I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
15
,
No
.
5
,
Octo
b
er
20
25
,
p
p
.
4
7
0
5
~
4
7
1
3
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
5
.
pp
4
7
0
5
-
4
7
1
3
4705
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
An
ef
ficien
t
di
rec
tion o
riente
d bloc
k
-
ba
sed v
ideo
inp
a
inting
using
mo
rpho
lo
g
ica
l opera
tions
a
nd
a
da
ptively
dime
nsio
ned
sea
rch re
g
io
n wi
t
h direc
tion
-
o
rien
t
ed block
-
ba
sed
in
pa
inting
Sh
y
ni Sh
a
j
a
ha
n
1
,
Y.
J
a
co
b Ve
t
ha
Ra
j
2
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
A
d
i
S
h
a
n
k
a
r
a
I
n
st
i
t
u
t
e
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
Te
c
h
n
o
l
o
g
y
,
K
a
l
a
d
y
,
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
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
N
e
s
a
mo
n
y
M
e
m
o
r
i
a
l
C
h
r
i
st
i
a
n
C
o
l
l
e
g
e
,
T
a
mi
l
n
a
d
u
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
l 2
5
,
2
0
2
4
R
ev
is
ed
J
u
l 3
,
2
0
2
5
Acc
ep
ted
J
u
l 1
2
,
2
0
2
5
Vid
e
o
i
n
p
a
i
n
ti
n
g
is
a
tec
h
n
i
q
u
e
i
n
c
o
m
p
u
ter v
isi
o
n
u
se
d
t
o
re
m
o
v
e
u
n
wa
n
ted
o
b
jec
ts
fro
m
v
id
e
o
se
q
u
e
n
c
e
s
wh
il
e
p
re
se
rv
i
n
g
v
isu
a
l
c
o
n
siste
n
c
y
,
so
th
a
t
m
o
d
ifi
c
a
ti
o
n
s
re
m
a
in
u
n
n
o
ti
c
e
a
b
l
e
to
th
e
h
u
m
a
n
e
y
e
.
Th
is
p
a
p
e
r
p
re
se
n
ts
a
n
a
c
c
u
ra
te
v
id
e
o
i
n
p
a
in
ti
n
g
m
o
d
e
l
b
a
se
d
o
n
th
e
a
d
a
p
ti
v
e
ly
d
ime
n
si
o
n
e
d
se
a
rc
h
re
g
io
n
with
d
irec
ti
o
n
-
o
r
ien
ted
b
lo
c
k
-
b
a
se
d
i
n
p
a
i
n
ti
n
g
(
AD
S
R
-
DO
BI
)
a
lg
o
rit
h
m
.
Th
e
m
o
d
e
l
o
p
e
ra
tes
in
fiv
e
m
a
in
p
h
a
se
s:
p
re
p
ro
c
e
ss
in
g
,
b
a
c
k
g
ro
u
n
d
se
p
a
ra
ti
o
n
,
m
o
r
p
h
o
lo
g
ica
l
o
p
e
ra
ti
o
n
s
,
o
b
jec
t
re
m
o
v
a
l,
a
n
d
v
id
e
o
in
p
a
i
n
ti
n
g
.
I
n
it
iall
y
,
th
e
i
n
p
u
t
v
i
d
e
o
is
c
o
n
v
e
rted
in
to
fra
m
e
s,
fo
ll
o
we
d
b
y
p
re
p
ro
c
e
ss
in
g
ste
p
s
su
c
h
a
s
d
e
i
o
n
izin
g
a
n
d
re
siz
in
g
.
Th
e
se
fra
m
e
s
a
re
th
e
n
p
ro
c
e
ss
e
d
u
sin
g
a
b
a
c
k
g
ro
u
n
d
su
b
trac
ti
o
n
m
o
d
u
le,
wh
e
re
o
b
jec
t
l
o
c
a
li
z
a
ti
o
n
a
n
d
fo
re
g
ro
u
n
d
d
e
tec
ti
o
n
a
re
p
e
rfo
rm
e
d
u
sin
g
th
e
b
in
o
m
ially
d
istri
b
u
ted
fo
re
g
ro
u
n
d
se
g
m
e
n
tati
o
n
n
e
tw
o
rk
(BDFg
S
e
g
Ne
t)
a
n
d
m
o
r
p
h
o
l
o
g
ica
l
tec
h
n
iq
u
e
s.
T
h
is
re
su
lt
s
in
se
g
m
e
n
ted
fo
re
g
ro
u
n
d
o
b
jec
ts
trac
k
e
d
a
c
ro
ss
fra
m
e
s.
Th
e
o
b
jec
t
re
m
o
v
a
l
p
h
a
se
e
li
m
in
a
tes
th
e
id
e
n
ti
fied
f
o
re
g
ro
u
n
d
o
b
jec
ts
a
n
d
d
e
fi
n
e
s
th
e
m
issin
g
re
g
i
o
n
s
(
h
o
les
)
to
b
e
fil
led
.
Th
e
AD
S
R
-
DO
BI
a
lg
o
rit
h
m
is
t
h
e
n
a
p
p
li
e
d
t
o
i
n
p
a
in
t
t
h
e
se
re
g
io
n
s
se
a
m
les
sly
.
Ex
p
e
rime
n
tal
re
su
l
ts
d
e
m
o
n
stra
te
th
a
t
t
h
is
a
p
p
ro
a
c
h
o
u
t
p
e
rfo
rm
s
e
x
isti
n
g
sta
te
-
of
-
th
e
-
a
rt
m
e
th
o
d
s in
b
o
th
a
c
c
u
ra
c
y
a
n
d
e
fficie
n
c
y
.
K
ey
w
o
r
d
s
:
B
o
n
o
m
ially
d
is
tr
ib
u
ted
f
o
r
eg
r
o
u
n
d
s
eg
m
en
tatio
n
n
etwo
r
k
Ad
ap
tiv
ely
d
im
e
n
s
io
n
ed
s
ea
r
c
h
r
eg
io
n
with
d
ir
ec
tio
n
o
r
ien
ted
b
lo
ck
b
ased
i
n
p
ain
tin
g
Mo
r
p
h
o
l
o
g
ical
o
p
er
atio
n
s
Su
m
o
f
s
q
u
a
r
ed
d
if
f
er
en
ce
s
Vid
eo
in
p
ain
tin
g
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
:
Sh
y
n
i Sh
ajah
an
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
,
A
d
i Sh
a
n
k
ar
a
I
n
s
titu
te
o
f
E
n
g
in
ee
r
in
g
an
d
T
ec
h
n
o
lo
g
y
Kala
d
y
,
Ker
ala,
I
n
d
ia
E
m
ail: sh
y
n
i.c
s
@
ad
is
h
an
k
ar
a.
ac
.
in
1.
I
NT
RO
D
UCT
I
O
N
I
n
p
ain
tin
g
i
n
v
o
l
v
es
f
illi
n
g
m
is
s
in
g
p
ix
els
in
im
ag
es
with
v
is
u
ally
p
lau
s
ib
le
co
n
ten
t
[
1
]
,
b
u
t
it
is
in
h
er
en
tly
ill
-
p
o
s
ed
with
n
o
u
n
iq
u
e
s
o
lu
tio
n
[
2
]
.
T
h
e
n
ee
d
f
o
r
in
p
ain
tin
g
h
as
in
cr
ea
s
ed
with
th
e
g
r
o
wth
o
f
h
ig
h
-
r
eso
lu
tio
n
m
u
ltime
d
ia
c
o
n
ten
t
[
3
]
.
Vid
eo
in
p
ain
tin
g
ex
ten
d
s
th
is
task
to
tem
p
o
r
al
d
ata,
aim
in
g
to
f
ill
m
is
s
in
g
r
eg
io
n
s
ac
r
o
s
s
f
r
am
es
with
co
h
er
en
t
an
d
r
ea
lis
tic
co
n
ten
t
[
4
]
,
[
5
]
.
C
h
allen
g
es
ar
is
e
f
r
o
m
ca
m
er
a
m
o
tio
n
an
d
co
m
p
le
x
o
b
ject
m
o
v
em
en
ts
in
r
ea
l
-
wo
r
ld
v
id
eo
s
[
6
]
.
Ap
p
l
y
in
g
im
ag
e
i
n
p
a
in
tin
g
m
o
d
els
f
r
am
e
-
by
-
f
r
a
m
e
o
f
ten
lead
s
to
tem
p
o
r
al
in
c
o
n
s
is
ten
cies
lik
e
f
lick
er
in
g
[
7
]
.
T
h
is
n
aiv
e
ap
p
r
o
ac
h
o
v
er
lo
o
k
s
v
id
e
o
d
y
n
am
ics
an
d
f
ails
to
ca
p
tu
r
e
m
o
tio
n
-
d
r
iv
en
a
p
p
ea
r
a
n
ce
ch
an
g
es
o
v
er
tim
e
[
8
]
,
h
ig
h
lig
h
t
in
g
th
e
im
p
o
r
tan
ce
o
f
s
p
atio
tem
p
o
r
al
co
h
e
r
en
ce
i
n
h
ig
h
-
q
u
ality
v
id
eo
in
p
ain
tin
g
.
As
v
id
eo
s
ex
h
ib
it
tem
p
o
r
al
r
eg
u
lar
ity
,
to
in
p
ain
t
a
g
iv
en
f
r
am
e,
it
is
n
atu
r
al
to
u
s
e
d
ata
f
r
o
m
o
th
er
f
r
am
es
as
th
e
d
ata
in
o
th
er
f
r
a
m
es
m
ay
co
r
r
esp
o
n
d
to
p
ar
ts
o
f
th
e
s
ce
n
e
b
eh
in
d
th
e
m
ask
ed
r
eg
io
n
[
9
]
.
B
o
t
h
s
p
atial
s
tr
u
ctu
r
e
an
d
tem
p
o
r
al
co
h
er
e
n
ce
ar
e
r
eq
u
ir
e
d
to
b
e
co
n
s
id
er
ed
in
h
ig
h
-
q
u
ality
v
id
eo
in
p
ain
tin
g
[
1
0
]
.
R
ec
o
v
er
in
g
m
is
s
in
g
v
id
e
o
co
n
ten
t r
eq
u
ir
es th
e
u
n
d
er
s
tan
d
in
g
o
f
n
o
t o
n
l
y
th
e
s
p
atial
co
n
te
x
t o
f
ea
ch
f
r
am
e
b
u
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
5
,
Octo
b
e
r
20
25
:
4
7
0
5
-
4
7
1
3
4706
also
th
e
m
o
tio
n
co
n
tex
t
ac
r
o
s
s
f
r
am
es
[
1
1
]
.
Fo
r
an
y
m
is
s
in
g
p
ix
els
th
at
lack
g
o
o
d
co
r
r
esp
o
n
d
en
ce
d
u
e
to
o
cc
lu
s
io
n
,
th
e
v
i
d
eo
in
p
ain
tin
g
m
eth
o
d
m
u
s
t h
allu
cin
ate
r
ea
s
o
n
ab
le
co
n
ten
t
[
1
2
]
.
T
h
e
s
tat
e
-
of
-
th
e
-
ar
t m
eth
o
d
s
ten
d
to
ca
p
tu
r
e
l
o
n
g
-
ter
m
c
o
r
r
esp
o
n
d
en
c
es
with
an
atten
tio
n
m
ec
h
a
n
is
m
,
s
o
th
e
a
v
ailab
l
e
co
n
ten
t
at
d
is
tan
t
f
r
am
es
ca
n
b
e
g
lo
b
ally
p
r
o
p
ag
ated
to
th
e
u
n
k
n
o
wn
r
eg
i
o
n
s
[
1
3
]
.
T
r
ad
itio
n
al
p
atc
h
-
b
ased
m
eth
o
d
s
f
i
n
d
s
im
ilar
Sp
atio
-
tem
p
o
r
al
p
atch
es
f
r
o
m
th
e
k
n
o
wn
r
eg
i
o
n
s
o
f
v
id
eo
s
t
o
f
ill
th
e
h
o
les,
wh
ich
f
o
r
m
u
la
te
th
e
p
r
o
b
lem
as
a
p
atch
-
b
ased
o
p
tim
izatio
n
task
[
1
4
]
.
T
h
ese
m
eth
o
d
s
r
ely
h
ea
v
ily
o
n
th
e
h
y
p
o
th
esis
th
at
th
e
m
is
s
in
g
co
n
ten
t
i
n
th
e
co
r
r
u
p
te
d
r
eg
io
n
ap
p
ea
r
s
i
n
n
eig
h
b
o
r
in
g
f
r
am
es,
wh
ich
g
r
ea
tly
lim
its
th
eir
g
en
er
aliza
tio
n
ab
ilit
y
[
1
5
]
.
I
n
ad
d
itio
n
,
p
atch
-
b
ased
m
et
h
o
d
s
ass
u
m
e
th
er
e
is
a
r
ef
er
en
ce
f
o
r
th
e
m
is
s
in
g
p
a
r
t
an
d
o
f
ten
f
ail
to
r
ec
o
v
er
n
o
n
-
r
ep
etitiv
e
an
d
c
o
m
p
lex
r
eg
i
o
n
s
(
e.
g
.
th
ey
ca
n
n
o
t
r
ec
o
v
er
a
m
is
s
in
g
f
ac
e
well)
[
1
6
]
.
I
n
r
ec
en
t
y
ea
r
s
,
a
n
u
m
b
e
r
o
f
d
ee
p
lear
n
in
g
-
b
ased
v
id
eo
in
p
ain
tin
g
m
eth
o
d
s
ar
e
p
r
o
p
o
s
ed
[
1
7
]
.
T
h
ese
ex
itin
g
d
ee
p
v
id
e
o
in
p
ain
tin
g
m
eth
o
d
s
ca
n
b
e
s
u
m
m
ar
ized
a
s
two
k
ey
m
o
d
u
les,
a
tem
p
o
r
al
f
ea
tu
r
e
ag
g
r
eg
atio
n
,
an
d
s
in
g
l
e
-
f
r
am
e
i
n
p
ain
tin
g
f
o
r
tem
p
o
r
al
co
n
s
is
ten
cy
[
1
8
]
.
Sid
d
av
atam
et
a
l.
[
1
9
]
p
r
o
p
o
s
ed
a
v
id
eo
in
p
ain
tin
g
m
eth
o
d
u
s
in
g
au
t
o
en
co
d
er
s
th
at
lear
n
s
th
e
b
ac
k
g
r
o
u
n
d
f
ir
s
t,
th
en
o
b
ject
f
ea
tu
r
es,
f
o
llo
wed
b
y
o
b
ject
r
e
m
o
v
al
an
d
b
ac
k
g
r
o
u
n
d
r
ec
o
n
s
tr
u
ctio
n
.
T
h
ey
u
s
e
d
a
p
r
e
-
tr
ain
e
d
YOL
O
m
o
d
el
f
o
r
o
b
ject
d
etec
tio
n
.
Alth
o
u
g
h
th
e
m
eth
o
d
s
h
o
we
d
im
p
r
o
v
ed
p
er
f
o
r
m
a
n
ce
,
it
f
ac
ed
lim
itatio
n
s
r
elate
d
to
d
ee
p
f
ak
e
task
s
.
Ke
et
a
l.
[
2
0
]
in
tr
o
d
u
ce
d
a
n
o
cc
lu
s
io
n
-
a
war
e
v
id
e
o
o
b
ject
in
p
ain
tin
g
ap
p
r
o
ac
h
with
th
e
Yo
u
T
u
b
e
-
VOI
b
e
n
ch
m
a
r
k
f
o
r
r
ea
lis
tic
o
cc
lu
s
io
n
s
.
T
h
eir
v
id
eo
o
b
ject
in
p
ain
tin
g
n
etwo
r
k
(
VOI
N)
u
s
ed
tem
p
o
r
al
GANs
an
d
s
p
atio
-
tem
p
o
r
al
atten
tio
n
f
o
r
s
h
ap
e
c
o
m
p
letio
n
an
d
te
x
tu
r
e
g
en
er
atio
n
.
W
h
ile
ef
f
ec
tiv
e
f
o
r
co
m
p
lex
o
b
jects,
VOI
N’
s
p
e
r
f
o
r
m
a
n
ce
d
eg
r
a
d
ed
with
in
ac
cu
r
ate
in
p
u
t.
Szeto
et
a
l.
[
2
1
]
p
r
o
p
o
s
ed
a
tem
p
o
r
ally
-
awa
r
e
in
ter
p
o
latio
n
n
et
wo
r
k
f
o
r
v
id
e
o
f
r
am
e
i
n
p
ain
tin
g
,
u
s
in
g
a
v
id
eo
p
r
ed
ictio
n
s
u
b
n
etwo
r
k
to
g
en
er
ate
in
ter
m
ed
iate
f
r
a
m
es
an
d
b
len
d
i
n
g
th
em
with
tem
p
o
r
ally
-
awa
r
e
in
ter
p
o
latio
n
(
T
AI
)
.
T
h
eir
m
et
h
o
d
o
u
tp
er
f
o
r
m
e
d
s
tate
-
of
-
th
e
-
ar
t
ap
p
r
o
ac
h
es
b
u
t
p
r
o
d
u
ce
d
b
lu
r
r
y
r
esu
lts
u
n
d
er
h
ea
v
y
ca
m
e
r
a
m
o
tio
n
.
Hu
a
n
g
an
d
L
in
[
2
2
]
in
tr
o
d
u
ce
d
a
v
id
eo
in
p
ain
tin
g
m
eth
o
d
b
ased
o
n
o
b
ject
m
o
tio
n
r
ate
an
d
co
lo
r
v
ar
ia
n
ce
,
u
s
in
g
an
ad
ap
tiv
e
f
o
r
e
g
r
o
u
n
d
m
o
d
el
an
d
ex
em
p
lar
-
b
ased
in
p
ain
tin
g
f
o
r
u
n
p
air
e
d
ar
ea
s
.
W
h
ile
th
eir
ap
p
r
o
ac
h
y
ield
ed
v
is
u
ally
p
leasin
g
r
esu
lts
,
it
s
tr
u
g
g
led
t
o
ac
cu
r
ately
esti
m
ate
m
o
tio
n
r
ates
wh
e
n
m
o
v
in
g
o
b
jects
o
v
e
r
lap
p
ed
.
I
n
p
ain
ted
v
id
e
o
s
h
av
e
b
ec
o
m
e
m
o
r
e
a
n
d
m
o
r
e
d
i
f
f
icu
lt
t
o
b
e
d
is
tin
g
u
is
h
ed
ev
en
b
y
e
y
es
in
p
ac
e
with
th
e
r
e
m
ar
k
ab
le
s
u
cc
ess
in
v
i
d
eo
in
p
ain
tin
g
m
eth
o
d
s
[
2
3
]
.
T
h
e
d
if
f
icu
lty
o
f
v
id
e
o
in
p
ain
tin
g
is
in
h
er
e
n
tly
tied
to
th
e
co
n
ten
t
o
f
th
e
v
id
eo
s
a
n
d
m
ask
s
b
ein
g
in
p
ain
te
d
.
S
o
,
co
n
te
n
t
-
in
f
o
r
m
e
d
d
iag
n
o
s
tic
ev
al
u
atio
n
is
p
er
f
o
r
m
ed
,
wh
ich
id
en
tifie
s
th
e
s
tr
en
g
th
s
an
d
wea
k
n
ess
es
o
f
m
o
d
e
r
n
in
p
ain
tin
g
m
eth
o
d
s
[
2
4
]
.
Mo
s
t
o
f
th
e
ex
i
s
tin
g
tech
n
iq
u
es
d
ev
elo
p
ed
f
o
r
v
id
eo
in
p
ai
n
tin
g
h
av
e
c
o
m
p
l
ex
ities
in
ter
m
s
o
f
co
m
p
u
tatio
n
a
n
d
ac
c
u
r
ac
y
.
Al
th
o
u
g
h
th
er
e
ar
e
s
ev
e
r
al
tech
n
iq
u
es,
th
er
e
is
a
co
n
s
tan
t
d
em
an
d
f
o
r
r
eliab
le
an
d
ef
f
icien
t
v
id
e
o
in
p
ain
tin
g
s
y
s
tem
s
.
T
h
er
ef
o
r
e,
t
h
is
p
ap
er
p
r
o
p
o
s
es
an
ef
f
icie
n
t
d
ir
ec
tio
n
o
r
ien
ted
f
ast
iter
ativ
e
b
lo
ck
-
b
ased
v
id
eo
in
p
ain
tin
g
m
o
d
el
u
s
in
g
ADSR
-
DOBI.
T
h
e
r
est
o
f
th
e
p
ap
er
is
o
r
g
a
n
ized
as
f
o
llo
ws,
s
ec
tio
n
1
s
u
r
v
ey
s
th
e
ex
is
tin
g
wo
r
k
s
r
elate
d
to
v
id
eo
in
p
ain
tin
g
.
S
ec
tio
n
2
e
x
p
lain
s
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
.
T
h
e
ex
p
er
im
en
tal
ev
al
u
atio
n
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
l
o
g
y
is
g
iv
e
n
in
s
ec
tio
n
3
an
d
s
ec
tio
n
4
c
o
n
clu
d
es th
e
p
ap
er
with
f
u
tu
r
e
e
n
h
an
ce
m
en
t.
2.
P
RO
P
O
SE
D
VID
E
O
I
NP
A
I
NT
I
NG
S
YS
T
E
M
I
n
th
is
p
ap
er
,
d
i
r
ec
tio
n
o
r
ien
ted
f
ast
iter
ativ
e
b
lo
c
k
-
b
ased
v
id
eo
in
p
ain
tin
g
u
s
in
g
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
a
n
d
SS
D
is
d
o
n
e
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
f
ir
s
t
d
etec
ts
th
e
f
o
r
e
g
r
o
u
n
d
o
b
ject
th
at
n
ee
d
s
to
b
e
r
em
o
v
ed
an
d
th
e
tar
g
et
r
eg
io
n
to
b
e
in
p
ain
ted
.
T
h
en
th
e
AD
SR
-
DO
B
I
alg
o
r
ith
m
is
u
tili
ze
d
f
o
r
th
e
p
u
r
p
o
s
e
o
f
in
p
ain
tin
g
,
wh
e
r
e
th
e
tar
g
et
r
eg
io
n
is
in
p
ain
ted
with
th
e
e
f
f
icien
t
b
lo
ck
m
atch
i
n
g
m
ec
h
an
is
m
.
T
h
e
b
lo
ck
d
iag
r
am
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
is
s
h
o
wn
in
Fig
u
r
e
1
.
2
.
1
.
P
r
o
ce
s
s
ing
I
n
th
is
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
,
th
e
in
p
u
t
v
id
eo
(
)
is
tak
en
f
r
o
m
p
u
b
licly
av
ailab
le
d
atasets
.
As
p
ar
t
o
f
p
r
e
-
p
r
o
ce
s
s
in
g
,
f
r
am
es
ar
e
ex
tr
ac
te
d
f
r
o
m
th
e
c
ap
tu
r
ed
v
id
e
o
s
f
o
r
th
e
f
u
r
th
er
p
r
o
ce
s
s
an
d
th
e
co
n
v
er
ted
f
r
am
es a
r
e
i
n
itialized
as,
(
)
=
{
(
1
)
,
(
2
)
,
.
.
.
.
.
.
(
)
}
(1
)
wh
er
e
,
(
)
is
th
e
n
u
m
b
e
r
o
f
f
r
am
es.
T
h
e
co
n
v
er
ted
f
r
am
es
ar
e
f
u
r
th
er
p
r
e
-
p
r
o
ce
s
s
ed
b
ased
o
n
th
e
f
o
llo
win
g
s
tep
s
.
a.
R
esize
im
ag
e:
I
n
th
is
s
ec
tio
n
,
th
e
f
r
am
es
(
)
ar
e
r
esized
f
o
r
r
ed
u
cin
g
th
e
co
m
p
u
tatio
n
al
ti
m
e
o
f
th
e
s
y
s
tem
.
I
m
a
g
e
r
e
s
i
z
i
n
g
r
e
f
e
r
s
t
o
t
h
e
s
c
a
l
in
g
o
f
i
m
a
g
e
s
.
I
t
h
e
l
p
s
in
r
ed
u
c
in
g
t
h
e
n
u
m
b
er
o
f
p
i
x
e
l
s
f
r
o
m
a
n
im
a
g
e
an
d
a
l
s
o
z
o
o
m
i
n
g
in
o
n
im
a
g
e
s
.
B
ec
au
s
e
th
e
lar
g
e
im
ag
es
ar
e
f
e
d
in
to
th
e
AI
alg
o
r
ith
m
v
ar
y
in
s
ize
th
er
ef
o
r
e
th
e
tr
ai
n
in
g
m
ig
h
t b
e
in
cr
ea
s
ed
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
efficien
t d
ir
ec
tio
n
o
r
ien
ted
b
lo
ck
-
b
a
s
ed
vid
eo
in
p
a
in
tin
g
u
s
in
g
…
(
S
h
yn
i S
h
a
ja
h
a
n
)
4707
b.
No
is
e
R
em
o
v
al:
I
n
th
is
s
ec
tio
n
,
th
e
n
o
is
es
ar
e
r
em
o
v
ed
d
u
e
to
th
e
p
r
esen
ce
o
f
b
lu
r
an
d
illu
m
in
atio
n
s
in
th
e
im
ag
es.
Hen
ce
th
e
p
r
o
p
o
s
ed
m
eth
o
d
u
s
es
th
e
tech
n
iq
u
e
o
f
Gau
s
s
ian
s
m
o
o
th
en
in
g
in
o
r
d
er
to
en
h
a
n
ce
th
e
im
ag
e
s
tr
u
ctu
r
es
at
a
d
if
f
er
en
t
s
ca
le.
T
h
e
v
is
u
al
ef
f
e
ct
o
f
t
h
is
b
l
u
r
r
i
n
g
t
ec
h
n
i
q
u
e
is
a
s
m
o
o
th
b
lu
r
r
es
em
b
li
n
g
t
h
a
t
o
f
v
ie
wi
n
g
th
e
im
a
g
e
t
h
r
o
u
g
h
a
t
r
a
n
s
l
u
c
en
t
s
cr
ee
n
.
T
h
e
d
eg
r
ee
o
f
s
m
o
o
th
in
g
is
d
eter
m
i
n
ed
b
y
th
e
s
tan
d
a
r
d
d
e
v
iatio
n
o
f
th
e
Gau
s
s
ian
.
Hen
ce
th
e
p
r
e
p
r
o
ce
s
s
ed
f
r
a
m
e
is
in
itialized
as,
(
)
=
{
(
1
)
,
(
2
)
,
.
.
.
.
.
.
(
)
}
(
2
)
w
h
er
e,
(
)
d
en
o
tes th
e
p
r
e
p
r
o
ce
s
s
ed
f
r
am
es.
Fig
u
r
e
1
.
B
lo
ck
d
iag
r
am
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
o
lo
g
y
2
.
2
.
B
a
c
k
g
ro
un
d
s
ub
t
ra
ct
io
n
B
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
is
a
w
id
ely
u
s
ed
m
eth
o
d
f
o
r
d
etec
tin
g
m
o
v
in
g
o
b
jects
in
v
id
eo
s
ca
p
tu
r
ed
b
y
s
tatic
ca
m
er
as.
I
t
h
elp
s
in
s
e
g
m
en
tin
g
f
o
r
eg
r
o
u
n
d
b
lo
b
s
in
to
in
d
iv
id
u
al
o
b
jects
an
d
tr
ac
k
in
g
th
em
ac
r
o
s
s
f
r
am
es.
I
n
th
is
wo
r
k
,
f
o
r
eg
r
o
u
n
d
s
eg
m
en
tatio
n
n
etwo
r
k
(
F
g
Seg
Net)
is
u
s
ed
f
o
r
th
is
p
u
r
p
o
s
e.
Fg
Seg
Net
is
a
r
ec
en
tly
d
e
v
elo
p
e
d
,
h
i
g
h
-
p
e
r
f
o
r
m
in
g
n
e
u
r
al
n
etwo
r
k
th
a
t
em
p
lo
y
s
en
c
o
d
er
-
d
ec
o
d
er
ar
ch
itectu
r
e.
T
h
e
en
co
d
er
,
m
ad
e
u
p
o
f
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NNs),
ex
tr
ac
ts
im
ag
e
f
ea
tu
r
es,
wh
ile
th
e
d
ec
o
d
er
,
u
s
in
g
a
tr
an
s
p
o
s
ed
C
NN
(
T
C
NN)
,
r
ec
o
n
s
tr
u
cts
th
e
f
ea
tu
r
e
m
ap
s
f
o
r
o
b
ject
s
eg
m
en
tatio
n
.
T
h
is
ar
ch
itectu
r
e
en
ab
les
ac
cu
r
ate
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
an
d
o
b
ject
id
en
tifi
ca
tio
n
.
T
o
en
s
u
r
e
s
tab
le
tr
ain
i
n
g
an
d
av
o
id
is
s
u
es
lik
e
v
an
is
h
in
g
o
r
ex
p
lo
d
in
g
g
r
ad
ien
ts
,
th
e
n
etwo
r
k
’
s
weig
h
ts
ar
e
in
itialized
u
s
in
g
a
B
in
o
m
i
al
d
is
tr
ib
u
tio
n
.
T
h
e
ar
ch
itectu
r
e
o
f
Fg
Seg
Net
is
s
h
o
wn
in
Fig
u
r
e
2
.
I
n
iti
all
y
,
t
h
e
i
n
p
u
t
im
ag
e
f
r
a
m
es a
r
e
f
e
d
in
to
t
h
r
e
e
C
NNs
a
n
d
t
h
e
o
u
t
p
u
ts
a
r
e
c
o
n
ca
te
n
a
te
d
a
n
d
a
p
p
l
ie
d
f
o
r
T
C
NN
.
B
ef
o
r
e
s
eg
m
en
tati
o
n
,
th
e
weig
h
ts
o
f
n
etwo
r
k
s
m
u
s
t
b
e
in
itialized
.
Her
e
th
e
weig
h
ts
ar
e
in
itialized
u
s
in
g
th
e
B
in
o
m
ial
d
is
tr
ib
u
ti
o
n
f
u
n
ctio
n
r
ath
e
r
th
a
n
ass
ig
n
in
g
t
h
e
r
a
n
d
o
m
n
u
m
b
e
r
s
.
T
h
u
s
,
th
e
weig
h
ts
ar
e
in
itialized
as,
=
(
)
−
(
3
)
w
h
e
r
e
,
d
en
o
tes
th
e
b
in
o
m
ial
p
r
o
b
ab
ilit
y
,
d
en
o
tes
th
e
p
r
o
b
ab
ilit
y
o
f
s
u
cc
ess
,
p
r
o
b
ab
il
ity
o
f
f
ailu
r
e
,
d
en
o
tes n
u
m
b
er
o
f
tr
ials
,
d
en
o
tes th
e
s
p
ec
if
ic
o
u
tco
m
es
i.e
.
,
weig
h
ts
.
2.
2
.
1
.
E
nc
o
de
r
net
wo
r
k
T
h
e
en
co
d
er
n
etwo
r
k
co
n
s
is
ts
o
f
th
r
ee
co
p
ies
o
f
C
NN,
ea
c
h
o
f
wh
ich
co
n
tain
s
th
e
f
ir
s
t
f
o
u
r
b
lo
ck
s
as
th
at
o
f
th
e
VGG
-
1
6
n
et
an
d
th
e
d
r
o
p
o
u
t
lay
er
s
.
T
h
e
in
p
u
t
im
ag
e
f
r
am
es
ar
e
f
ed
to
ea
ch
C
NN
wh
er
e
th
e
co
n
v
o
l
u
tio
n
lay
er
tr
an
s
f
o
r
m
s
th
e
in
p
u
ts
in
to
th
e
f
ea
tu
r
e
m
a
p
s
o
f
s
ize
×
ℎ
.
T
h
e
tr
an
s
f
o
r
m
ati
o
n
o
f
f
ea
tu
r
e
m
ap
s
ca
n
b
e
e
x
p
r
ess
ed
as,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
5
,
Octo
b
e
r
20
25
:
4
7
0
5
-
4
7
1
3
4708
(
)
=
ℑ
⌊
∗
(
)
+
⌋
(
4
)
ℑ
=
(
,
ℎ
)
ℎ
,
ℎ
=
∗
(
)
+
(
5
)
w
h
er
e,
(
)
d
en
o
tes
th
e
o
u
tp
u
t
f
ea
tu
r
e
m
ap
s
o
f
co
n
v
o
lu
tio
n
,
ℑ
d
en
o
tes
th
e
R
e
LU
ac
tiv
atio
n
f
u
n
ctio
n
,
is
th
e
weig
h
t
v
alu
es,
d
e
n
o
tes
th
e
b
ias
o
f
th
e
n
etwo
r
k
.
T
h
u
s
,
th
e
e
x
tr
ac
ted
f
ea
tu
r
es
m
a
p
s
a
r
e
d
o
w
n
-
s
a
m
p
l
ed
b
y
u
s
i
n
g
t
h
e
p
o
o
li
n
g
l
a
ye
r
.
Th
e
p
o
o
l
in
g
l
a
y
er
u
t
iliz
es
th
e
m
a
x
p
o
o
li
n
g
o
p
er
a
ti
o
n
t
h
a
t
s
t
o
r
es
o
n
l
y
t
h
e
ma
x
-
p
o
o
li
n
g
i
n
d
ic
es
i
.
e
.
th
e
l
o
c
ati
o
n
s
o
f
m
a
x
i
m
u
m
f
e
at
u
r
e
v
al
u
e
i
n
e
ac
h
p
o
o
li
n
g
wi
n
d
o
w
to
c
a
p
t
u
r
e
a
n
d
s
to
r
e
b
o
u
n
d
a
r
y
i
n
f
o
r
m
a
ti
o
n
.
T
h
e
o
u
tp
u
t
o
f
t
h
e
p
o
o
l
in
g
la
y
er
i
s
c
o
m
p
u
t
ed
as
,
(
)
=
(
)
−
+
1
(
6
)
w
h
e
r
e
,
(
)
d
en
o
tes
th
e
o
u
tp
u
t
f
ea
tu
r
e
m
ap
s
o
f
th
e
p
o
o
lin
g
lay
e
r
,
d
en
o
tes
th
e
s
tr
id
es
o
f
th
e
k
er
n
el.
At
th
e
e
n
d
o
f
t
h
is
p
r
o
ce
s
s
,
t
h
e
d
r
o
p
o
u
t l
a
y
e
r
is
u
ti
liz
ed
t
o
a
v
o
i
d
t
h
e
p
r
o
b
le
m
o
f
o
v
e
r
f
it
ti
n
g
.
Fig
u
r
e
2
.
Ar
c
h
itectu
r
e
o
f
Fg
Seg
Net
2
.
2
.
2
.
De
co
d
er
ne
t
w
o
r
k
T
h
e
o
u
tp
u
t
o
f
th
e
en
c
o
d
er
n
et
wo
r
k
is
co
n
ca
t
en
at
ed
t
o
f
o
r
m
th
e
f
ea
t
u
r
e
m
ap
s
o
f
d
i
f
f
e
r
e
n
t
s
ca
l
es
a
n
d
th
e
n
t
h
e
m
a
p
s
a
r
e
f
e
d
t
o
T
C
N
N
f
o
r
d
e
co
d
i
n
g
t
h
e
f
e
at
u
r
e
m
ap
s
.
He
r
e
th
e
f
ea
t
u
r
e
m
a
p
s
a
r
e
o
p
e
r
a
te
d
wit
h
t
h
e
tr
a
n
s
p
o
s
e
d
c
o
n
v
o
l
u
t
io
n
t
o
e
n
l
a
r
g
e
t
h
e
f
e
at
u
r
e
m
ap
s
.
Fin
ally
,
a
s
ig
m
o
id
f
u
n
ctio
n
is
u
s
ed
in
th
e
last
lay
er
as
th
e
p
r
o
b
a
b
ilit
y
v
alu
es
f
o
r
ea
c
h
p
ix
el
to
o
b
tain
d
is
cr
ete
b
i
n
ar
y
class
lab
els
f
o
r
eg
r
o
u
n
d
an
d
b
ac
k
g
r
o
u
n
d
.
T
h
e
s
eg
m
en
ted
im
ag
e
ca
n
b
e
o
b
tai
n
ed
as,
(
)
=
1
1
+
(
)
(
7
)
T
h
e
cr
o
s
s
-
en
tr
o
p
y
is
u
tili
ze
d
a
s
a
lo
s
s
f
u
n
ctio
n
,
wh
ich
is
ex
p
r
ess
ed
as,
=
1
∑
−
(
(
)
(
)
=
1
+
(
1
−
(
)
)
(
1
−
(
)
)
)
(
8
)
wh
er
e
,
d
en
o
tes
th
e
lo
s
s
f
u
n
ct
io
n
,
d
en
o
tes
th
e
n
u
m
b
er
o
f
p
ix
els
in
th
e
f
r
am
e,
(
)
is
th
e
g
r
o
u
n
d
tr
u
t
h
lab
el,
(
)
d
en
o
tes
th
e
s
eg
m
en
ted
lab
els
b
y
th
e
n
etwo
r
k
.
T
h
e
s
eg
m
en
ted
f
r
am
es
ar
e
f
u
r
th
er
e
n
h
an
ce
d
to
r
em
o
v
e
th
e
im
p
er
f
ec
ti
o
n
s
in
s
eg
m
en
tatio
n
u
s
in
g
m
o
r
p
h
o
lo
g
i
ca
l o
p
er
atio
n
s
.
2
.
2
.
3
.
O
bje
ct
re
m
o
v
a
l
T
h
e
s
eg
m
en
te
d
o
b
jects
o
r
t
ar
g
et
o
b
jects
s
p
ec
ified
b
y
th
e
m
o
r
p
h
o
lo
g
ical
p
r
o
ce
s
s
(
)
ar
e
r
em
o
v
ed
f
r
o
m
ea
ch
im
ag
e
b
y
f
o
r
m
in
g
th
e
h
o
le
r
eg
io
n
t
o
b
e
in
p
ain
ted
.
T
h
e
s
tatic
p
o
r
tio
n
o
f
th
e
h
o
le
ca
n
b
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
efficien
t d
ir
ec
tio
n
o
r
ien
ted
b
lo
ck
-
b
a
s
ed
vid
eo
in
p
a
in
tin
g
u
s
in
g
…
(
S
h
yn
i S
h
a
ja
h
a
n
)
4709
fill
ed
b
y
av
ailab
le
b
ac
k
g
r
o
u
n
d
in
f
o
r
m
atio
n
u
s
in
g
th
e
v
i
d
eo
i
n
p
ain
tin
g
alg
o
r
ith
m
.
Oth
e
r
wis
e,
im
ag
e
in
p
ain
tin
g
is
p
er
f
o
r
m
e
d
b
ased
o
n
th
e
s
u
r
r
o
u
n
d
in
g
im
ag
e
s
tatis
tic
s
.
Hen
ce
,
th
e
f
r
am
e
with
th
e
h
o
le
r
eg
i
o
n
is
in
itialized
as,
ℎ
(
)
=
⌊
ℎ
(
1
)
,
ℎ
(
2
)
,
ℎ
(
3
)
,
…
.
,
ℎ
(
)
⌋
(
9
)
w
h
er
e
ℎ
(
)
d
en
o
tes th
e
f
r
am
es wit
h
f
o
r
m
ed
h
o
le
r
eg
io
n
,
ℎ
(
)
d
en
o
tes
th
e
ℎ
f
r
am
e.
2
.
2
.
4
.
Video
inp
a
inting
T
h
is
s
ec
tio
n
ex
p
lain
s
th
e
v
id
eo
in
p
ain
tin
g
p
r
o
ce
s
s
u
s
i
n
g
th
e
d
ir
ec
tio
n
-
o
r
ien
ted
b
l
o
ck
-
b
ase
d
in
p
ain
tin
g
(
DOBI)
alg
o
r
ith
m
.
T
h
e
tar
g
et
r
eg
io
n
,
id
en
tifi
ed
as
th
e
h
o
le
a
r
ea
,
is
f
illed
with
m
atch
in
g
b
ac
k
g
r
o
u
n
d
co
n
ten
t.
I
n
itially
,
th
e
b
o
u
n
d
ar
y
p
o
in
ts
o
f
th
e
ta
r
g
et
r
eg
i
o
n
a
r
e
d
eter
m
in
ed
to
s
ea
r
ch
f
o
r
s
u
itab
le
p
atch
es.
W
h
ile
u
s
in
g
a
f
ix
e
d
p
atch
s
ize
in
cr
ea
s
es
s
ea
r
ch
p
o
in
ts
,
ad
ap
tin
g
th
e
s
ea
r
ch
r
e
g
io
n
s
ize
im
p
r
o
v
es
ef
f
icien
cy
with
o
u
t
s
ac
r
if
icin
g
q
u
ality
.
W
h
en
m
o
tio
n
v
ar
ies
am
o
n
g
a
d
jace
n
t
b
l
o
ck
s
,
a
la
r
g
er
s
ea
r
c
h
ar
ea
is
n
ee
d
ed
.
T
h
u
s
,
a
n
ad
a
p
tiv
ely
d
im
en
s
io
n
ed
s
ea
r
ch
r
e
g
io
n
is
u
s
ed
,
lead
in
g
to
th
e
p
r
o
p
o
s
ed
a
d
ap
tiv
ely
d
im
en
s
io
n
ed
s
ea
r
ch
r
eg
io
n
-
b
a
s
ed
DOBI
(
ADS
R
-
DOBI)
m
et
h
o
d
,
as sh
o
wn
in
Fig
u
r
e
3
.
Fig
u
r
e
3
.
Fra
m
es with
tar
g
et
a
n
d
s
o
u
r
ce
r
eg
io
n
f
o
r
in
p
ain
tin
g
I
n
itially
,
th
e
tar
g
et
r
eg
io
n
(
)
h
as
b
ee
n
s
elec
ted
an
d
th
e
n
ea
r
e
s
t
b
o
u
n
d
ar
y
p
o
in
ts
wer
e
d
ete
cted
.
Fro
m
th
e
o
u
ts
id
e
o
f
th
e
d
etec
ted
b
o
u
n
d
ar
y
,
th
e
b
lo
ck
wit
h
th
e
h
ig
h
est
m
atch
in
g
p
r
o
b
ab
ilit
y
is
s
elec
ted
.
Du
r
in
g
th
is
p
r
o
ce
s
s
,
th
e
d
im
en
s
io
n
o
f
th
e
s
ea
r
c
h
r
eg
io
n
is
ch
o
s
en
ad
ap
tiv
ely
as,
=
{
,
}
(
1
0
)
=
{
,
(
|
−
1
|
,
|
−
2
|
,
|
−
3
|
)
}
(
1
1
)
=
{
,
(
|
−
1
|
,
|
−
2
|
,
|
−
3
|
)
}
(
1
2
)
w
h
er
e,
(
)
d
en
o
tes
th
e
d
im
en
s
io
n
o
f
s
ea
r
ch
r
eg
i
o
n
,
(
)
,
(
)
ar
e
th
e
ad
ap
tiv
e
d
is
p
lace
m
en
t
in
th
e
h
o
r
izo
n
tal
(
)
an
d
v
er
tical
(
)
d
ir
ec
tio
n
,
(
,
)
d
en
o
tes
t
h
e
n
u
m
b
er
o
f
s
ea
r
ch
p
o
in
ts
.
T
h
e
a
d
ap
tiv
e
SR
is
b
o
u
n
d
ed
s
u
ch
th
at,
≤
,
≤
(
1
3
)
Mo
r
eo
v
er
,
in
o
r
d
er
to
d
eter
m
in
e
th
e
p
ix
el,
th
e
m
o
s
t
ac
cu
r
ate
to
b
e
r
ep
air
ed
th
e
co
n
f
id
e
n
ce
o
f
th
e
r
ep
air
e
d
p
ix
els
n
ee
d
s
t
o
b
e
u
p
d
ated
.
A
n
in
cr
ea
s
e
in
b
o
th
th
e
co
n
f
i
d
e
n
ce
o
f
th
e
n
eig
h
b
o
r
in
g
p
ix
els
an
d
th
e
p
r
io
r
ity
o
f
th
e
n
eig
h
b
o
r
i
n
g
p
ix
els
co
n
s
titu
tes
th
e
m
o
s
t
ac
cu
r
ate
p
ix
el
t
o
b
e
r
ep
air
e
d
s
o
th
at
t
h
e
im
p
r
o
v
ed
o
u
tp
u
t
ca
n
b
e
r
etr
iev
ed
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
5
,
Octo
b
e
r
20
25
:
4
7
0
5
-
4
7
1
3
4710
T
h
er
ef
o
r
e,
th
e
p
r
io
r
ity
o
f
a
g
iv
en
b
lo
ck
ca
n
b
e
d
ef
in
e
d
as,
=
.
(
1
4
)
=
∑
∈
∪
|
|
,
0
≤
≤
1
(
1
5
)
=
|
.
|
,
0
≤
≤
1
(
1
6
)
w
h
er
e,
d
en
o
tes
th
e
p
r
io
r
ity
,
,
ar
e
th
e
co
n
f
id
en
ce
ter
m
a
n
d
d
ata
ter
m
,
is
th
e
u
n
it
v
ec
to
r
o
r
th
o
g
o
n
al
to
im
ag
e
g
r
ad
ie
n
t,
is
th
e
u
n
it
v
ec
to
r
o
r
th
o
g
o
n
al
to
th
e
p
o
i
n
t
,
is
th
e
n
o
r
m
aliza
tio
n
v
ec
t
o
r
.
T
h
u
s
,
th
e
p
ix
el
with
th
e
h
ig
h
est p
r
io
r
ity
is
tr
e
ated
as th
e
in
itial
s
ea
r
ch
ce
n
tr
e
to
ch
o
o
s
e
th
e
tar
g
et
p
atch
to
b
e
f
illed
.
T
h
en
th
e
s
elec
tio
n
o
f
th
e
b
est
m
atch
i
n
g
b
lo
ck
(
)
is
d
o
n
e
b
ased
o
n
th
e
s
u
m
o
f
s
q
u
a
r
ed
d
if
f
er
en
ce
s
(
SS
D)
ca
lcu
lated
b
etwe
en
th
e
k
n
o
wn
p
ix
els
o
f
t
h
e
tar
g
et
r
eg
io
n
an
d
s
ea
r
ch
r
e
g
io
n
.
Fro
m
t
h
is
s
tep
,
th
e
ar
ea
h
as
b
ee
n
id
en
tifie
d
th
at
s
atis
f
ies th
e
f
o
llo
win
g
cr
iter
io
n
as,
=
∈
(
,
)
(
1
7
)
w
h
er
e,
(
•
)
is
th
e
SS
D
b
etwe
en
t
h
e
b
lo
ck
(
)
an
d
th
e
b
lo
ck
(
)
.
Hen
ce
th
e
co
r
r
esp
o
n
d
in
g
p
o
s
itio
n
to
th
e
u
n
k
n
o
wn
p
ix
el
o
f
th
e
ta
r
g
et
r
eg
io
n
is
f
illed
b
y
ass
ig
n
in
g
th
e
k
n
o
wn
p
ix
els
o
f
th
e
m
atch
in
g
b
l
o
ck
o
b
tain
ed
.
Af
te
r
f
illi
n
g
th
e
co
n
f
id
en
ce
v
alu
e
is
u
p
d
ated
as,
=
,
∀
∈
∩
(
1
8
)
T
h
e
s
tep
s
ar
e
r
ep
ea
ted
u
n
til
t
h
e
tar
g
et
r
eg
io
n
g
ets
f
illed
.
T
h
e
p
s
eu
d
o
-
co
d
e
o
f
th
e
p
r
o
p
o
s
ed
ADSR
-
DOBI
is
s
h
o
wn
in
b
elo
w
Fig
u
r
e
4
.
Fig
u
r
e
4
.
Ps
eu
d
o
co
d
e
o
f
th
e
ADSR
-
DO
B
I
alg
o
r
ith
m
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
I
n
th
is
s
ec
tio
n
,
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
v
id
eo
in
p
ai
n
tin
g
m
eth
o
d
is
an
aly
ze
d
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
l
o
g
y
is
im
p
lem
e
n
ted
i
n
th
e
wo
r
k
in
g
p
latf
o
r
m
o
f
PY
T
HON.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
efficien
t d
ir
ec
tio
n
o
r
ien
ted
b
lo
ck
-
b
a
s
ed
vid
eo
in
p
a
in
tin
g
u
s
in
g
…
(
S
h
yn
i S
h
a
ja
h
a
n
)
4711
3
.
1
.
Da
t
a
ba
s
e
des
cr
iptio
n
Fo
r
th
e
p
er
f
o
r
m
an
ce
an
aly
s
i
s
,
th
e
p
r
o
p
o
s
ed
wo
r
k
u
s
es
th
e
Yo
u
T
u
b
e
-
v
id
eo
o
b
ject
s
eg
m
en
tatio
n
(
VOS)
d
ataset
th
at
is
p
u
b
lica
lly
av
ailab
le
o
n
th
e
in
ter
n
et.
Yo
u
T
u
b
e
-
VOS
co
n
tain
s
4
,
4
5
3
v
id
eo
s
.
Fro
m
th
e
d
ataset,
8
0
%
o
f
d
ata
was
u
s
e
d
f
o
r
tr
ain
in
g
an
d
2
0
%
d
ata
f
o
r
test
in
g
.
T
h
e
co
llected
d
at
aset
h
as
m
o
r
e
th
a
n
7
,
8
0
0
u
n
iq
u
e
o
b
jects,
1
9
0
k
h
i
g
h
-
q
u
ality
m
an
u
al
a
n
n
o
tatio
n
s
,
an
d
m
o
r
e
th
a
n
3
4
0
m
in
u
tes in
d
u
r
atio
n
.
3
.
2
.
P
er
f
o
r
m
a
nce
a
na
ly
s
is
T
h
e
p
r
o
p
o
s
ed
B
DFg
Seg
Net
m
eth
o
d
is
b
ased
o
n
q
u
ality
m
etr
ics
s
u
ch
as
s
en
s
it
iv
ity
,
s
p
ec
if
icity
,
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F
-
m
ea
s
u
r
e,
f
alse
p
o
s
itiv
e
r
ate
(
FP
R
)
,
f
alse
n
eg
ativ
e
r
ate
(
F
NR
)
,
an
d
Ma
tth
ews
co
r
r
elatio
n
co
ef
f
icien
t
(
MCC
)
.
T
ab
le
1
s
h
o
ws
th
e
co
m
p
ar
ativ
e
an
aly
s
is
o
f
th
e
p
r
o
p
o
s
ed
ADSR
-
DO
B
I
alg
o
r
ith
m
a
n
d
t
h
e
ex
is
tin
g
D
OS
-
b
ased
alg
o
r
ith
m
.
T
h
e
an
a
ly
s
is
h
as
b
ee
n
d
o
n
e
b
y
v
ar
y
i
n
g
th
e
s
ize
o
f
t
h
e
f
r
am
e.
Fo
r
a
2
8
5
4
1
-
p
ix
el
f
r
a
m
e,
th
e
tim
e
tak
en
f
o
r
in
p
ain
t
in
g
is
0
.
0
1
8
s
ec
o
n
d
s
less
er
th
an
th
e
ex
is
tin
g
DOS
-
b
ased
m
eth
o
d
.
Similar
ly
,
f
o
r
a
5
4
3
0
0
-
p
i
x
el
f
r
am
e,
th
e
tim
e
t
ak
en
f
o
r
in
p
ain
tin
g
is
1
.
2
2
6
7
s
ec
o
n
d
s
less
er
wh
en
co
m
p
ar
ed
with
th
e
DOS
alg
o
r
ith
m
.
L
ess
er
tim
e,
tak
en
f
o
r
i
n
p
ain
tin
g
th
e
g
i
v
en
a
r
ea
,
s
h
o
ws
th
e
ef
f
icien
cy
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
T
ab
le
2
p
r
esen
ts
a
p
er
f
o
r
m
a
n
ce
co
m
p
a
r
is
o
n
o
f
th
e
p
r
o
p
o
s
ed
ADSR
-
DO
B
I
m
eth
o
d
with
ex
is
tin
g
tech
n
iq
u
es
u
s
in
g
q
u
ality
m
etr
i
cs
s
u
ch
as
PS
NR
,
SS
I
M,
MSE
,
an
d
R
MSE
.
A
h
ig
h
er
PS
NR
an
d
SS
I
M
in
d
icate
b
etter
im
ag
e
q
u
ality
,
wh
ile
lo
wer
MSE
an
d
R
MSE
r
ef
lect
r
ed
u
ce
d
er
r
o
r
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
s
h
o
ws
a
0
.
8
7
d
B
im
p
r
o
v
em
e
n
t
in
PS
NR
o
v
er
FF
B
MA
an
d
a
0
.
0
7
in
cr
ea
s
e
in
SS
I
M
co
m
p
ar
ed
to
B
B
GDS.
Ad
d
itio
n
ally
,
MSE
an
d
R
MSE
v
alu
es
ar
e
co
n
s
is
ten
tly
lo
wer
th
an
th
o
s
e
o
f
ex
is
tin
g
m
eth
o
d
s
.
Ov
e
r
all,
ADSR
-
DO
B
I
o
u
tp
er
f
o
r
m
s
o
th
e
r
tech
n
iq
u
es
ac
r
o
s
s
all
m
etr
ics,
d
em
o
n
s
tr
atin
g
its
ef
f
ec
tiv
en
ess
in
v
id
e
o
in
p
ain
tin
g
.
T
ab
le
3
p
r
esen
ts
a
p
er
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
b
etwe
en
th
e
p
r
o
p
o
s
ed
B
DFg
Seg
Net
an
d
ex
is
tin
g
m
eth
o
d
s
u
s
in
g
v
ar
i
o
u
s
q
u
ality
m
etr
ics.
Hig
h
er
v
alu
es
o
f
ac
cu
r
ac
y
,
s
p
ec
i
f
icity
,
s
en
s
itiv
ity
,
p
r
ec
is
io
n
,
F
-
m
ea
s
u
r
e,
an
d
MCC
in
d
icate
b
etter
p
er
f
o
r
m
a
n
ce
,
wh
il
e
lo
wer
FP
R
an
d
FN
R
v
al
u
es
r
ef
lect
g
r
ea
ter
ef
f
icien
cy
.
As
o
b
s
er
v
e
d
,
ex
is
tin
g
m
eth
o
d
s
s
u
ch
as
C
NN,
ANN,
DNN,
an
d
Seg
Net
s
h
o
w
r
elativ
ely
lo
w
p
er
f
o
r
m
an
ce
ac
r
o
s
s
th
ese
m
et
r
ics.
I
n
c
o
n
tr
ast,
th
e
p
r
o
p
o
s
ed
B
DFg
Seg
Net
d
em
o
n
s
tr
ates
s
u
p
er
io
r
r
esu
lts
in
all
ev
alu
ated
m
etr
ics,
h
ig
h
lig
h
tin
g
its
ef
f
ec
tiv
en
ess
with
o
p
tim
a
l f
ea
tu
r
e
s
elec
tio
n
.
T
ab
le
1
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
o
f
p
r
o
p
o
s
ed
ADSR
-
DOBI
alg
o
r
ith
m
with
th
e
DOS
-
b
ased
alg
o
r
ith
m
To
t
a
l
f
r
a
me
si
z
e
(
i
n
p
i
x
e
l
s)
R
e
m
o
v
e
d
a
r
e
a
s
i
z
e
(
i
n
p
i
x
e
l
s)
Ti
me
t
a
k
e
n
i
n
s
e
c
o
n
d
s
DOS
-
b
a
se
d
a
l
g
o
r
i
t
h
m
P
r
o
p
o
se
d
A
D
S
R
-
D
O
B
I
2
8
5
4
1
2
9
6
0
.
1
5
0
.
1
3
2
5
4
3
0
0
3
4
8
0
1
.
8
5
6
0
.
6
2
9
3
3
7
0
0
0
5
2
1
0
2
.
5
7
0
6
0
.
5
6
7
4
9
2
9
8
2
6
0
1
5
2
.
7
5
2
2
0
.
9
3
4
4
1
9
5
8
6
0
2
5
8
0
6
1
0
.
2
9
3
2
3
.
4
3
7
2
T
ab
le
2
.
Per
f
o
r
m
an
ce
an
aly
s
is
o
f
p
r
o
p
o
s
ed
ADSR
-
DOBI
m
e
th
o
d
b
ased
o
n
d
if
f
er
e
n
t im
ag
e
q
u
ality
m
etr
ics
Te
c
h
n
i
q
u
e
s
P
S
N
R
S
S
I
M
M
S
E
R
M
S
E
DOS
1
4
.
1
9
0
.
7
5
2
1
.
9
5
4
.
6
9
B
B
G
D
S
1
7
.
1
6
0
.
7
8
1
7
.
2
9
4
.
1
6
F
F
B
M
A
2
1
.
3
2
0
.
8
1
5
.
8
1
3
.
9
7
P
r
o
p
o
se
d
A
D
S
R
-
D
O
B
I
2
2
.
1
9
0
.
8
5
1
4
.
2
2
3
.
7
7
T
ab
le
3
.
Per
f
o
r
m
an
ce
an
aly
s
is
o
f
p
r
o
p
o
s
ed
B
DFg
Seg
Net
m
eth
o
d
b
ased
o
n
q
u
ality
m
etr
ics
P
e
r
f
o
r
ma
n
c
e
me
t
r
i
c
s/
t
e
c
h
n
i
q
u
e
s
S
e
g
N
et
C
N
N
ANN
DNN
P
r
o
p
o
se
d
B
D
F
g
S
e
g
N
e
t
A
c
c
u
r
a
c
y
9
2
.
2
3
9
1
.
3
1
9
4
.
3
4
9
2
.
2
6
9
7
.
1
4
S
p
e
c
i
f
i
c
i
t
y
8
6
.
5
6
8
7
.
2
7
8
5
.
9
2
9
2
.
3
9
9
3
.
6
5
S
e
n
s
i
t
i
v
i
t
y
8
9
.
7
2
9
0
.
2
4
8
5
.
2
4
9
0
.
6
2
9
5
.
1
7
P
r
e
c
i
s
i
o
n
9
2
.
5
6
9
0
.
3
1
8
7
.
3
4
9
4
.
8
3
9
7
.
2
9
F
-
M
e
a
s
u
r
e
s
9
3
.
6
7
9
2
.
9
8
8
8
.
2
9
9
3
.
6
2
9
7
.
4
8
FPR
4
6
.
1
4
4
4
.
1
5
5
9
.
8
2
3
0
.
5
1
1
1
.
1
6
F
N
R
4
2
.
3
2
4
9
.
4
2
5
5
.
6
8
3
7
.
5
8
8
.
9
8
M
C
C
8
5
.
4
2
8
2
.
3
1
8
0
.
9
2
8
7
.
2
8
9
4
.
5
3
4.
CO
NCLU
SI
O
N
Vid
eo
in
p
ain
tin
g
r
em
o
v
es
o
r
r
esto
r
es
m
is
s
in
g
r
eg
io
n
s
in
a
v
id
eo
u
s
in
g
s
p
atial
an
d
tem
p
o
r
a
l
in
f
o
r
m
atio
n
.
T
h
is
p
ap
er
p
r
o
p
o
s
es
an
ef
f
ici
en
t
ADSR
-
DO
B
I
b
lo
ck
m
atc
h
in
g
alg
o
r
ith
m
f
o
r
in
p
ain
tin
g
,
wh
er
e
th
e
ta
r
g
et
r
eg
io
n
is
id
en
tifie
d
an
d
f
illed
with
m
atch
in
g
b
ac
k
g
r
o
u
n
d
c
o
n
ten
t.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
i
s
ev
alu
ated
ag
ain
s
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
5
,
Octo
b
e
r
20
25
:
4
7
0
5
-
4
7
1
3
4712
ex
is
tin
g
tech
n
iq
u
es
u
s
in
g
q
u
a
lity
m
etr
ics.
R
esu
lts
s
h
o
w
th
at
ADSR
-
DO
B
I
ac
h
iev
es
s
u
p
er
io
r
p
er
f
o
r
m
a
n
ce
,
with
a
PS
NR
v
alu
e
o
f
2
2
.
1
9
an
d
an
in
p
ain
tin
g
tim
e
o
f
3
.
4
3
s
ec
o
n
d
s
—
b
o
th
b
etter
th
a
n
ex
is
tin
g
m
et
h
o
d
s
.
T
h
ese
f
in
d
in
g
s
d
em
o
n
s
tr
ate
th
e
ef
f
icien
cy
o
f
ADSR
-
DOBI.
Fu
tu
r
e
wo
r
k
ca
n
im
p
r
o
v
e
th
e
m
eth
o
d
f
u
r
th
e
r
b
y
ad
d
r
ess
in
g
o
cc
lu
d
ed
r
e
g
io
n
s
a
n
d
en
h
an
cin
g
m
o
tio
n
h
an
d
lin
g
ca
p
ab
ilit
ies.
RE
F
E
R
E
NC
E
S
[
1
]
S
.
Le
e
,
S
.
W
.
O
h
,
D
.
W
o
n
,
a
n
d
S
.
J
.
K
i
m
,
“
C
o
p
y
-
a
n
d
-
p
a
s
t
e
n
e
t
w
o
r
k
s
f
o
r
d
e
e
p
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
a
rX
i
v
:
1
9
0
8
.
1
1
5
8
7
,
A
u
g
.
2
0
1
9
,
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
1
9
0
8
.
1
1
5
8
7
.
[
2
]
H
.
Z
h
a
n
g
,
L
.
M
a
i
,
H
.
Ji
n
,
Z.
W
a
n
g
,
N
.
X
u
,
a
n
d
J.
C
o
l
l
o
m
o
ss
e
,
“
A
n
i
n
t
e
r
n
a
l
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
t
o
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
i
n
2
0
1
9
I
E
EE/
C
VF
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
V
i
s
i
o
n
(
I
C
C
V)
,
O
c
t
.
2
0
1
9
,
p
p
.
2
7
2
0
–
2
7
2
9
,
d
o
i
:
1
0
.
1
1
0
9
/
i
c
c
v
.
2
0
1
9
.
0
0
2
8
1
.
[
3
]
H
.
O
u
y
a
n
g
,
T.
W
a
n
g
,
a
n
d
Q
.
C
h
e
n
,
“
I
n
t
e
r
n
a
l
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
b
y
i
mp
l
i
c
i
t
l
o
n
g
-
r
a
n
g
e
p
r
o
p
a
g
a
t
i
o
n
,
”
i
n
2
0
2
1
I
EEE
/
C
V
F
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
C
o
m
p
u
t
e
r V
i
s
i
o
n
(
I
C
C
V)
,
O
c
t
.
2
0
2
1
,
p
p
.
1
4
5
5
9
–
1
4
5
6
8
,
d
o
i
:
1
0
.
1
1
0
9
/
i
c
c
v
4
8
9
2
2
.
2
0
2
1
.
0
1
4
3
1
.
[
4
]
R
.
L
i
u
e
t
a
l
.
,
“
F
u
s
e
F
o
r
mer
:
f
u
s
i
n
g
f
i
n
e
-
g
r
a
i
n
e
d
i
n
f
o
r
mat
i
o
n
i
n
t
r
a
n
s
f
o
r
mers
f
o
r
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
i
n
2
0
2
1
I
EEE
/
C
V
F
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
C
o
m
p
u
t
e
r V
i
s
i
o
n
(
I
C
C
V)
,
O
c
t
.
2
0
2
1
,
p
p
.
1
4
0
2
0
–
1
4
0
2
9
,
d
o
i
:
1
0
.
1
1
0
9
/
i
c
c
v
4
8
9
2
2
.
2
0
2
1
.
0
1
3
7
8
.
[
5
]
G
.
Tu
d
a
v
e
k
a
r
,
S
.
R
.
P
a
t
i
l
,
a
n
d
S
.
S
.
S
a
r
a
f
,
“
D
u
a
l
‐
t
r
e
e
c
o
m
p
l
e
x
w
a
v
e
l
e
t
t
r
a
n
sf
o
r
m
a
n
d
s
u
p
e
r
‐
r
e
so
l
u
t
i
o
n
b
a
se
d
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
a
p
p
l
i
c
a
t
i
o
n
t
o
o
b
j
e
c
t
r
e
mo
v
a
l
a
n
d
e
r
r
o
r
c
o
n
c
e
a
l
me
n
t
,
”
C
A
AI
T
ra
n
sa
c
t
i
o
n
s
o
n
I
n
t
e
l
l
i
g
e
n
c
e
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
5
,
n
o
.
4
,
p
p
.
3
1
4
–
3
1
9
,
N
o
v
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
4
9
/
t
r
i
t
.
2
0
1
9
.
0
0
4
5
.
[
6
]
R
.
X
u
,
X
.
Li
,
B
.
Zh
o
u
,
a
n
d
C
.
C
.
L
o
y
,
“
D
e
e
p
f
l
o
w
-
g
u
i
d
e
d
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
i
n
2
0
1
9
I
EEE
/
C
VF
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
rn
Re
c
o
g
n
i
t
i
o
n
(
C
V
PR)
,
Ju
n
.
2
0
1
9
,
p
p
.
3
7
1
8
–
3
7
2
7
,
d
o
i
:
1
0
.
1
1
0
9
/
c
v
p
r
.
2
0
1
9
.
0
0
3
8
4
.
[
7
]
Y.
-
L.
C
h
a
n
g
,
Z.
Y
.
Li
u
,
K
.
-
Y
.
L
e
e
,
a
n
d
W
.
H
s
u
,
“
F
r
e
e
-
f
o
r
m
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
w
i
t
h
3
D
g
a
t
e
d
c
o
n
v
o
l
u
t
i
o
n
a
n
d
t
e
mp
o
r
a
l
P
a
t
c
h
G
A
N
,
”
i
n
2
0
1
9
I
EEE
/
C
VF
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
(
I
C
C
V)
,
O
c
t
.
2
0
1
9
,
p
p
.
9
0
6
5
–
9
0
7
4
,
d
o
i
:
1
0
.
1
1
0
9
/
i
c
c
v
.
2
0
1
9
.
0
0
9
1
6
.
[
8
]
D
.
K
i
m,
S
.
W
o
o
,
J.
-
Y
.
Le
e
,
a
n
d
I
.
S
.
K
w
e
o
n
,
“
D
e
e
p
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
i
n
2
0
1
9
I
EEE
/
C
VF
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
(
C
VP
R)
,
J
u
n
.
2
0
1
9
,
p
p
.
5
7
8
5
–
5
7
9
4
,
d
o
i
:
1
0
.
1
1
0
9
/
c
v
p
r
.
2
0
1
9
.
0
0
5
9
4
.
[
9
]
D
.
La
o
,
P
.
Z
h
u
,
P
.
W
o
n
k
a
,
a
n
d
G
.
S
u
n
d
a
r
a
mo
o
r
t
h
i
,
“
F
l
o
w
-
g
u
i
d
e
d
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
w
i
t
h
s
c
e
n
e
t
e
mp
l
a
t
e
s
,
”
2
0
2
1
I
E
EE/
C
VF
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
C
o
m
p
u
t
e
r
Vi
s
i
o
n
(
I
C
C
V)
,
M
o
n
t
r
e
a
l
,
Q
C
,
C
a
n
a
d
a
,
2
0
2
1
,
p
p
.
1
4
5
7
9
-
1
4
5
8
8
,
d
o
i
:
1
0
.
1
1
0
9
/
i
c
c
v
4
8
9
2
2
.
2
0
2
1
.
0
1
4
3
3
.
[
1
0
]
Z.
Li
,
C
.
-
Z
.
L
u
,
J.
Q
i
n
,
C
.
-
L
.
G
u
o
,
a
n
d
M
.
-
M
.
C
h
e
n
g
,
“
T
o
w
a
r
d
s
a
n
e
n
d
-
to
-
e
n
d
f
r
a
mew
o
r
k
f
o
r
f
l
o
w
-
g
u
i
d
e
d
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
a
rXi
v
:
2
2
0
4
.
0
2
6
6
3
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
2
2
0
4
.
0
2
6
6
3
.
[
1
1
]
C
.
W
a
n
g
,
H
.
H
u
a
n
g
,
X
.
H
a
n
,
a
n
d
J.
W
a
n
g
,
“
V
i
d
e
o
i
n
p
a
i
n
t
i
n
g
b
y
j
o
i
n
t
l
y
l
e
a
r
n
i
n
g
t
e
mp
o
r
a
l
st
r
u
c
t
u
r
e
a
n
d
sp
a
t
i
a
l
d
e
t
a
i
l
s,”
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
AAAI
C
o
n
f
e
r
e
n
c
e
o
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
3
,
n
o
.
0
1
,
p
p
.
5
2
3
2
–
5
2
3
9
,
Ju
l
.
2
0
1
9
,
d
o
i
:
1
0
.
1
6
0
9
/
a
a
a
i
.
v
3
3
i
0
1
.
3
3
0
1
5
2
3
2
.
[
1
2
]
X
.
Z
o
u
,
L.
Y
a
n
g
,
D
.
Li
u
,
a
n
d
Y
.
Jae
Le
e
,
“
P
r
o
g
r
e
ssi
v
e
t
e
m
p
o
r
a
l
f
e
a
t
u
r
e
a
l
i
g
n
m
e
n
t
n
e
t
w
o
r
k
f
o
r
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
2
0
2
1
I
EEE
/
C
V
F
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
(
C
V
PR)
.
I
EEE,
p
p
.
1
6
4
4
3
–
1
6
4
5
2
,
Ju
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
c
v
p
r
4
6
4
3
7
.
2
0
2
1
.
0
1
6
1
8
.
[
1
3
]
J.
R
e
n
,
Q
.
Zh
e
n
g
,
Y
.
Z
h
a
o
,
X
.
X
u
,
a
n
d
C
.
Li
,
“
D
LFo
r
mer
:
d
i
s
c
r
e
t
e
l
a
t
e
n
t
t
r
a
n
sf
o
r
mer
f
o
r
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
i
n
2
0
2
2
I
EE
E/
C
V
F
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
(
C
V
PR)
,
Ju
n
.
2
0
2
2
,
p
p
.
3
5
0
1
–
3
5
1
0
,
d
o
i
:
1
0
.
1
1
0
9
/
c
v
p
r
5
2
6
8
8
.
2
0
2
2
.
0
0
3
5
0
.
[
1
4
]
R
.
L
i
u
,
Z.
W
e
n
g
,
Y
.
Zh
u
,
a
n
d
B
.
Li
,
“
T
e
mp
o
r
a
l
a
d
a
p
t
i
v
e
a
l
i
g
n
m
e
n
t
n
e
t
w
o
r
k
f
o
r
d
e
e
p
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
T
w
e
n
t
y
-
N
i
n
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
re
n
c
e
o
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
J
u
l
.
2
0
2
0
,
p
p
.
9
2
7
–
9
3
3
,
d
o
i
:
1
0
.
2
4
9
6
3
/
i
j
c
a
i
.
2
0
2
0
/
1
2
9
.
[
1
5
]
C
.
W
a
n
g
,
X
.
C
h
e
n
,
S
.
M
i
n
,
J.
W
a
n
g
,
a
n
d
Z.
-
J.
Zh
a
,
“
S
t
r
u
c
t
u
r
e
-
g
u
i
d
e
d
d
e
e
p
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
I
EE
E
T
ra
n
sa
c
t
i
o
n
s
o
n
C
i
r
c
u
i
t
s a
n
d
S
y
s
t
e
m
s
f
o
r
V
i
d
e
o
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
3
1
,
n
o
.
8
,
p
p
.
2
9
5
3
–
2
9
6
5
,
A
u
g
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
t
c
sv
t
.
2
0
2
0
.
3
0
3
4
4
2
2
.
[
1
6
]
Y.
-
L.
C
h
a
n
g
,
Z.
Y
.
L
i
u
,
a
n
d
W
.
H
s
u
,
“
V
O
R
N
e
t
:
sp
a
t
i
o
-
t
e
m
p
o
r
a
l
l
y
c
o
n
s
i
st
e
n
t
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
f
o
r
o
b
j
e
c
t
r
e
mo
v
a
l
,
”
i
n
2
0
1
9
I
EEE/
C
VF
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
W
o
rks
h
o
p
s
(
C
VPR
W)
,
J
u
n
.
2
0
1
9
,
p
p
.
1
7
8
5
–
1
7
9
4
,
d
o
i
:
1
0
.
1
1
0
9
/
c
v
p
r
w
.
2
0
1
9
.
0
0
2
2
9
.
[
1
7
]
K
.
Z
h
a
n
g
,
J.
F
u
,
a
n
d
D
.
L
i
u
,
“
I
n
e
r
t
i
a
-
g
u
i
d
e
d
f
l
o
w
c
o
m
p
l
e
t
i
o
n
a
n
d
s
t
y
l
e
f
u
si
o
n
f
o
r
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
i
n
2
0
2
2
I
EEE
/
C
VF
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
a
n
d
P
a
t
t
e
r
n
R
e
c
o
g
n
i
t
i
o
n
(
C
V
PR)
,
Ju
n
.
2
0
2
2
,
p
p
.
5
9
7
2
–
5
9
8
1
,
d
o
i
:
1
0
.
1
1
0
9
/
c
v
p
r
5
2
6
8
8
.
2
0
2
2
.
0
0
5
8
9
.
[
1
8
]
X
.
D
i
n
g
,
Y
.
P
a
n
,
K
.
Lu
o
,
Y
.
H
u
a
n
g
,
J.
O
u
y
a
n
g
,
a
n
d
G
.
Y
a
n
g
,
“
L
o
c
a
l
i
z
a
t
i
o
n
o
f
d
e
e
p
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
b
a
se
d
o
n
s
p
a
t
i
o
t
e
m
p
o
r
a
l
c
o
n
v
o
l
u
t
i
o
n
a
n
d
r
e
f
i
n
e
m
e
n
t
n
e
t
w
o
r
k
,
”
i
n
2
0
2
1
I
E
EE
I
n
t
e
rn
a
t
i
o
n
a
l
S
y
m
p
o
si
u
m
o
n
C
i
rc
u
i
t
s
a
n
d
S
y
s
t
e
m
s
(
I
S
C
AS
)
,
M
a
y
2
0
2
1
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
i
sca
s
5
1
5
5
6
.
2
0
2
1
.
9
4
0
1
6
7
5
.
[
1
9
]
I
.
S
i
d
d
a
v
a
t
a
m,
A
.
D
a
l
v
i
,
D
.
P
a
w
a
d
e
,
A
.
B
h
a
t
t
,
J.
V
a
r
t
a
k
,
a
n
d
A
.
G
u
p
t
a
,
“
A
n
o
v
e
l
a
p
p
r
o
a
c
h
f
o
r
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
u
si
n
g
a
u
t
o
e
n
c
o
d
e
r
s
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
E
n
g
i
n
e
e
ri
n
g
a
n
d
E
l
e
c
t
ro
n
i
c
B
u
si
n
e
ss
,
v
o
l
.
1
3
,
n
o
.
6
,
p
p
.
4
8
–
6
1
,
D
e
c
.
2
0
2
1
,
d
o
i
:
1
0
.
5
8
1
5
/
i
j
i
e
e
b
.
2
0
2
1
.
0
6
.
0
5
.
[
2
0
]
L.
K
e
,
Y
.
-
W
.
Ta
i
,
a
n
d
C
.
-
K
.
T
a
n
g
,
“
O
c
c
l
u
si
o
n
-
a
w
a
r
e
v
i
d
e
o
o
b
j
e
c
t
i
n
p
a
i
n
t
i
n
g
,
”
i
n
2
0
2
1
I
EE
E/
C
VF
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
(
I
C
C
V)
,
O
c
t
.
2
0
2
1
,
p
p
.
1
4
4
4
8
–
1
4
4
5
8
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
4
8
9
2
2
.
2
0
2
1
.
0
1
4
2
0
.
[
2
1
]
R
.
S
z
e
t
o
,
X
.
S
u
n
,
K
.
Lu
,
a
n
d
J.
J.
C
o
r
so
,
“
A
t
e
m
p
o
r
a
l
l
y
a
w
a
r
e
i
n
t
e
r
p
o
l
a
t
i
o
n
n
e
t
w
o
r
k
f
o
r
v
i
d
e
o
f
r
a
me
i
n
p
a
i
n
t
i
n
g
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
P
a
t
t
e
rn
A
n
a
l
y
si
s
a
n
d
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
2
,
n
o
.
5
,
p
p
.
1
0
5
3
–
1
0
6
8
,
M
a
y
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
1
9
.
2
9
5
1
6
6
7
.
[
2
2
]
H.
-
Y
.
H
u
a
n
g
a
n
d
C
.
-
H
.
L
i
n
,
“
V
i
d
e
o
i
n
p
a
i
n
t
i
n
g
u
si
n
g
o
b
j
e
c
t
m
o
t
i
o
n
r
a
t
e
a
n
d
c
o
l
o
r
v
a
r
i
a
n
c
e
i
n
s
p
a
t
i
o
t
e
mp
o
r
a
l
d
o
mai
n
,
”
J
o
u
r
n
a
l
o
f
I
n
t
e
l
l
i
g
e
n
t
&
F
u
zzy
S
y
st
e
m
s
,
v
o
l
.
4
0
,
n
o
.
3
,
p
p
.
5
6
0
9
–
5
6
2
2
,
M
a
r
.
2
0
2
1
,
d
o
i
:
1
0
.
3
2
3
3
/
JI
F
S
-
2
0
0
5
4
2
.
[
2
3
]
B
.
Y
u
,
W
.
Li
,
X
.
Li
,
J
.
Lu
,
a
n
d
J.
Zh
o
u
,
“
F
r
e
q
u
e
n
c
y
a
w
a
r
e
s
p
a
t
i
o
t
e
mp
o
r
a
l
t
r
a
n
sf
o
r
mers
f
o
r
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
d
e
t
e
c
t
i
o
n
,
”
i
n
2
0
2
1
I
EEE/
C
VF
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
C
o
m
p
u
t
e
r
Vi
s
i
o
n
(
I
C
C
V)
,
O
c
t
.
2
0
2
1
,
p
p
.
8
1
6
8
–
8
1
7
7
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
4
8
9
2
2
.
2
0
2
1
.
0
0
8
0
8
.
[
2
4
]
R
.
S
z
e
t
o
a
n
d
J
.
J.
C
o
r
so
,
“
T
h
e
d
e
v
i
l
i
s
i
n
t
h
e
d
e
t
a
i
l
s
a
d
i
a
g
n
o
st
i
c
e
v
a
l
u
a
t
i
o
n
b
e
n
c
h
mar
k
f
o
r
v
i
d
e
o
i
n
p
a
i
n
t
i
n
g
,
”
a
rXi
v
:
2
1
0
5
.
0
5
3
3
2
,
2
0
2
2
,
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
2
1
0
5
.
0
5
3
3
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
efficien
t d
ir
ec
tio
n
o
r
ien
ted
b
lo
ck
-
b
a
s
ed
vid
eo
in
p
a
in
tin
g
u
s
in
g
…
(
S
h
yn
i S
h
a
ja
h
a
n
)
4713
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
S
h
y
n
i
S
h
a
ja
h
a
n
h
o
ld
s
a
B
.
Tec
h
.
a
n
d
M
.
Tec
h
.
i
n
c
o
m
p
u
ter
sc
ie
n
c
e
fro
m
CUSAT
a
n
d
a
P
h
.
D.
fr
o
m
No
o
ru
l
Isla
m
Ce
n
tre
fo
r
Hig
h
e
r
Ed
u
c
a
ti
o
n
(2
0
2
3
).
S
h
e
is
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
i
n
t
h
e
De
p
a
rtme
n
t
o
f
C
o
m
p
u
ter S
c
ien
c
e
a
t
Ad
i
S
h
a
n
k
a
ra
In
stit
u
te
o
f
En
g
in
e
e
ri
n
g
a
n
d
Tec
h
n
o
l
o
g
y
,
Ka
lad
y
.
He
r
re
se
a
r
c
h
in
tere
sts
in
c
lu
d
e
ima
g
e
p
ro
c
e
ss
in
g
,
c
o
d
in
g
tec
h
n
i
q
u
e
s,
DBMS
,
d
a
ta
m
in
i
n
g
,
m
a
c
h
in
e
le
a
rn
in
g
,
d
a
ta
sc
ien
c
e
,
AI,
a
n
d
m
u
lt
ime
d
ia.
S
h
e
h
a
s
p
u
b
li
s
h
e
d
e
x
ten
siv
e
ly
in
n
a
ti
o
n
a
l
a
n
d
i
n
tern
a
ti
o
n
a
l
j
o
u
r
n
a
ls.
Sh
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
sh
y
n
i
.
c
s@
a
d
ish
a
n
k
a
ra
.
a
c
.
in
.
Y.
J
a
c
o
b
Ve
th
a
R
a
j
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
th
e
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
t
Ne
s
a
m
o
n
y
M
e
m
o
ria
l
Ch
risti
a
n
Co
ll
e
g
e
,
In
d
ia.
He
re
c
e
iv
e
d
a
B.
E.
d
e
g
re
e
in
c
o
m
p
u
ter
sc
ien
c
e
fro
m
Na
ti
o
n
a
l
En
g
in
e
e
rin
g
Co
l
leg
e
,
Ko
v
il
p
a
tt
i,
In
d
ia,
a
n
d
a
lso
re
c
e
iv
e
d
M
.
Tec
h
.
a
n
d
P
h
.
D.
d
e
g
re
e
s
fro
m
M
.
S
.
U
n
iv
e
rsit
y
,
Ti
r
u
n
e
l
v
e
li
,
In
d
ia.
His
a
re
a
o
f
in
tere
st
i
s
ima
g
e
p
ro
c
e
ss
in
g
.
He
h
a
s
d
e
v
e
lo
p
e
d
m
a
n
y
a
p
p
li
c
a
ti
o
n
so
ftwa
re
p
ro
g
ra
m
s
a
n
d
h
a
s
p
u
b
li
sh
e
d
n
u
m
e
ro
u
s
n
a
ti
o
n
a
l
a
n
d
i
n
tern
a
ti
o
n
a
l
re
se
a
rc
h
p
a
p
e
rs
.
H
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
jac
o
b
v
e
th
a
ra
j@g
m
a
il
.
c
o
m
Evaluation Warning : The document was created with Spire.PDF for Python.