n
g
c
li
e
n
ts
tran
sf
o
r
m
th
e
ir
b
u
si
n
e
ss
e
s
w
it
h
d
a
ta
-
d
riv
e
n
i
n
sig
h
ts
a
n
d
in
n
o
v
a
ti
v
e
so
lu
ti
o
n
s.
He
can
be
c
o
n
tac
ted
at
e
m
a
il
:
ra
jes
a
m
p
a
th
ra
jan
@g
m
a
il
.
c
o
m
.
Dr
.
Th
a
y
a
la
g
a
r
a
n
M
e
r
li
n
I
n
b
a
m
a
la
r
se
rv
e
s
as
an
A
ss
o
c
iate
P
r
o
f
e
ss
o
r
in
th
e
De
p
a
rtme
n
t
of
El
e
c
tro
n
ics
a
n
d
In
stru
m
e
n
tatio
n
En
g
in
e
e
rin
g
at
S
a
v
e
e
th
a
En
g
in
e
e
rin
g
Co
ll
e
g
e
,
Ch
e
n
n
a
i,
I
n
d
ia.
S
h
e
e
a
rn
e
d
h
e
r
B.
E.
in
e
lec
tro
n
ics
a
n
d
i
n
stru
m
e
n
tatio
n
e
n
g
in
e
e
rin
g
in
2
0
0
6
f
ro
m
Ka
ru
n
y
a
In
stit
u
te
of
T
e
c
h
n
o
lo
g
y
,
Co
im
b
a
to
re
,
a
ff
il
iate
d
w
it
h
A
n
n
a
Un
iv
e
rsity
,
a
n
d
h
er
M
.
E.
in
a
p
p
l
ied
e
lec
tro
n
ics
in
2
0
0
8
f
ro
m
S
t.
Jo
se
p
h
’s
Co
ll
e
g
e
of
En
g
in
e
e
rin
g
,
Ch
e
n
n
a
i,
a
ls
o
a
ff
il
iate
d
w
it
h
A
n
n
a
Un
iv
e
rsit
y
.
S
h
e
c
o
m
p
lete
d
h
e
r
P
h
.
D.
at
A
n
n
a
Un
iv
e
rsit
y
in
2024.
W
it
h
16
y
e
a
rs
of
te
a
c
h
in
g
e
x
p
e
rien
c
e
,
sh
e
h
a
s
c
o
n
tri
b
u
ted
to
n
u
m
e
ro
u
s
n
a
ti
o
n
a
l
a
n
d
in
ter
n
a
ti
o
n
a
l
jo
u
r
n
a
ls,
as
w
e
ll
as
p
a
ten
ts.
He
r
a
re
a
s
of
e
x
p
e
rti
se
in
c
lu
d
e
ima
g
e
p
ro
c
e
ss
in
g
,
in
stru
m
e
n
tatio
n
,
a
n
d
c
o
n
tro
l
.
S
h
e
can
be
c
o
n
tac
ted
at
e
m
a
il
:
m
e
rli
n
in
b
a
m
a
l
a
r@g
m
a
i
l.
c
o
m
.
Na
g
e
shv
a
r
a
n
Na
n
d
h
i
n
i
h
a
s
done
h
e
r
B.
T
e
c
h
.
(
in
f
o
r
m
a
ti
o
n
tec
h
n
o
l
o
g
y
)
at
Ko
n
g
u
En
g
i
n
e
e
rin
g
Co
ll
e
g
e
,
Ero
d
e
a
n
d
M
.
E
.
(Co
m
p
u
ter
S
c
ien
c
e
a
n
d
E
n
g
in
e
e
rin
g
)
at
P
e
ru
m
a
l
M
a
n
im
e
k
a
lai
Co
ll
e
g
e
of
En
g
in
e
e
rin
g
,
Ho
su
r.
S
h
e
h
a
s
sta
rted
h
e
r
tea
c
h
in
g
c
a
re
e
r
in
th
e
y
e
a
r
2
0
1
5
a
n
d
sh
e
h
a
s
m
o
re
th
a
n
7
y
e
a
rs
of
tea
c
h
in
g
e
x
p
e
rien
c
e
.
S
h
e
h
a
s
o
rg
a
n
ize
d
se
m
in
a
rs,
w
o
rk
sh
o
p
s
f
o
r
th
e
b
e
n
e
f
it
of
stu
d
e
n
ts
a
n
d
g
u
id
e
d
se
v
e
ra
l
stu
d
e
n
ts’
p
ro
jec
ts.
S
h
e
h
a
s
p
u
b
li
s
h
e
d
4
p
a
p
e
rs
in
in
tern
a
ti
o
n
a
l
jo
u
rn
a
l
s
a
n
d
p
re
se
n
ted
5
p
a
p
e
rs
in
v
a
ri
o
u
s
n
a
ti
o
n
a
l
/
in
tern
a
ti
o
n
a
l
c
o
n
f
e
re
n
c
e
s.
S
h
e
h
a
s
a
u
th
o
re
d
3
b
o
o
k
s
a
n
d
b
o
o
k
c
h
a
p
ters
.
S
h
e
is
th
e
m
e
m
b
e
r
of
c
o
m
p
u
ter
so
c
iety
of
In
d
ia.
S
h
e
is
c
u
rre
n
tl
y
s
e
rv
in
g
as
an
As
sista
n
t
P
r
o
f
e
ss
o
r
in
th
e
De
p
a
rtm
e
n
t
of
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
,
P
.
S
.
V.
Co
ll
e
g
e
of
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
,
Krish
n
a
g
iri
,
T
a
m
il
n
a
d
u
,
In
d
ia.
S
h
e
can
be
c
o
n
tac
t
ed
at
e
m
a
il
:
n
a
n
d
h
in
i0
6
5
@g
m
a
il
.
c
o
m
.
S
h
a
n
m
u
g
a
m
S
u
j
a
t
h
a
is
an
a
d
ju
n
c
t
p
r
o
f
e
ss
o
r,
S
a
v
e
e
th
a
S
c
h
o
o
l
of
En
g
in
e
e
rin
g
,
S
a
v
e
e
th
a
In
stit
u
te
of
M
e
d
ica
l
a
n
d
T
e
c
h
n
ica
l
S
c
ien
c
e
s,
Ch
e
n
n
a
i,
T
a
m
il
n
a
d
u
,
In
d
ia.
S
h
e
p
u
b
li
sh
e
d
h
e
r
re
se
a
rc
h
a
rti
c
les
in
m
a
n
y
in
tern
a
ti
o
n
a
l
a
n
d
n
a
ti
o
n
a
l
c
o
n
f
e
re
n
c
e
s
a
n
d
jo
u
rn
a
ls
.
He
r
re
s
e
a
rc
h
a
re
a
s
in
c
lu
d
e
n
e
two
rk
se
c
u
rit
y
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
S
h
e
can
be
c
o
n
tac
ted
at
e
m
a
il
:
su
jath
a
sm
v
r@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.