I
n
t
ern
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
A
d
v
a
n
ces
i
n
A
p
p
l
i
ed
S
ci
en
ces
(
I
J
A
A
S
)
V
o
l
.
7
,
N
o
.
2
,
J
une
201
8
,
p
p
.
1
56
~
1
61
IS
S
N
:
225
2
-
88
14
,
D
OI
:
10.
115
91
/ij
a
a
s
.
v7
.
i
2
.
p
p
15
6
-
1
61
1
56
Jo
u
r
n
al
h
om
e
pa
ge
:
h
t
t
p
:
/
/
i
a
e
s
co
r
e
.
c
o
m/
o
n
l
i
n
e
/
i
n
d
e
x
.
p
h
p
/
I
J
A
A
S
V
i
s
i
o
n B
a
s
e
d A
p
pr
oa
c
h
t
o
Si
g
n L
a
ng
uage
R
ec
og
ni
t
i
o
n
H
i
m
a
ns
hu
G
u
pt
a
,
A
ni
r
ud
dh
R
a
m
j
i
w
a
l
,
J
a
s
m
i
n
T
.
J
o
s
e
S
c
hool
of Com
p
ut
e
r
E
ngi
ne
e
ri
ng
,
V
IT
U
ni
ve
rs
i
t
y,
V
e
l
l
ore
-
63201
4,
Indi
a
A
rt
i
cl
e
I
nf
o
A
BS
TR
A
C
T
Ar
t
i
c
l
e
h
i
s
t
o
r
y
:
R
ecei
v
ed
D
ec
1
9
,
2
01
7
Re
v
i
s
e
d
M
ar
4
,
2
0
1
8
A
ccep
t
e
d
Ma
y
2
,
20
1
8
W
e
propos
e
a
n a
l
gori
t
hm
for a
ut
om
a
t
i
c
a
l
l
y
re
c
og
ni
z
i
ng s
om
e
c
e
rt
a
i
n a
m
ount
of ge
s
t
ure
s
from
ha
nd m
ove
m
e
nt
s
t
o he
l
p de
a
f a
nd dum
b a
nd ha
rd he
a
ri
n
g
pe
opl
e
.
H
a
nd
ge
s
t
ure
re
c
ogni
t
i
on
i
s
qui
t
e
a
c
ha
l
l
e
ngi
ng probl
e
m
i
n i
t
s
form
.
W
e
ha
ve
c
ons
i
de
re
d a
fi
x
e
d s
e
t
of m
a
nua
l
c
om
m
a
nds
a
nd
a
s
pe
c
i
fi
c
e
nvi
ronm
e
nt
,
a
n
d d
e
ve
l
op a
e
ff
e
c
t
i
v
e
,
pro
c
e
dur
e
for ge
s
t
ure
r
e
c
o
gni
t
i
on.
O
ur
a
pproa
c
h
c
ont
a
i
ns
s
t
e
ps
for s
e
g
m
e
nt
i
ng t
he
ha
n
d r
e
gi
on,
l
o
c
a
t
i
n
g t
he
fi
nge
rs
,
a
nd fi
na
l
l
y
c
l
a
s
s
i
f
y
i
ng t
he
g
e
s
t
ur
e
w
hi
c
h i
n ge
n
e
ra
l
t
e
rm
s
m
e
a
ns
de
t
e
c
t
i
ng
,
t
ra
c
k
i
ng a
nd
r
e
c
ogni
s
i
ng
.
T
h
e
a
l
gori
t
hm
i
s
non
-
c
ha
ngi
ng
t
o rot
a
t
i
ons
,
t
ra
ns
l
a
t
i
ons
a
n
d s
c
a
l
e
of t
h
e
ha
nd.
W
e
w
i
l
l
be
de
m
ons
t
ra
t
i
ng t
he
ef
f
ec
t
i
v
en
es
s
o
f
t
he
t
e
c
hn
i
que
on
re
a
l
i
m
a
ge
r
y
.
Ke
y
wo
r
d
:
D
e
t
e
c
t
i
o
n
G
e
s
tur
e
R
e
c
o
g
n
i
t
i
o
n
,
S
eg
m
en
t
T
r
ack
Copy
r
i
ght
©
201
8
Ins
t
i
t
ut
e
o
f
A
d
v
anc
e
d
E
ngi
n
e
e
r
i
ng and S
c
i
e
nc
e
.
A
l
l
ri
g
h
t
s re
se
rv
e
d
.
C
or
r
e
s
po
n
di
n
g
A
u
t
h
or
:
H
i
m
a
ns
hu
G
up
t
a
,
S
c
h
ool
o
f
C
om
put
e
r
E
ngi
ne
e
r
i
ng
,
V
IT
U
n
i
v
e
r
s
i
t
y
,
V
e
l
l
o
r
e
-
632
014
,
I
n
d
ia
.
E
m
a
i
l
:
gupt
a
hi
m
a
ns
h
u1
2@
g
m
a
i
l
.
c
o
m
1.
I
N
T
R
O
D
U
C
T
I
O
N
T
he
s
i
gn
l
a
n
g
ua
ge
i
s
ba
s
i
c
a
l
l
y
a
l
a
ng
ua
ge
i
n
w
hi
c
h t
he
r
e
i
s
m
a
nua
l
c
om
m
uni
c
a
t
i
on t
o
c
on
ve
y
m
e
a
ni
ng,
c
o
nt
r
a
r
y
t
o c
o
n
ve
y
e
d s
ou
n
d pa
t
t
e
r
ns
.
T
hi
s
i
n
v
ol
v
e
s
s
im
ul
t
a
ne
o
u
s
c
om
bi
na
t
i
ons
of
or
i
e
nt
a
t
i
on
,
ha
n
d
s
ha
pe
s
a
nd m
ove
m
e
nt
of
t
h
e
pe
r
s
o
n’
s
ha
nds
,
a
r
m
s
a
nd
b
ody
t
o
e
x
pr
e
s
s
hi
s
t
ho
u
gh
t
s
.
A
num
be
r
o
f
s
i
g
n
l
a
ng
ua
ge
s
s
ha
r
e
a
l
ot
of
s
im
il
a
r
i
t
i
e
s
w
i
t
h
t
he
s
po
ke
n
l
a
ng
ua
ge
s
,
t
ha
t
de
pe
n
ds
pr
i
m
a
r
i
l
y
o
n
s
o
un
d,
a
nd
l
i
ng
ui
s
t
s
c
on
s
i
de
r
e
d
b
o
t
h
t
o
be
t
he
t
y
pe
s
of
na
t
u
r
a
l
l
a
ng
ua
ge
.
A
l
t
h
o
ug
h
t
he
r
e
a
r
e
m
a
ny
s
i
gni
f
i
c
a
nt
di
f
f
e
r
e
nc
e
s
be
t
w
e
e
n
s
po
ke
n a
n
d s
i
gne
d
l
a
n
gua
g
e
s
,
l
i
ke
a
s
h
o
w t
he
y
us
e
s
p
a
c
e
g
r
a
m
m
a
t
i
c
a
l
l
y
,
s
i
gn l
a
ng
ua
ge
s
s
ho
w t
h
e
s
a
m
e
l
i
n
g
u
i
s
t
i
c
p
r
o
p
er
t
i
es
an
d
u
s
e
t
h
e
s
am
e
l
an
g
u
ag
e
f
acu
l
t
y
as
d
o
s
pok
e
n
la
ngu
a
g
e
s
.
S
t
i
l
l
i
t
i
s
not
k
no
w
n
t
ha
t
h
o
w
m
a
ny
s
i
gn
l
a
n
gua
ge
s
e
xi
s
t
s
.
O
ne
o
f
t
he
ve
r
y
c
om
m
on
m
i
s
c
onc
e
pt
i
o
n
i
s
a
l
l
s
i
gn l
a
n
g
ua
ge
s
a
r
e
s
a
m
e
w
o
r
l
d
w
i
de
o
r
s
i
gn l
a
ng
ua
ge
i
s
i
nt
e
r
na
t
i
ona
l
.
A
s
i
de
f
r
om
t
he
pi
d
gi
n
I
nt
e
r
n
a
t
i
ona
l
S
i
gn
,
a
l
m
os
t
e
ve
r
y
c
o
unt
r
y
h
a
s
i
t
s
ow
n,
na
t
i
v
e
s
i
gn
l
a
n
gua
g
e
,
a
n
d
s
om
e
c
ount
r
i
e
s
ha
ve
m
or
e
t
ha
n
o
ne
b
e
c
a
us
e
of
di
f
f
e
r
e
nt
l
a
ng
ua
ge
s
i
n t
he
c
ou
nt
r
y
i
t
s
e
l
f
ba
s
e
d o
n t
he
r
e
gi
o
ns
.
T
he
2
0
13 e
di
t
i
on
of
E
t
hn
ol
o
g
ue
l
i
s
t
s
1
3
7
s
i
gn l
a
n
g
ua
ge
s
.
S
om
e
s
i
gn l
a
ng
ua
ge
s
ha
ve
a
c
hi
e
ve
d s
om
e
f
or
m
of
l
e
ga
l
r
eco
g
n
i
t
i
o
n
,
w
h
i
l
e t
h
er
e ar
e
m
an
y
ot
he
r
s
w
hi
c
h
h
a
ve
no
s
t
a
t
us
a
t
a
l
l
.
I
n S
out
h A
s
i
a
,
I
nd
o
-
P
a
ki
s
t
a
n
i
S
i
gn L
a
n
g
ua
ge
(
I
P
S
L
)
i
s
t
he
p
r
e
d
om
i
na
nt
s
i
gn l
a
ng
ua
g
e
us
e
d
by
a
t
l
e
a
s
t
s
e
ve
r
a
l
h
un
d
r
e
d t
h
o
us
a
nd
de
a
f
s
i
g
ne
r
s
.
As
w
i
t
h m
a
ny
s
i
gn l
a
n
g
u
a
ge
s
,
i
t
i
s
hi
g
hl
y
di
f
f
i
c
u
l
t
t
o
c
o
u
n
t
num
be
r
s
wi
t
h
a
ny
c
e
r
t
a
i
nt
y
,
a
s
t
he
C
e
ns
us
of
I
n
di
a
ha
s
n
o
t
l
i
s
t
e
d
s
i
gn
l
a
ng
ua
ge
s
a
n
d
m
os
t
o
f
t
he
s
t
u
di
e
s
ha
ve
f
oc
us
e
d
on
t
he
no
r
t
h
a
n
d
ur
ba
n
a
r
e
a
s
.
G
e
s
t
ur
e
s
a
r
e
ve
r
y
p
owe
r
f
ul
m
ode
o
f
c
om
m
uni
c
a
t
i
on a
m
on
g
h
um
a
ns
.
A
m
ongs
t
di
f
f
e
r
e
nt
t
y
pe
s
of
m
oda
l
i
ty
of
b
o
dy
,
t
he
ha
n
d
g
e
s
t
ur
e
i
s
t
he
s
i
m
pl
e
s
t
a
nd na
t
ur
a
l
wa
y
of
c
o
m
m
uni
c
a
t
i
on
m
ode
.
T
he
R
e
a
l
t
im
e
,
vi
s
i
o
n ba
s
e
d
ha
n
d
ge
s
t
u
r
e
r
e
c
o
g
ni
t
i
on i
s
m
or
e
pr
oba
bl
e
be
c
a
us
e
of
t
he
l
a
t
e
s
t
a
dv
a
nc
e
s
i
n
t
he
f
i
e
l
d o
f
c
om
put
e
r
vi
s
i
o
n
a
nd
pa
t
t
e
r
n
r
e
c
og
ni
t
i
on
a
n
d
im
a
ge
pr
oc
e
s
s
i
ng
b
ut
i
t
ha
s
y
e
t
,
t
o
be
t
ot
a
l
l
y
e
x
pl
o
r
e
d
f
or
H
um
a
n
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
A
A
S
I
S
S
N
:
225
2
-
88
14
V
i
s
i
on
B
as
e
d
A
ppr
o
ac
h
t
o
Si
g
n
L
a
ng
u
age
R
e
c
og
ni
t
i
on
(
H
i
m
ans
h
u
Gu
pt
a
)
1
57
Co
m
p
u
t
e
r
I
n
t
e
r
a
c
t
i
o
n
(
H
CI
)
.
W
i
t
h
t
h
e
w
i
d
e
a
p
p
l
i
c
a
t
i
o
n
s
o
f
H
CI
,
n
o
w
d
a
y
s
,
i
t
b
e
c
o
m
e
s
a
c
t
i
v
e
f
o
c
u
s
o
f
r
es
ear
c
h
.
I
n
r
ece
n
t
y
ear
s
,
v
i
s
i
o
n
-
ba
s
e
d
a
ut
om
a
t
i
c
ha
nd ge
s
t
ur
e
r
e
c
o
gni
t
i
o
n
ha
s
be
e
n a
n
a
c
t
i
ve
r
e
s
ear
ch
t
o
p
i
c
w
i
t
h a
l
ot
o
f
m
ot
i
va
t
i
ng
a
p
pl
i
c
a
t
i
ons
s
uc
h a
s
s
i
gn l
a
ng
ua
ge
i
nt
e
r
p
r
e
t
a
t
i
on
,
r
o
b
ot
c
o
nt
r
ol
a
nd
h
um
a
n c
om
put
e
r
i
nt
e
r
a
c
t
i
on
(
H
C
I
)
.
B
ut
t
he
pr
obl
e
m
i
s
c
ha
l
l
e
ngi
ng
be
c
a
us
e
o
f
a
n
u
m
be
r
of
i
s
s
ue
s
w
hi
c
h i
nc
l
u
de
s
t
he
co
m
p
l
i
cat
ed
n
at
u
r
e
o
f
d
y
n
am
i
c
an
d
s
t
a
t
i
c
ha
nd
ge
s
t
u
r
e
s
,
oc
c
l
us
i
o
ns
a
n
d
c
om
pl
e
x
ba
c
k
gr
ou
n
ds
.
W
e
ne
e
d
ve
r
y
i
n
t
e
n
s
i
v
e
c
o
m
p
u
t
e
r
r
e
s
o
u
r
c
e
s
t
o
a
t
t
a
c
k
a
p
r
o
b
l
e
m
i
n
i
t
s
g
e
n
e
r
a
l
i
t
y
w
h
i
c
h
r
e
q
u
i
r
e
s
e
l
a
b
o
r
a
t
e
a
l
g
o
r
i
t
h
m
s
.
O
u
r
m
oti
va
t
i
on t
o
do t
hi
s
w
o
r
k i
s
a
r
ob
ot
na
vi
ga
t
i
on
p
r
o
bl
e
m
,
be
c
a
us
e
we
a
r
e
hi
g
h
l
y
i
n
t
e
r
e
s
t
e
d
i
n
c
o
n
t
r
o
l
l
i
n
g
a
r
o
b
ot
by
ha
nd
s
i
g
ns
gi
ve
n
by
a
h
um
a
n b
e
i
ng.
W
e
a
r
e
i
nt
e
r
e
s
t
e
d
i
n c
om
put
a
t
i
ona
l
l
y
e
f
f
i
c
i
e
nt
a
l
g
o
r
i
t
hm
b
ecau
s
e o
f
r
ea
l
-
t
im
e
ope
r
a
t
i
o
na
l
r
e
q
ui
r
e
m
e
nt
s
.
I
n ha
n
d g
e
s
t
ur
e
r
e
c
o
gni
t
i
on p
r
obl
e
m
in e
a
r
l
i
e
r
a
p
p
r
oa
c
he
s
t
h
er
e
w
as
a
u
s
e o
f
m
a
r
ke
r
s
o
n t
he
f
i
nge
r
t
i
p
s
i
n a
r
ob
ot
c
o
nt
r
ol
c
o
nt
e
xt
.
T
o i
de
nt
i
f
y
w
h
i
c
h a
r
e
t
he
f
i
n
g
e
r
s
t
ha
t
ar
e act
i
v
e i
n
t
h
e g
es
t
u
r
e t
o
d
et
ect
t
h
e co
l
o
r
an
d
p
r
es
e
n
ce
o
f
t
h
e m
ar
k
er
s
,
an
as
s
o
ci
at
ed
al
g
o
r
i
t
h
m
w
as
u
s
ed
.
B
ut
w
ha
t
m
a
ke
s
t
hi
s
a
s
a
n
i
nf
e
a
s
i
bl
e
a
p
pr
oa
c
h i
s
pl
a
c
i
ng m
a
r
ke
r
s
o
n t
he
us
e
r
’
s
ha
n
d
w
hi
c
h i
s
ve
r
y
i
n
co
n
v
e
n
i
en
t
.
R
ecen
t
m
et
h
o
d
s
u
s
e m
o
r
e accu
r
at
e co
m
p
u
t
er
t
ech
n
i
q
u
es
an
d
t
h
ey
d
o
n
’
t
r
e
q
u
i
r
e m
ar
k
er
s
.
H
an
d
ge
s
t
u
r
e
r
e
c
o
gn
i
t
i
on i
s
pe
r
f
or
m
e
d t
hr
ou
g
h a
c
ur
va
t
u
r
e
s
pa
c
e
m
e
t
hod i
n,
w
hi
c
h i
n
vol
ve
s
f
i
n
di
n
g t
he
b
ou
n
da
r
y
c
ont
ou
r
s
of
t
he
ha
n
d.
I
t
i
s
a
r
ob
us
t
a
p
p
r
oa
c
h
t
ha
t
i
s
t
r
a
ns
l
a
t
i
on,
s
c
a
l
e
a
nd
r
ot
a
t
i
o
n i
n
va
r
i
a
nt
o
n t
he
ha
n
d p
os
e
,
y
e
t
i
t
i
s
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
d
e
m
a
n
d
i
n
g
.
A
m
u
l
t
i
-
s
y
s
t
e
m
c
a
m
e
r
a
i
s
us
e
d t
o
pi
c
k t
he
C
O
G
of
t
he
ha
n
d a
n
d
p
oi
nt
s
i
t
w
i
t
h
f
a
r
t
h
e
s
t
d
i
s
t
a
n
c
e
s
f
r
o
m
t
h
e
c
e
nt
e
r
,
gi
vi
n
g
us
t
he
l
oc
a
t
i
on
s
of
t
he
f
i
n
ge
r
t
i
ps
,
w
hi
c
h
a
r
e
t
he
n
us
e
d t
o
o
b
t
a
i
n
a
s
k
e
l
e
t
o
n
i
m
a
g
e
,
a
n
d
t
h
e
n
f
i
n
a
l
l
y
f
o
r
t
h
e
g
e
s
t
u
r
e
r
e
c
o
g
n
i
t
i
o
n
a
n
d
i
t
i
s
p
r
o
p
o
s
e
d
i
n
a
v
i
s
i
o
n
-
ba
s
e
d
ha
n
d
pos
e
r
e
c
o
g
ni
t
i
on
t
e
c
h
ni
q
ue
u
s
i
ng
s
ke
l
e
t
on
i
m
a
ge
s
.
A
t
e
c
h
ni
q
ue
wa
s
p
r
o
p
os
e
d
f
or
ge
s
t
ur
e
r
e
c
o
gni
t
i
o
n
f
or
s
i
g
n
l
a
ng
ua
ge
r
e
c
o
gni
t
i
o
n.
P
a
r
t
i
c
l
e
f
i
l
t
e
r
s
,
F
our
i
e
r
de
s
c
r
i
pt
or
s
,
pr
i
nc
i
pa
l
c
om
pone
nt
a
na
l
y
s
i
s
,
or
i
e
nt
a
t
i
o
n
hi
s
t
o
gr
a
m
s
,
ne
ur
a
l
ne
t
w
o
r
ks
a
nd s
pe
c
i
a
l
i
z
e
d m
a
ppi
ngs
a
r
c
hi
t
e
c
t
ur
e
a
r
e
i
nc
l
ude
d i
n t
h
e
3D
a
nd
2D
ha
n
d
g
es
t
u
r
e
r
eco
g
n
i
t
i
on
w
hi
c
h
a
r
e
t
he
c
om
put
e
r
vi
s
i
o
n
t
o
ol
s
.
O
ur
f
oc
u
s
i
s
t
he
r
e
c
o
g
ni
t
i
on
o
f
a
f
i
xe
d
s
e
t
o
f
m
a
nua
l
c
om
m
a
nds
by
a
r
o
bot
,
i
n a
r
e
a
s
o
na
bl
y
s
t
r
uc
t
ur
e
d e
n
vi
r
o
nm
e
nt
i
n r
e
a
l
t
im
e
.
T
he
r
e
f
or
e
t
h
e
im
por
t
a
nt
t
hi
ng
f
o
r
u
s
i
s
s
p
e
e
d
,
h
e
n
c
e
s
i
m
p
l
i
c
i
t
y
o
f
t
h
e
a
l
g
o
r
i
t
h
m
.
W
e
de
ve
l
o
p
a
n
d
i
m
pl
e
m
e
nt
s
uc
h
a
p
r
o
c
e
d
ur
e
i
n
t
hi
s
w
or
k.
O
ur
w
a
y
i
n
vol
ve
s
s
e
gm
e
nt
i
ng
t
he
ha
nd
ba
s
e
d
on
s
i
z
e
c
o
ns
t
r
a
i
nt
s
a
n
d
s
ki
n
c
ol
or
s
t
a
t
i
s
t
i
c
s
.
B
a
s
e
d
o
n t
he
pr
e
pr
oc
e
s
s
i
ng
s
t
e
ps
w
hi
c
h a
r
e
f
i
ndi
n
g t
he
f
a
r
t
he
s
t
p
oi
nt
f
r
om
t
he
c
e
nt
e
r
of
g
r
a
vi
t
y
a
nd f
i
n
di
n
g t
he
cen
t
er
o
f
gr
a
vi
t
y
(
C
O
G)
of
t
he
ha
n
d r
e
gi
o
n,
we
w
i
l
l
de
r
i
ve
a
s
i
g
na
l
w
hi
c
h c
a
r
r
i
e
s
i
nf
or
m
a
t
i
on o
n t
he
a
c
t
i
vi
t
y
of
t
he
f
i
n
g
e
r
s
i
n
t
h
e
s
i
g
n
.
F
i
n
a
l
l
y
w
e
i
d
e
n
t
i
f
y
t
h
e
s
i
g
n
b
a
s
e
d
o
n
t
h
a
t
s
i
g
n
a
l
.
O
u
r
a
l
g
o
r
i
t
h
m
i
s
n
o
n
-
c
h
a
ng
ing
to
r
o
t
a
t
i
o
n
s
,
t
r
a
n
s
l
a
t
i
o
n
s
a
n
d
s
c
a
l
e
of
t
he
ha
n
d.
A
l
s
o,
t
hi
s
t
e
c
h
ni
q
ue
d
oe
s
not
ne
e
d a
ny
s
t
or
a
ge
o
f
a
ha
n
d g
e
s
t
ur
e
da
t
a
ba
s
e
i
n
t
he
r
ob
ot
’
s
m
e
m
o
r
y
.
2.
R
ELA
TED
W
O
R
K
E
v
e
n
af
t
er
m
u
ch
o
f
t
h
e
ef
f
o
r
t
s
t
h
er
e
ar
e
m
an
y
s
ch
o
l
ar
l
y
ar
t
i
cl
es
i
n
w
h
i
c
h
d
i
f
f
er
e
n
t
au
t
h
o
r
s
h
av
e
u
s
e
d
di
f
f
e
r
e
nt
t
e
c
h
n
i
que
s
.
I
n
[
1]
t
h
e
au
t
h
o
r
i
n
t
e
n
d
s
t
o
c
r
eat
e
a
t
w
o
d
ev
i
ce
o
n
w
e
ar
ab
l
e
a
n
d
a
n
o
t
h
er
o
n
e
t
h
at
i
s
d
es
k
ba
s
e
d.
T
he
de
v
i
c
e
w
i
l
l
r
e
c
or
d
vi
de
o o
f
pe
r
s
o
n t
a
l
ki
ng i
n a
s
l
.
W
e
a
r
a
bl
e
de
v
i
c
e
wi
l
l
r
e
c
or
d
t
he
vi
de
o of
pe
r
s
o
n
w
e
a
r
i
ng i
t
w
hi
l
e
de
s
k
ba
s
e
d
wi
l
l
r
e
c
or
d
vi
d
e
o o
f
ot
he
r
pe
opl
e
.
W
h
i
l
e
t
h
e
r
e
a
r
e
t
w
o
a
u
t
h
o
r
s
yun
ta
o
c
ui y a
n
d
j
ohn
w
e
ng
z
[
2
]
r
es
ear
ch
e
d
d
i
s
cr
i
m
i
n
an
t
an
a
l
y
s
i
s
w
h
i
ch
a
r
e ch
ar
act
e
r
i
zed
as
f
o
l
l
o
w
s
:
o
n
e
h
as
t
w
o
t
y
p
es
o
f
m
u
l
t
i
v
ar
i
at
e o
b
s
er
v
at
i
o
n
s
.
T
h
e f
i
r
s
t
,
cal
l
ed
t
r
ai
n
i
n
g
s
am
p
l
es
,
ar
e t
h
o
s
e
w
h
o
s
e cl
as
s
i
d
en
t
i
t
y
ar
e k
n
o
w
n
.
T
h
e
s
e
c
o
nd t
y
pe
,
r
e
f
e
r
r
e
d t
o a
s
t
e
s
t
s
a
m
pl
e
s
,
c
ons
i
s
t
s
o
f
o
bs
e
r
va
t
i
o
ns
f
or
w
hi
c
h c
l
a
s
s
i
de
n
t
i
t
y a
r
e
un
kn
o
wn a
n
d
w
hi
c
h ha
ve
t
o be
a
s
s
i
g
ne
d t
o
one
o
f
t
he
c
l
a
s
s
.
O
ne
of
t
he
a
ut
h
or
s
a
i
d t
ha
t
a
s
e
r
i
e
s
of
hy
pot
he
s
e
s
a
n
d t
e
s
t
s
,
w
h
e
r
e a
h
y
p
o
t
h
es
i
s
o
f
m
o
d
el
p
ar
am
et
er
s
at
each
s
t
e
p
i
s
g
en
e
r
at
ed
i
n
t
h
e
d
i
r
ect
i
o
n
o
f
t
h
e p
a
r
am
et
er
s
p
ace
(
f
r
om
t
he
pr
e
vi
ous
hy
p
ot
he
s
i
s
)
a
c
hi
e
vi
n
g t
he
gr
e
a
t
e
s
t
d
ec
r
e
as
e i
n
m
i
s
-
co
r
r
es
p
o
n
d
e
n
ce [
3
]
.
W
h
i
l
e
t
h
er
e w
as
a
cat
eg
o
r
y
i
n
w
h
i
ch
a ap
p
r
o
ac
h
cal
l
ed
“r
u
l
e
d
b
as
e
d
ap
p
r
o
ac
h
” i
s
u
s
ed
.
R
u
l
e
-
b
a
s
ed
a
p
pr
oa
c
he
s
c
ons
i
s
t
of
a
s
e
t
of
m
a
nua
l
l
y
e
nc
o
de
d r
ul
e
s
be
t
we
e
n f
e
a
t
ur
e
i
np
ut
s
.
G
i
ve
n a
n i
n
p
ut
ge
s
t
u
r
e a
s
et
o
f
f
eat
u
r
es
as
i
n
[
4
]
ar
e
e
xt
r
a
c
t
e
d
a
n
d
c
om
pa
r
e
d
t
o
t
he
e
nc
ode
d
r
ul
e
s
,
t
he
r
ul
e
t
ha
t
m
a
t
c
he
s
t
he
i
n
pu
t
i
s
o
ut
p
ut
t
e
d
a
s
t
he
ge
s
t
u
r
e
.
T
h
er
e
h
as
b
ee
n
m
a
ny
r
e
vi
e
w
s
wr
i
t
t
e
n
on
hi
dd
e
n
m
a
r
ko
v
m
ode
l
.
Hi
d
de
n
-
m
a
r
kov
-
m
o
d
e
l
s
(H
M
M
)
a
re
a
ve
r
y
s
t
r
o
ng
a
nd
p
ow
e
r
f
ul
t
e
c
hni
que
f
or
d
e
t
a
i
l
e
d e
xa
m
i
n
a
t
i
on
of
no
n
de
t
e
r
m
i
ni
s
t
i
c
t
im
e
s
i
gna
l
s
.
T
he
y
a
r
e
w
i
de
l
y
us
e
d i
n
ge
s
t
ur
e
r
e
c
o
g
ni
t
i
on,
s
pe
e
c
h r
e
c
o
g
ni
t
i
on a
n
d s
i
g
n l
a
ng
ua
g
e
r
e
c
o
gni
t
i
o
n,
s
i
nc
e
t
he
y
ha
v
e
t
he
a
b
i
l
i
t
y
t
o
h
a
n
d
l
e
s
p
a
t
i
o
-
t
e
m
p
o
r
a
l
v
a
r
i
a
t
i
ons
.
I
n 2
00
2 s
ua
t
a
ky
ol
[
5
]
f
o
u
n
d
a s
p
eci
al
cl
as
s
o
f
H
M
M
s
,
t
h
e s
o
c
a
l
l
e
d l
i
ne
a
r
m
ode
l
s
,
onl
y
ha
ve
f
o
r
w
a
r
d t
r
a
ns
i
t
i
on
s
,
i
.
e
.
A
t
r
a
ns
i
t
i
o
n ne
ve
r
l
e
a
ds
t
o
a
not
he
r
s
t
a
t
e
,
whi
c
h
co
n
n
ect
s
d
i
r
e
c
t
l
y
o
r
i
n
d
i
r
e
c
t
l
y
t
o
t
h
e
c
u
r
r
e
n
t
s
t
a
t
e
.
T
h
i
s
t
o
p
o
l
o
g
y
i
s
s
u
p
p
o
s
e
d
t
o
m
o
d
e
l
t
h
e
c
a
u
s
a
l
i
t
y
c
o
n
d
i
t
i
o
n
o
f
t
i
m
e
s
i
g
n
a
l
s
.
T
h
er
e
h
as
b
een
a
s
i
g
n
i
f
i
can
t
a
m
o
u
n
t
o
f
r
es
e
a
r
c
h
ba
s
e
d
o
n
s
ki
n
c
ol
or
.
I
n
[
6
]
ex
t
r
act
s
s
k
i
n
,
cl
o
t
h
es
an
d
e
l
bo
w
r
e
gi
on
b
y
m
a
t
c
hi
ng
t
e
m
pl
a
t
e
of
i
ni
t
i
a
l
pe
r
s
o
n’
s
r
e
gi
on
.
T
he
n
i
t
t
r
a
c
ks
t
he
f
a
c
e
,
ha
nds
a
n
d
e
l
bo
w
a
t
e
a
c
h
f
r
a
m
e
us
i
ng s
k
i
n c
ol
o
u
r
a
n
d t
e
m
pl
a
t
e
of
s
h
a
pe
.
W
hi
l
e
i
n 9
0s
[
7
]
t
he
r
e
ha
ve
be
e
n e
f
f
or
t
s
f
o
r
c
ol
or
o
f
h
um
a
n
s
ki
n
w
hi
c
h i
s
de
s
c
r
i
be
d
by
s
phe
r
i
c
a
l
i
nf
l
ue
nc
e
f
i
e
l
d
o
f
c
o
l
or
pr
ot
ot
y
pe
s
.
I
t
s
r
e
p
r
es
e
n
t
at
i
v
e co
l
o
r
f
eat
u
r
es
ar
e
e
xt
r
a
c
t
e
d
by
u
s
i
ng t
he
m
e
t
hod
of
c
ol
o
r
p
r
ot
ot
y
pe
de
ns
i
t
y
e
s
t
im
a
t
i
on i
n r
c
e
t
r
a
i
ni
ng
pr
oc
e
du
r
e
.
I
n
a
pa
pe
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
252
-
88
14
IJ
A
A
S
V
o
l
.
7
,
N
o
.
2
,
J
un
e
201
8
:
156
–
1
61
1
58
pu
bl
i
s
he
d
i
n
2
00
9
a
ut
ho
r
s
l
i
wi
c
ki
,
s
t
e
p
ha
n,
a
nd
m
a
r
k
e
ve
r
i
ng
ha
m
[
8
]
c
l
a
s
s
i
f
i
e
s
pi
xe
l
s
a
s
ha
n
d
or
no
n
-
ha
n
d
by
a
c
om
bi
na
t
i
on of
t
h
r
e
e
c
om
pone
nt
s
:
a
s
i
gne
r
-
s
p
e
c
i
f
i
c
s
k
i
n
c
o
l
o
r
m
o
d
e
l
,
a
s
p
a
t
i
a
l
l
y
-
v
a
r
yin
g
‘
non
-
s
k
in’
c
o
l
or
m
ode
l
,
a
nd a
s
pa
t
i
a
l
c
ohe
r
e
nc
e
p
r
i
o
r
.
E
m
i
l
m
.
P
e
t
r
i
u
e
t
a
l
.
[
9
]
h
a
a
r
l
i
k
e
f
e
a
t
u
r
e
s
t
o
t
e
l
l
t
h
e
r
a
t
i
o
b
e
t
w
e
e
n
d
a
r
k
a
nd
b
r
i
g
ht
a
r
e
a
s
,
c
a
n
ope
r
a
t
e
m
uc
h
f
a
s
t
e
r
t
h
a
n
pi
xe
l
ba
s
e
d s
y
s
t
e
m
.
I
t
a
l
s
o
r
o
b
us
t
s
n
oi
s
e
a
nd
l
i
ght
i
n
g
c
ha
nge
s
.
H
o
we
ve
r
,
i
t
i
s
di
f
f
i
c
ul
t
t
o s
e
p
a
r
a
t
e
o
ut
h
a
n
d
f
r
om
t
he
im
a
ge
be
c
a
us
e
of
m
i
xt
ur
e
o
f
s
ki
n c
ol
o
r
i
n t
ha
t
s
t
i
l
l
s
o
m
e
a
u
t
h
o
r
s
r
e
s
e
a
r
c
he
d a
bo
ut
t
ha
t
.
I
n
[
10
]
pe
r
f
or
m
i
ng
r
e
gi
o
n
-
ba
s
e
d s
e
gm
e
nt
a
t
i
on
of
t
he
ha
nd
,
e
l
i
m
i
n
a
t
i
n
g
s
m
a
l
l
f
a
l
s
e
-
al
ar
m
r
eg
i
o
n
s
t
h
a
t
w
er
e d
ecl
ar
e
d
as
“h
a
n
d
-
l
i
k
e
,
”
b
a
s
e
d
o
n
t
h
e
i
r
c
o
l
o
r
s
t
a
t
i
s
t
i
c
s
.
F
i
ndi
ng t
he
c
e
nt
e
r
of
gr
a
vi
t
y
(
c
o
g)
o
f
t
he
ha
nd
r
e
gi
o
n a
s
w
e
l
l
a
s
t
he
f
a
r
t
h
e
s
t
di
s
t
a
nc
e
i
n
t
he
ha
n
d r
e
gi
o
n f
r
om
t
he
c
og
.
I
n
on
e
of
t
he
pa
pe
r
s
w
i
t
h t
he
he
l
p
of
ki
ne
t
i
c
s
e
ns
or
s
w
e
c
a
n a
c
qui
r
e
c
ol
o
ur
i
m
a
ge
i
nput
a
n
d de
pt
h
m
a
p i
np
ut
a
s
s
a
i
d i
n [
1
1
]
.
W
i
t
h t
he
he
l
p o
f
bot
h,
ha
n
d c
a
n
be
s
e
pa
r
a
t
e
d
o
ut
.
C
ol
o
u
r
i
m
a
ge
i
n
put
w
i
l
l
he
l
p t
o
l
o
c
a
t
e
t
h
e
h
a
n
d
w
h
i
l
e
d
e
p
t
h
i
m
a
g
e
i
n
p
u
t
w
i
l
l
d
i
s
t
i
n
g
u
i
s
h
ha
nd
f
r
om
ot
he
r
ob
je
c
t
s
i
n
s
c
r
e
e
n
r
e
m
ovi
n
g
pr
obl
e
m
f
r
om
c
l
ut
t
e
r
e
d ba
c
k
g
r
o
u
nds
o
r
ot
he
r
bo
dy
p
a
r
t
s
m
i
xi
ng wi
t
h ha
n
ds
.
S
om
e
a
ut
ho
r
s
di
s
c
us
s
e
d
i
n
t
he
i
r
pa
pe
r
s
a
nd
t
r
i
e
d
t
o
s
ol
ve
i
l
l
um
i
na
t
i
on
pr
o
bl
e
m
by
a
ddi
n
g
a
n
ot
he
r
l
e
ve
l
t
o
ha
n
d
de
t
e
c
t
i
on
but
[
1
2
]
doe
s
n
ot
s
o
w
e
ca
n
s
a
y
t
ha
t
t
hi
s
w
i
l
l
be
t
he
l
e
a
s
t e
f
f
e
c
t
i
ve
m
e
tho
d.
I
t
us
e
s
ht
s
a
l
go
r
i
t
hm
f
or
ha
n
d t
r
a
c
ki
n
g
a
nd s
e
gm
e
nt
a
t
i
on.
I
t
us
e
s
e
d
ge
de
t
e
c
t
i
on
t
e
c
hni
que
t
o
i
t
.
A
n
um
be
r
o
f
a
ut
h
or
s
c
o
nd
uc
t
e
d
t
he
i
r
r
e
s
e
a
r
c
h
ba
s
e
d
on
t
he
f
r
a
m
e
s
.
I
n
[
1
3
]
a
t
r
a
c
k
i
n
g
s
y
s
t
e
m
i
s
u
s
ed
t
o o
bt
a
i
n
he
a
d
a
nd
ha
n
d
p
os
i
t
i
ons
,
t
he
s
e
c
o
-
or
di
na
t
e
s
a
r
e
t
he
n
c
l
us
t
e
r
e
d
t
o
obt
a
i
n a
q
ua
nt
i
z
e
d
de
s
c
r
i
pt
i
on
f
o
r
each
f
r
am
e w
h
i
ch
ar
e t
h
en
t
em
p
o
r
al
l
y
co
n
c
at
en
at
ed
t
o
cr
e
at
e b
i
f
r
am
e f
ea
t
u
r
es
.
I
n
2
0
1
3
o
n
e
o
f
t
h
e
a
u
t
h
or
in
h
er
t
ech
n
i
q
u
e
s
ai
d
[
1
4
]
t
h
a
t
i
t
u
s
e
s
vi
de
o i
np
ut
f
r
om
m
i
c
r
os
of
t
ki
ne
t
i
c
3ge
a
r
t
e
c
h
nol
o
gy
w
hi
c
h s
e
n
s
e
t
he
m
o
t
i
o
n
o
f
t
h
e
h
an
d
i
n
a s
k
el
et
al
f
r
a
m
es
an
d
g
i
v
e t
h
e
v
al
u
e o
f
co
o
r
d
i
n
at
es
i
n
each
f
r
am
e.
T
h
e p
r
o
p
o
s
ed
m
e
t
hod
t
a
ke
s
i
nt
o
a
c
c
o
unt
bo
t
h
t
he
l
oc
a
l
a
n
d
gl
oba
l
f
e
a
t
u
r
e
s
as
s
o
ci
at
ed
w
i
t
h
a
g
es
t
u
re
.
In
[
15
]
f
o
r
r
e
c
o
gn
ition
of
dy
na
m
i
c
ge
s
t
ur
e
s
i
m
pl
i
e
s
t
he
a
na
l
y
s
i
s
of
t
e
m
por
a
l
s
e
q
u
e
nc
e
s
o
f
f
r
a
m
e
s
,
w
hi
c
h c
o
nt
a
i
n t
he
va
r
i
ous
pos
e
s
a
s
s
um
e
d
by
t
h
e
ha
nd
d
u
r
i
n
g
i
t
s
m
ove
m
e
nt
.
S
t
e
p
by
S
t
e
p
P
r
oc
e
s
s
of
P
r
o
p
os
e
d
W
o
r
k
a
s
s
h
own
in
F
igur
e
1
.
F
i
gu
r
e
1.
S
te
p
b
y
s
t
e
p
p
r
oc
e
s
s
o
f
pr
op
os
e
d
w
or
k
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
A
A
S
I
S
S
N
:
225
2
-
88
14
V
i
s
i
on
B
as
e
d
A
ppr
o
ac
h
t
o
Si
g
n
L
a
ng
u
age
R
e
c
og
ni
t
i
on
(
H
i
m
ans
h
u
Gu
pt
a
)
1
59
3.
P
R
O
P
O
SE
D
W
O
R
K
W
e
a
s
k
us
e
r
t
o
u
pl
oa
d
t
he
vi
d
e
o
of
hi
s
ha
n
d
w
he
r
e
he
i
s
us
i
ng
s
i
g
n
l
a
ng
ua
ge
t
o
c
om
m
unic
a
t
e
.
Vi
de
o
i
s
co
l
l
ect
i
o
n
o
f
f
r
am
es
t
h
at
ar
e p
l
ay
ed
t
o
g
et
h
er
.
I
f
w
e
d
i
v
i
d
e t
h
e
v
i
d
e
o
i
n
t
o
f
r
am
es
an
d
co
n
s
i
d
er
each
f
r
am
e
one
by
o
ne
.
W
e
pr
oc
e
s
s
e
a
c
h
f
r
a
m
e
a
s
s
i
ngl
e
im
a
ge
a
nd m
ove
o
n t
o
ne
xt
f
r
a
m
e
t
il
l
t
he
e
nd
of
t
he
vi
d
e
o w
e
w
i
l
l
b
e
a
b
l
e
t
o
p
r
o
ces
s
each
f
r
a
m
e
i
n
v
i
d
eo
.
3.
1.
L
oc
al
i
s
i
n
g
H
an
d
R
e
gi
o
n
W
e
a
s
s
um
e
t
ha
t
t
he
us
e
r
ha
s
u
pl
oa
de
d t
he
vi
de
o i
n a
c
c
e
p
t
a
bl
e
f
or
m
a
t
a
nd
d
e
c
e
n
t
qu
a
lity.
T
h
e
n our
f
i
r
s
t
t
a
s
k i
s
t
o s
e
gm
e
nt
out
t
h
e
ha
n
d i
n t
he
vi
de
o f
r
om
t
he
ba
c
k
gr
o
un
d.
W
e
a
c
hi
e
ve
t
h
i
s
r
e
s
ul
t
i
n t
w
o s
t
e
ps
.
F
i
r
s
t
,
we
f
i
n
d
t
he
pi
xe
l
s
t
ha
t
b
e
l
on
g
t
o
t
he
ha
nd
r
e
g
i
o
n
.
R
e
fe
r
F
i
g
ur
e
1
.
I
t
i
s
be
e
n
o
bs
e
r
ve
d t
ha
t
YC
bC
r
c
ol
o
u
r
s
p
a
c
e
g
i
v
e
s
b
e
t
t
e
r
c
l
u
s
t
e
r
i
n
g
r
e
s
u
l
t
a
n
d
c
o
m
p
u
t
a
t
i
o
n
a
l
e
f
f
i
c
i
e
nc
y
[
1
6]
.
S
o
we
c
on
ve
r
t
ou
r
i
m
a
ge
i
nto
Y
C
bC
r
c
ol
o
u
r
s
pa
c
e
(re
fe
r
t
o
E
q
.
1
).
I
n
Y
C
b
C
r
c
o
l
o
u
r
s
p
a
c
e
,
Y
i
s
l
um
i
na
nc
e
,
C
b
a
n
d
C
r
a
r
e
c
hr
om
a
t
i
c
i
t
y
o
f
bl
u
e
a
n
d
r
e
d
c
ol
o
ur
.
C
b
a
nd
c
r
a
r
e
t
w
o
di
m
e
ns
i
o
n
a
l
i
n
d
e
p
e
nde
nt
.
D
e
m
ons
t
r
a
t
e
d
i
n
F
i
g
ur
e
2
t
he
c
on
ve
r
t
e
d
R
G
B
c
ol
o
ur
s
pa
c
e
t
o
Y
C
bC
r
c
ol
o
ur
s
pa
c
e
:
=
0
.
2
9
9
0
0
.
5
8
7
0
0
.
1
4
4
0
−
0
.
1
6
8
7
−
0
.
3
3
1
3
0
.
5
0
0
0
0
.
5
0
0
0
−
0
.
4
1
8
7
−
0
.
0
8
1
3
+
0
1
2
8
1
2
8
(
1)
A
f
t
er
co
n
v
er
t
i
n
g
t
h
e i
m
ag
e w
e e
x
t
r
act
t
h
e
s
k
i
n
r
e
g
i
o
n
i
n
t
h
e i
m
ag
e.
F
o
r
o
u
r
s
y
s
t
em
t
o
accep
t
p
e
o
p
l
e
w
i
t
h
di
f
f
e
r
e
nt
s
ki
n
t
one
s
we
a
ppl
y
s
ki
n
t
hr
e
s
hol
d
t
o
s
e
pa
r
a
t
e
o
ut
t
he
s
ki
n
[
17]
.
8
0≤
C
b≤
1
20
a
n
d
13
3≤
C
r
≤
17
3
F
i
g
ur
e
2
.
Co
n
v
e
r
t
e
d
i
m
a
g
e
f
r
o
m
RG
B
t
o
Y
C
b
Cr
c
o
l
o
r
s
p
a
c
e
3.
2.
F
i
n
d
i
n
g
C
e
n
t
r
o
i
d
A
f
t
e
r
we
f
o
un
d out
t
he
s
ki
n c
ol
o
ur
e
d
r
eg
i
o
n
s
i
n
o
u
r
i
m
ag
e w
e cal
c
u
l
at
e cen
t
r
o
i
d
,
cen
t
r
e o
f
g
r
a
v
i
t
y
(
CO
G
)
b
y
u
s
i
n
g
f
o
l
l
o
w
i
n
g
f
o
r
m
u
l
a
̅
=
∑
=
0
=
∑
=
0
(
2)
Wh
e
re
x
j
an
d
y
j
ar
e
x
an
d
y
co
o
r
d
i
n
at
es
o
f
t
h
e
j
th
pi
xe
l
i
n
t
h
e
ha
n
d
a
r
e
a
,
a
n
d
n
de
n
ot
e
s
t
he
num
be
r
o
f
pi
x
e
l
s
i
n
t
h
e
ar
ea
.
A
f
t
e
r
w
e
o
b
t
a
i
n
c
e
n
t
r
o
i
d
(
r
e
f
e
r
t
o E
q.
2
)
we
f
i
n
d t
he
di
s
t
a
nc
e
f
r
om
t
h
e
m
os
t
e
xt
r
e
m
e
poi
nt
i
n t
he
ha
n
d t
o t
h
e
c
e
nt
r
e
,
n
or
m
a
l
l
y
t
hi
s
f
a
r
t
he
s
t
di
s
t
a
nc
e
i
s
t
he
di
s
t
a
nc
e
f
r
om
t
he
c
e
nt
r
oi
d
t
o t
i
p
of
t
he
l
on
ge
s
t
a
c
t
i
ve
f
i
n
g
e
r
i
n
t
h
e
p
a
r
t
i
c
u
l
a
r
g
e
s
t
u
r
e
[
4
].
W
e t
h
e
n
d
r
aw
a ci
r
cl
e o
f
w
h
o
s
e
r
a
d
i
u
s
i
s
f
a
r
t
h
e
s
t
d
i
s
t
a
n
c
e
f
r
o
m
c
e
n
t
r
o
i
d
.
T
h
i
s
c
i
r
c
l
e
c
o
n
t
a
i
n
t
h
e
w
h
o
l
e
g
es
t
u
r
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
252
-
88
14
IJ
A
A
S
V
o
l
.
7
,
N
o
.
2
,
J
un
e
201
8
:
156
–
1
61
1
60
F
i
g
ur
e
.
3
.
E
x
p
ec
t
ed
i
m
ag
es
af
t
er
c
o
n
v
er
s
i
o
n
3
.
3
.
G
e
s
tu
r
e
Cl
a
s
s
i
fi
c
a
ti
o
n
N
o
w
w
e
ha
ve
t
o
i
de
nt
i
f
y
wha
t
t
he
ge
s
t
u
r
e
m
e
a
ns
.
F
o
r
t
hi
s
w
e
ha
ve
us
e
d
t
e
m
pl
a
t
e
m
a
t
c
hing
d
ue
t
o
i
t
s
s
i
m
p
l
i
c
i
t
y
a
n
d
p
r
e
d
i
c
t
a
b
l
e
r
e
s
p
o
n
s
e
w
i
t
h
a
l
i
m
i
t
e
d
t
r
a
i
n
i
n
g
s
e
t
.
T
e
m
p
l
a
t
e
m
a
t
c
h
i
n
g
i
s
b
a
s
i
c
a
l
l
y
t
h
e
t
w
o
di
m
e
ns
i
ona
l
c
r
os
s
-
c
o
r
r
el
at
i
o
n
o
f
a
g
r
ay
s
cal
e
i
m
ag
e w
i
t
h
a
g
r
ay
s
cal
e t
em
p
l
at
e,
h
en
ce
es
t
i
m
at
i
n
g
t
h
e
d
eg
r
ee o
f
s
i
m
i
l
ar
i
t
y
b
et
w
een
t
h
e t
w
o
.
W
e co
m
p
ar
e
t
he
u
nk
n
ow
n g
e
s
t
ur
e
w
i
t
h t
he
pr
e
s
e
t
t
e
m
pl
a
te
m
ode
l
s
of
i
n
di
vi
d
ua
l
g
es
t
u
r
es
.
4.
C
O
N
CL
U
S
I
O
N
D
e
a
f
a
n
d
D
um
b
pe
opl
e
r
e
l
y
o
n
s
i
gn
l
a
n
g
ua
g
e
i
nt
e
r
pr
e
t
e
r
s
f
or
c
om
m
uni
c
a
t
i
on.
Ho
w
e
ve
r
,
t
he
y
c
a
nn
ot
de
pe
n
d o
n i
nt
e
r
p
r
e
t
e
r
s
e
ve
r
y
da
y
i
n l
i
f
e
m
a
i
nl
y
due
t
o t
he
hi
g
h c
os
t
s
a
nd t
he
di
f
f
i
c
ul
t
y
i
n f
i
ndi
ng a
n
d
s
c
h
e
d
u
l
i
n
g
q
u
a
l
i
f
i
e
d
i
n
t
e
r
p
r
e
t
e
r
s
.
T
h
i
s
s
y
s
t
e
m
w
i
l
l
h
e
l
p
d
i
s
a
b
l
e
d
p
e
r
s
o
n
s
i
n
i
m
p
r
o
v
i
n
g
t
h
e
i
r
q
u
a
l
i
t
y
o
f
l
i
f
e
s
i
g
n
i
f
i
c
a
n
t
l
y
.
T
hi
s
pa
pe
r
p
r
o
pos
e
s
a
ha
n
d
ge
s
t
u
r
e
r
e
c
o
gn
i
t
i
on m
e
t
ho
d b
a
s
e
d
on
Y
C
bC
r
c
ol
ou
r
s
p
a
c
e
,
C
OG
a
n
d
t
e
m
p
l
a
t
e
m
a
t
c
hi
n
g.
T
he
p
r
op
os
e
d m
e
t
hod c
a
n de
t
e
c
t
ha
n
d
ge
s
t
u
r
e
f
r
om
ba
c
k
g
r
o
u
nd
.
I
t
a
c
c
e
pt
s
di
f
f
e
r
e
nt
s
ki
n
t
one
s
a
n
d di
f
f
e
r
e
nt
i
l
l
um
i
na
t
ion c
o
ndi
t
i
o
ns
.
T
hi
s
m
e
t
hod
u
s
e
s
c
h
r
om
a
t
i
c
ity
i
n YC
bC
r
c
o
l
ou
r
s
pa
c
e
,
a
vo
i
di
n
g
t
h
e ef
f
ect
o
n
h
an
d
g
e
s
t
u
r
e a
n
d
u
s
es
C
O
G
t
o
s
eg
m
en
t
t
he
h
a
nd
.
I
t
t
he
n r
e
c
og
ni
s
e
s
t
he
ge
s
t
ur
e
wi
t
h
t
he
h
e
l
p
o
f
t
e
m
pl
a
t
e
m
a
t
c
hi
n
g.
T
hi
s
m
e
tho
d i
s
r
ob
us
t
t
o
di
f
f
e
r
e
nt
ba
c
kg
r
ou
n
ds
a
nd
pr
o
vi
de
s
be
t
t
e
r
r
e
s
ul
t
.
T
hi
s
m
e
t
ho
d
w
i
l
l
be
a
g
oo
d
a
ddi
t
i
o
n
t
o
t
he
on
g
oi
n
g
r
e
s
e
a
r
c
h
i
n
t
he
f
i
e
l
d
of
H
um
a
n
C
o
m
put
e
r
I
nt
e
r
f
a
c
e
(
H
C
I
).
A
C
K
N
O
W
L
E
D
G
E
ME
NT
S
W
e
w
o
ul
d l
i
ke
t
o t
ha
n
k a
l
l
t
h
e
r
e
s
e
a
r
c
he
r
s
wo
r
ki
n
g
on
t
hi
s
f
i
e
l
d
w
h
o i
n
one
w
a
y
o
r
a
n
ot
he
r
gui
de
d
us
on
a
c
hi
e
vi
n
g
ou
r
goa
l
s
.
W
e
w
oul
d
a
l
s
o
l
i
ke
t
o
e
x
pr
e
s
s
o
ur
a
p
pr
e
c
i
a
t
i
o
n
t
o
ou
r
gui
de
a
t
VI
T
Uni
ve
r
s
i
t
y
wh
o
w
a
s
ki
n
d
e
n
ou
gh
t
o
s
ha
r
e
he
r
vi
e
w
s
wi
t
h
us
a
nd
o
f
f
e
r
e
d
s
o
m
e
s
ug
ge
s
t
i
o
ns
i
n
m
a
ki
ng
t
hi
s
p
r
o
je
c
t
a
s
uc
c
e
s
s
R
EF
ER
E
N
C
ES
[1]
S
t
ar
n
er
, T
h
ad
, J
o
s
h
u
a W
eav
er
, a
n
d
A
l
ex
P
en
t
l
an
d
. "
R
eal
-
t
i
m
e
a
m
e
ri
c
a
n s
i
gn l
a
n
gua
ge
re
c
ogn
i
t
i
o
n us
i
ng de
s
k a
nd
w
ear
ab
l
e co
m
p
u
t
er
b
as
ed
v
i
d
eo
.
"
IE
E
E
T
ra
ns
a
c
t
i
ons
on P
a
t
t
e
rn
A
na
l
y
s
i
s
a
nd M
a
c
hi
n
e
Int
e
l
l
i
ge
n
c
e
20
.
12 (1998)
:
1371
-
1375.
[2]
T
an
i
b
at
a
,
N
obuhi
ko,
N
obut
a
k
a
S
hi
m
a
da
,
a
nd Y
o
s
hi
a
ki
S
hi
ra
i
.
"
E
xt
ra
c
t
i
on of ha
n
d fe
a
t
ur
e
s
for re
c
ogni
t
i
on of s
i
g
n
l
a
ngua
g
e
w
ords
.
"
Int
e
r
nat
i
ona
l
c
onf
e
r
e
nc
e
on
v
i
s
i
on i
nt
e
r
f
ac
e
.
20
02
.
[3]
Coope
r
,
H
e
l
e
n,
a
nd Ri
c
h
a
rd Bo
w
de
n.
"
L
e
a
rni
n
g s
i
gns
from
s
ubt
i
t
l
e
s
:
A
w
e
a
kl
y
s
upe
rvi
s
e
d
a
pproa
c
h
t
o s
i
g
n
l
a
ngua
g
e
r
e
c
ogn
i
t
i
on.
"
Com
put
e
r
V
i
s
i
on
and P
a
t
t
e
r
n R
e
c
ogni
t
i
on,
2009.
C
V
P
R
20
09.
IE
E
E
Conf
e
r
e
nc
e
on
. I
E
E
E
,
2009.
[4]
M
a
l
i
m
a
,
A
s
a
nt
e
ra
bi
K
i
ghom
a
,
E
rol
Ö
z
gür
,
a
nd
M
üj
da
t
Çe
t
i
n.
"
A
fa
s
t
a
l
gori
t
h
m
for vi
s
i
on
-
ba
s
e
d ha
nd ge
s
t
ur
e
re
c
ogni
t
i
on fo
r r
obot
c
on
t
rol
.
"
(2
006):
1
-
4.
[5]
C
h
en
, Q
i
n
g
, N
i
c
o
l
as
D
. G
eo
r
g
an
as
, an
d
E
m
i
l
M
. P
et
r
i
u
. "
R
eal
-
ti
m
e
v
is
io
n
-
ba
s
e
d ha
nd ge
s
t
ure
re
c
ogni
t
i
on us
i
n
g
h
aar
-
l
i
k
e
f
ea
t
u
r
es
.
"
Ins
t
rum
e
nt
a
t
i
o
n a
nd M
e
a
s
ure
m
e
nt
T
e
c
hno
l
og
y
Confe
r
e
nc
e
P
roc
e
e
d
i
ngs
,
200
7.
IM
T
C 2007.
I
E
E
E
. I
E
E
E
, 2
0
0
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
A
A
S
I
S
S
N
:
225
2
-
88
14
V
i
s
i
on
B
as
e
d
A
ppr
o
ac
h
t
o
Si
g
n
L
a
ng
u
age
R
e
c
og
ni
t
i
on
(
H
i
m
ans
h
u
Gu
pt
a
)
1
61
[6]
D
ong
,
G
uo,
Y
onghua
Y
a
n,
a
nd M
.
X
i
e
.
"
V
i
s
i
on
-
ba
s
e
d
ha
nd g
e
s
t
ure
re
c
ogn
i
t
i
on
f
or hum
a
n
-
v
eh
i
cl
e
i
n
t
er
act
i
o
n
."
P
r
oc
.
o
f
t
he
Int
e
r
nat
i
onal
c
onf
e
r
e
n
c
e
on Cont
r
ol
,
A
ut
om
at
i
on
and Co
m
put
e
r
V
i
s
i
on
.
V
ol
.
1
.
1998
.
[7]
Re
n,
Z
hou,
J
i
ng
j
i
ng M
e
ng,
a
nd J
uns
ong Y
ua
n.
"
D
e
pt
h c
a
m
e
ra
b
a
s
e
d ha
nd
ge
s
t
ur
e
re
c
ogni
t
i
on a
n
d i
t
s
a
pp
l
i
c
a
t
i
on
s
i
n hum
a
n
-
c
om
put
e
r
-
in
te
r
a
c
tio
n
.
"
Inf
or
m
at
i
on
,
Com
m
uni
c
at
i
ons
and Si
gnal
P
r
oc
e
s
s
i
ng (
IC
ICS)
2011
8t
h
Int
e
r
nat
i
ona
l
Co
nf
e
r
e
nc
e
on
. I
E
E
E
, 2
0
1
1
.
[8]
G
a
rg,
P
ra
ga
t
i
,
N
a
ve
e
n A
gg
a
rw
a
l
,
a
nd S
a
n
j
e
e
v
S
ofa
t
.
"
V
i
s
i
on b
a
s
e
d ha
nd ge
s
t
ur
e
re
c
o
gni
t
i
on
.
"
Wor
l
d A
c
ade
m
y
of
Sc
i
e
n
c
e
,
E
ng
i
ne
e
r
i
ng and T
e
c
hn
ol
ogy
49
.
1 (2009
):
972
-
977.
[9]
P
ort
a
,
Ma
r
c
o
.
"
Vi
si
o
n
-
ba
s
e
d us
e
r i
n
t
e
rf
a
c
e
s
:
m
e
t
hods
a
nd
a
ppl
i
c
a
t
i
ons
.
"
Int
e
r
nat
i
onal
J
our
nal
of
H
um
an
-
Com
put
e
r
St
udi
e
s
57
.
1 (20
02):
27
-
73.
[10]
G
eet
h
a
,
M
.
,
e
t
a
l
.
"
A
vi
s
i
on
ba
s
e
d d
y
n
a
m
i
c
ge
s
t
ure
re
c
ogni
t
i
o
n of i
ndi
a
n s
i
gn
l
a
ngua
g
e
on k
i
ne
c
t
b
a
s
e
d d
e
pt
h
i
m
ag
es
."
E
m
e
r
gi
ng T
r
e
nds
i
n Com
m
uni
c
at
i
on,
C
ont
r
ol
,
Si
gnal
P
r
oc
e
s
s
i
ng
&
Co
m
put
i
ng A
ppl
i
c
a
t
i
ons
(
C2SP
CA
)
,
2013 Int
e
r
nat
i
on
al
Conf
e
r
e
nc
e
o
n
.
IE
E
E
,
2013
.
[11]
M
urt
h
y
,
G
.
R.
S
.
,
a
nd R.
S
.
J
a
do
n.
"
A
re
vi
e
w
of
vi
s
i
on ba
s
e
d ha
nd ge
s
t
ure
s
re
c
og
ni
t
i
on.
"
Int
e
r
nat
i
onal
J
our
nal
of
Inf
or
m
at
i
on T
e
c
hnol
ogy
and
Kn
owl
e
dge
Manag
e
m
e
nt
2.
2
(2009)
:
405
-
410.
[12]
A
k
y
ol
,
S
ua
t
,
a
n
d U
l
ri
c
h
Ca
n
z
l
e
r.
"
A
n i
nform
a
t
i
on t
e
rm
i
na
l
us
i
n
g vi
s
i
on b
a
s
e
d s
i
gn l
a
ngua
ge
re
c
ogni
t
i
on
.
"
I
T
EA
Wor
k
s
hop on V
i
r
t
ual
H
om
e
E
n
v
i
r
onm
e
nt
s
,
V
H
E
Mi
ddl
e
war
e
Con
s
or
t
i
um
.
V
ol
.
12.
2002.
[13]
G
hot
ka
r,
A
rc
ha
na
S
.
,
a
nd G
a
j
a
na
n K
.
K
ha
ra
t
e
.
"
H
a
nd s
e
gm
e
nt
a
t
i
on t
e
c
hni
qu
e
s
t
o ha
nd ge
s
t
ure
re
c
ogni
t
i
on fo
r
na
t
ura
l
hum
a
n
c
om
put
e
r i
nt
e
ra
c
t
i
on.
"
Int
e
r
nat
i
on
al
J
our
nal
of
H
um
an Com
put
e
r
Int
e
r
ac
t
i
on (
IJ
H
CI)
3.
1 (2012):
15.
[14]
Cui
,
Y
unt
a
o
,
a
nd J
u
y
a
ng
W
e
ng.
"
A
ppe
a
ra
n
c
e
-
b
a
s
e
d
ha
nd s
i
gn re
c
ogni
t
i
on from
i
nt
e
ns
i
t
y
i
m
ag
e
s
eq
u
en
ces
."
Co
m
put
e
r
V
i
s
i
on
a
nd Im
age
U
nde
r
s
t
andi
ng
78.
2
(2
000):
157
-
176.
[15]
L
i
w
i
c
ki
,
S
t
e
pha
n,
a
nd M
a
rk E
ve
ri
ngh
a
m
.
"
A
ut
om
a
t
i
c
re
c
o
gni
t
i
on
of fi
ng
e
rs
pe
l
l
e
d w
ords
i
n bri
t
i
s
h s
i
gn
l
an
g
u
ag
e."
Com
put
e
r
V
i
s
i
on an
d P
at
t
e
r
n R
e
c
og
ni
t
i
on Wor
k
s
hop
s
,
2009.
CV
P
R
Wor
k
s
hops
2009
.
IE
E
E
Com
put
e
r
Soc
i
e
t
y
Con
f
e
r
e
nc
e
on
.
I
E
E
E
,
20
09.
[16]
Qi
u
-
y
u
,
Z
ha
ng
,
e
t
a
l
.
"
H
a
nd ge
s
t
ure
s
e
gm
e
nt
a
t
i
o
n m
e
t
hod ba
s
e
d on Y
CbCr c
ol
or s
pa
c
e
a
nd k
-
m
e
a
ns
c
l
us
t
e
r
i
ng.
"
Int
e
rna
t
i
ona
l
J
ou
rna
l
of
S
i
gna
l
P
r
oc
e
s
s
i
ng,
Im
a
g
e
P
roc
e
s
s
i
ng a
nd
P
a
t
t
e
rn
Re
c
ogn
i
t
i
o
n 8.
5
(2015):
1
05
-
116.
[17]
Ba
s
i
l
i
o,
J
orge
A
l
be
rt
o M
a
rc
i
a
l
,
e
t
a
l
.
"
E
xp
l
i
c
i
t
i
m
a
ge
d
e
t
e
c
t
i
on
us
i
ng Y
CbCr s
pa
c
e
c
ol
or
m
ode
l
a
s
s
ki
n de
t
e
c
t
i
on.
"
A
ppl
i
c
a
t
i
ons
of
M
a
t
he
m
a
t
i
c
s
a
n
d Com
put
e
r E
ng
i
ne
e
r
i
ng (2011)
:
123
-
128.
Evaluation Warning : The document was created with Spire.PDF for Python.