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.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
,
p
p
.
3
8
7
9
~
3
8
9
1
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
1
3
.
i4
.
p
p
3
8
7
9
-
3
8
9
1
3879
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Desig
n of an e
ff
e
c
tive multi
ple
o
bj
e
cts
tracki
ng
f
ra
m
ewo
rk f
o
r
dy
na
mic v
ideo
scenes
Su
nil
K
um
a
r
K
a
ra
na
m
1
,
N
a
ra
s
im
ha
M
urt
hy
P
o
k
a
le
K
a
v
y
a
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
&
En
g
i
n
e
e
r
i
n
g
,
B
M
S
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
A
f
f
i
l
i
a
t
e
d
t
o
V
i
sv
e
sv
a
r
a
y
a
Te
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
si
t
y
,
B
e
l
a
g
a
v
i
,
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
&
En
g
i
n
e
e
r
i
n
g
,
R
N
S
I
T
I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
,
A
f
f
i
l
i
a
t
e
d
to
V
i
s
v
e
sv
a
r
a
y
a
Te
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
s
i
t
y
,
B
e
l
a
g
a
v
i
,
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
an
3
0
,
2
0
2
4
R
ev
is
ed
Feb
2
4
,
2
0
2
4
Acc
ep
ted
Ma
r
2
1
,
2
0
2
4
No
wa
d
a
y
s,
t
h
e
a
p
p
l
ica
ti
o
n
s
c
o
rre
sp
o
n
d
in
g
t
o
v
id
e
o
su
r
v
e
il
lan
c
e
sy
ste
m
s
a
re
g
e
tt
in
g
p
o
p
u
lar
d
u
e
to
t
h
e
ir
wid
e
ra
n
g
e
o
f
d
e
p
lo
y
m
e
n
t
in
v
a
ri
o
u
s
p
l
a
c
e
s
su
c
h
a
s
sc
h
o
o
ls,
ro
a
d
s,
a
n
d
a
irp
o
r
ts.
De
sp
it
e
th
e
c
o
n
ti
n
u
o
u
s
e
v
o
l
u
ti
o
n
a
n
d
in
c
re
a
sin
g
d
e
p
lo
y
m
e
n
t
o
f
o
b
jec
t
-
trac
k
in
g
fe
a
tu
re
s
in
v
i
d
e
o
s
u
rv
e
il
lan
c
e
a
p
p
li
c
a
ti
o
n
s,
t
h
e
l
o
o
p
h
o
les
stil
l
n
e
e
d
t
o
b
e
so
l
v
e
d
d
u
e
t
o
t
h
e
li
m
it
e
d
fu
n
c
ti
o
n
a
li
ti
e
s
o
f
v
i
d
e
o
-
trac
k
i
n
g
sy
ste
m
s.
T
h
e
e
x
isti
n
g
v
i
d
e
o
su
rv
e
il
lan
c
e
sy
ste
m
s
p
o
se
h
i
g
h
p
ro
c
e
ss
in
g
o
v
e
rh
e
a
d
d
u
e
t
o
t
h
e
larg
e
r
siz
e
o
f
v
id
e
o
fil
e
s.
Ho
we
v
e
r,
t
h
e
tra
d
it
i
o
n
a
l
li
tera
tu
r
e
re
p
o
rt
q
u
it
e
so
p
h
isti
c
a
ted
sc
h
e
m
e
s
wh
ich
m
ig
h
t
s
u
c
c
e
ss
fu
ll
y
re
tai
n
h
ig
h
e
r
o
b
jec
t
d
e
tec
ti
o
n
a
c
c
u
ra
c
y
fr
o
m
th
e
v
id
e
o
sc
e
n
e
s
b
u
t
n
e
e
d
s
m
o
re
e
ffe
c
ti
v
e
n
e
ss
re
g
a
rd
in
g
c
o
m
p
u
tati
o
n
a
l
c
o
m
p
lex
it
y
u
n
d
e
r
li
m
it
e
d
c
o
m
p
u
ti
n
g
re
so
u
rc
e
s.
Th
e
stu
d
y
t
h
e
re
b
y
i
d
e
n
ti
fies
th
e
sc
o
p
e
o
f
e
n
h
a
n
c
e
m
e
n
t
in
tra
d
it
i
o
n
a
l
o
b
jec
t
-
trac
k
in
g
f
u
n
c
ti
o
n
s.
F
u
r
th
e
r,
it
in
t
ro
d
u
c
e
s
a
n
o
v
e
l
,
c
o
st
-
e
ffe
c
ti
v
e
trac
k
in
g
m
o
d
e
l
b
a
se
d
o
n
G
a
u
ss
ian
m
ix
t
u
re
m
o
d
e
l
(G
M
M
)
a
n
d
Ka
lma
n
fil
ter
(KF)
th
a
t
c
a
n
a
c
c
u
ra
tely
id
e
n
ti
f
y
n
u
m
e
ro
u
s
m
o
b
il
e
o
b
jec
ts
fro
m
a
d
y
n
a
m
ic
v
i
d
e
o
sc
e
n
e
a
n
d
e
n
s
u
re
s
c
o
m
p
u
ti
n
g
e
ffi
c
ien
c
y
.
T
h
e
stu
d
y
'
s
o
u
tc
o
m
e
sh
o
ws
t
h
a
t
t
h
e
p
ro
p
o
se
d
stra
teg
ic
m
o
d
e
ll
in
g
o
ffe
rs
b
e
tt
e
r
trac
k
in
g
p
e
rfo
rm
a
n
c
e
fo
r
d
y
n
a
m
ic
o
b
jec
ts
wit
h
c
o
s
t
-
e
ffe
c
ti
v
e
c
o
m
p
u
tati
o
n
c
o
m
p
a
re
d
to
th
e
p
o
p
u
lar b
a
se
li
n
e
a
p
p
ro
a
c
h
e
s.
K
ey
w
o
r
d
s
:
C
o
s
t e
v
alu
atio
n
Dy
n
am
ic
s
ce
n
e
I
n
ter
n
et
o
f
th
in
g
s
Mo
b
ile
o
b
ject
tr
ac
k
i
n
g
v
i
d
eo
s
u
r
v
eillan
ce
Ob
ject
d
etec
tio
n
ac
cu
r
ac
y
Pu
b
lic
s
af
ety
Secu
r
ity
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
:
Kar
an
am
Su
n
il Ku
m
a
r
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
,
B
MS
C
o
lleg
e
o
f
E
n
g
in
ee
r
in
g
B
u
ll T
em
p
le
R
d
,
B
asav
an
ag
u
d
i,
B
en
g
alu
r
u
,
Kar
n
atak
a
5
6
0
0
1
9
,
I
n
d
ia
E
m
ail: su
n
ilk
ar
an
am
r
esear
ch
2
0
2
0
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
g
r
o
wth
o
f
th
e
g
l
o
b
al
s
u
r
v
eillan
ce
m
ar
k
et
h
as
m
ad
e
d
y
n
am
ic
o
b
ject
d
etec
tio
n
an
d
tr
ac
k
in
g
f
r
o
m
v
id
eo
s
ce
n
es
p
o
p
u
lar
in
r
ec
e
n
t
y
ea
r
s
.
T
h
e
ad
v
a
n
ce
m
en
t
o
f
c
o
m
p
u
ter
v
is
io
n
tech
n
o
lo
g
y
an
d
im
ag
e
p
r
o
ce
s
s
in
g
m
ak
es
th
is
m
ar
k
et
s
ize
g
r
o
w
f
aster
.
T
h
e
p
r
im
e
r
ea
s
o
n
b
eh
in
d
its
r
ap
id
d
e
v
elo
p
m
en
t
is
u
r
b
an
izatio
n
co
n
s
tr
u
ctio
n
an
d
th
e
wid
e
r
a
n
g
e
o
f
d
e
p
lo
y
m
e
n
t
o
f
s
u
r
v
eillan
ce
s
y
s
tem
s
o
v
er
lar
g
e
b
u
il
d
in
g
s
,
p
u
b
lic
p
lace
s
,
p
ar
k
s
,
r
o
ad
s
,
an
d
air
p
o
r
ts
.
M
o
n
ito
r
in
g
an
d
s
u
r
v
eillan
ce
s
y
s
tem
s
p
lay
a
cr
u
cial
r
o
le
in
v
ar
io
u
s
asp
ec
ts
,
v
iz.
,
tr
af
f
ic
m
o
v
em
en
t
m
an
ag
e
m
en
t,
au
to
m
o
ti
v
e
s
af
ety
,
ac
tiv
ity
-
b
ased
r
ec
o
g
n
itio
n
f
o
r
cy
b
er
-
s
e
cu
r
ity
ap
p
licatio
n
s
,
an
d
s
p
o
r
ts
an
aly
s
is
[
1
]
,
[
2
]
.
Her
e
ar
is
e
th
e
r
eq
u
i
r
em
en
ts
o
f
r
eliab
le
an
d
ac
cu
r
ate
m
u
lt
ip
le
-
o
b
ject
tr
ac
k
in
g
(
MO
T
)
s
o
th
at
th
e
p
u
r
p
o
s
e
o
f
p
u
b
lic
s
af
ety
c
o
n
ce
r
n
s
ca
n
b
e
f
u
lf
illed
u
n
d
er
i
n
ter
co
n
n
ec
ted
s
m
ar
t
cities.
T
h
e
p
r
im
e
m
o
tiv
e
o
f
s
in
g
le
o
r
m
u
ltip
le
o
b
ject
tr
ac
k
in
g
(
MO
T
)
is
to
co
n
s
is
ten
tly
lo
ca
lize
an
d
id
en
tify
s
ev
e
r
al
o
b
jects
in
a
v
id
eo
s
eq
u
en
ce
w
h
ich
f
ac
ilit
ates
v
id
eo
an
aly
s
is
ap
p
licatio
n
s
o
f
v
id
e
o
s
u
r
v
eilla
n
ce
s
y
s
tem
s
.
Mo
s
t
co
n
v
en
tio
n
al
wo
r
k
s
o
n
MO
T
f
o
llo
w
th
e
id
ea
o
f
a
tr
ac
k
in
g
-
by
-
d
etec
tio
n
f
r
am
ewo
r
k
d
u
e
t
o
its
s
im
p
licity
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
3
8
7
9
-
3
8
9
1
3880
ef
f
ec
tiv
en
ess
in
f
u
lf
illi
n
g
tr
ac
k
in
g
r
eq
u
ir
em
en
ts
.
T
r
ad
iti
o
n
al
MO
T
tr
ac
k
in
g
co
n
s
is
ts
o
f
two
s
tag
es
o
f
o
p
er
atio
n
s
[
3
]
–
[
8
]
.
I
n
th
e
f
ir
s
t
s
tag
e
o
f
o
p
er
atio
n
s
,
th
e
f
r
am
ewo
r
k
em
p
lo
y
s
an
o
b
ject
d
etec
to
r
to
d
etec
t
o
b
jects
o
f
in
ter
est
in
th
e
cu
r
r
en
t
v
id
eo
f
r
am
e,
wh
er
ea
s
,
in
th
e
s
ec
o
n
d
s
tag
e
o
f
o
p
er
atio
n
s
,
th
e
d
e
tecte
d
o
b
jects
ar
e
ass
o
ciate
d
with
th
e
tr
ac
k
s
f
r
o
m
th
e
p
r
ev
io
u
s
f
r
am
es
to
c
o
n
s
tr
u
ct
th
e
tr
ajec
to
r
ies
f
u
r
th
er
.
Her
e
th
e
s
y
s
tem
ass
o
ciate
s
th
e
d
etec
ted
o
b
jects
b
etwe
en
f
r
am
es
u
s
in
g
f
ea
tu
r
es
th
at
co
u
ld
b
e
eith
er
lo
ca
tio
n
o
r
ap
p
ea
r
a
n
ce
[
9
]
–
[
1
1
]
.
T
h
e
r
ec
e
n
t
p
r
o
g
r
ess
in
tr
ac
k
in
g
-
by
-
d
etec
tio
n
s
tr
ateg
y
h
as
ev
o
lv
e
d
to
war
d
s
s
o
lv
in
g
th
e
am
b
ig
u
ities
ass
o
ciate
d
with
o
b
ject
d
etec
tio
n
.
I
t
ca
n
also
h
an
d
le
th
e
co
n
s
tr
ain
ts
th
at
r
esu
lt
in
o
b
ject
d
etec
tio
n
f
ailu
r
es
.
Ho
wev
er
,
o
b
ject
d
etec
tio
n
is
also
clo
s
ely
s
tu
d
ied
with
m
o
ti
o
n
esti
m
atio
n
,
wh
ich
is
ca
p
a
b
le
o
f
id
en
tif
y
in
g
an
o
b
ject'
s
m
o
b
ilit
y
b
etwe
en
two
co
n
s
ec
u
tiv
e
f
r
am
es
[
1
2
]
.
T
h
e
s
eg
m
en
tatio
n
p
lay
s
a
s
ig
n
if
ican
t
r
o
le
in
d
ev
elo
p
in
g
ap
p
licatio
n
s
o
r
tec
h
n
iq
u
es
f
o
r
tr
ac
k
in
g
th
e
v
id
eo
o
r
th
e
f
r
am
e
s
eq
u
en
ce
s
in
th
e
v
id
eo
.
T
h
er
e
ar
e
s
tu
d
i
es
wh
ich
h
av
e
also
b
ee
n
wo
r
k
ed
in
th
is
d
ir
ec
tio
n
wh
er
e
a
s
ig
n
if
ican
t
s
tu
d
y
is
b
e
in
g
co
n
d
u
cted
b
y
th
e
au
th
o
r
s
[
1
3
]
,
wh
er
e
t
h
e
o
b
jectiv
e
f
u
n
ct
io
n
f
o
r
o
p
tim
izin
g
th
e
ac
cu
r
ac
y
o
f
th
e
s
eg
m
en
t
atio
n
u
s
es
two
p
ar
am
eter
s
:
i)
en
tr
o
p
y
an
d
ii)
clu
s
ter
in
g
in
d
ices.
Fu
r
th
er
,
th
e
v
alid
atio
n
o
f
t
h
e
m
eth
o
d
h
as
e
x
p
er
im
en
ted
with
tr
a
d
itio
n
al
s
eg
m
en
tatio
n
tech
n
iq
u
es
th
at
i
n
clu
d
e:
i)
s
tatis
tical
r
eg
io
n
m
er
g
in
g
,
ii)
wate
r
s
h
e
d
an
d
K
-
m
ea
n
.
Alth
o
u
g
h
th
ey
h
av
e
test
ed
th
is
m
eth
o
d
o
n
f
o
u
r
d
if
f
er
e
n
t
d
atasets
,
all
th
ese
d
atasets
ar
e
h
eter
o
g
e
n
eo
u
s
im
a
g
es,
n
o
t
v
i
d
eo
s
eq
u
en
ce
s
.
Min
h
as
et
a
l.
[
1
4
]
p
r
o
p
o
s
e
a
n
o
v
el
co
n
ce
p
t
o
f
b
u
ild
in
g
a
s
em
an
tic
s
eg
m
en
tatio
n
n
etwo
r
k
f
r
o
m
s
k
in
f
ea
tu
r
es o
f
h
ig
h
s
ig
n
if
ican
c
e
th
at
f
i
n
e
-
tu
n
es th
e
o
b
ject
b
o
u
n
d
ar
ies
in
f
o
r
m
atio
n
at
d
if
f
er
en
t
s
ca
les.
T
h
e
m
eth
o
d
is
b
e
in
g
test
ed
a
n
d
v
alid
ated
o
n
m
an
y
h
u
m
an
ac
tiv
it
y
d
atab
ases
.
C
h
en
g
et
a
l.
[
1
5
]
i
n
tr
o
d
u
ce
s
a
f
r
a
m
ewo
r
k
n
am
el
y
ViT
r
ac
k
wh
ich
ta
r
g
ets
to
e
f
f
icien
tly
im
p
lem
e
n
t
m
u
lti
-
v
id
eo
tr
ac
k
i
n
g
s
y
s
tem
s
o
n
ed
g
e
to
f
ac
ilit
ates
th
e
v
id
eo
s
u
r
v
eillan
ce
r
eq
u
ir
em
e
n
ts
.
T
h
e
p
r
o
b
lem
f
o
r
m
u
latio
n
in
t
h
e
s
tu
d
y
ad
d
r
ess
es
th
e
co
r
e
r
esear
ch
c
h
allen
g
es
in
th
r
ee
p
r
im
e
ar
ea
s
o
f
v
id
eo
tr
ac
k
in
g
in
s
u
r
v
eillan
ce
s
y
s
tem
s
s
u
ch
as
i)
c
o
m
p
r
ess
ed
s
en
s
in
g
(
C
S)
[
1
6
]
–
[
1
8
]
,
ii)
o
b
ject
r
ec
o
g
n
i
tio
n
,
an
d
iii)
o
b
ject
tr
ac
k
in
g
.
Xin
g
et
a
l.
[
1
9
]
e
x
p
lo
r
ed
th
e
e
v
o
lu
tio
n
o
f
in
t
ellig
en
t
tr
an
s
p
o
r
tatio
n
s
y
s
tem
s
wh
er
e
v
eh
icu
lar
m
o
v
em
en
t
tr
ac
k
in
g
is
a
n
im
p
o
r
tan
t
co
n
ce
r
n
f
o
r
tr
af
f
ic
s
u
r
v
eillan
ce
.
T
h
e
au
th
o
r
s
m
o
s
tly
em
p
h
asized
o
n
d
esig
n
in
g
a
r
ea
l
-
tim
e
tr
ac
k
i
n
g
s
y
s
tem
o
f
v
eh
icu
lar
m
o
v
e
m
en
t
co
n
s
id
er
in
g
co
m
p
lex
f
o
r
m
o
f
s
ce
n
es
f
r
o
m
ca
p
tu
r
ed
v
id
eo
f
ee
d
s
.
T
h
e
a
u
th
o
r
s
in
tr
o
d
u
ce
th
e
tr
ac
k
in
g
m
o
d
el
n
am
ely
N
o
is
y
OT
Net
wh
ich
r
ea
lis
es
th
e
p
r
o
b
lem
o
f
o
b
ject
tr
ac
k
in
g
o
n
co
m
p
lex
v
i
d
eo
s
ce
n
es
as
r
ein
f
o
r
ce
m
en
t
lear
n
in
g
with
p
ar
am
eter
s
p
ac
e
p
r
o
b
lem
.
T
h
e
s
tu
d
y
e
x
p
lo
r
es
tr
ad
itio
n
al
v
eh
icle
tr
ac
k
i
n
g
m
eth
o
d
s
s
u
ch
as
co
r
r
elatio
n
f
ilter
-
b
ased
m
eth
o
d
[
2
0
]
–
[
2
2
]
,
d
ee
p
lea
r
n
in
g
-
b
a
s
ed
m
eth
o
d
s
[
2
3
]
,
[
2
4
]
f
o
r
v
e
h
icle
tr
ac
k
i
n
g
p
u
r
p
o
s
es.
I
t
f
in
d
s
th
at
co
r
r
elatio
n
-
b
ased
m
eth
o
d
s
a
n
d
d
ee
p
lear
n
in
g
-
b
ased
m
e
th
o
d
s
ad
o
p
t
s
tatic
lear
n
in
g
ap
p
r
o
ac
h
u
n
lik
e
r
ein
f
o
r
ce
m
e
n
t le
ar
n
in
g
[
2
5
]
,
[
2
6
]
.
Ab
d
elali
et
a
l.
[
2
7
]
also
a
d
d
r
ess
es
th
e
p
r
o
b
lem
o
f
v
eh
ic
u
lar
tr
af
f
ic
s
u
r
v
eillan
ce
an
d
r
o
a
d
v
io
latio
n
s
an
d
f
u
r
th
er
attem
p
ts
to
d
esig
n
an
ap
p
r
o
ac
h
to
tack
le
th
is
i
s
s
u
e.
I
n
th
is
r
e
g
ar
d
th
e
s
tu
d
y
in
tr
o
d
u
ce
s
a
f
u
lly
au
to
m
ated
m
eth
o
d
ical
ap
p
r
o
a
ch
n
am
ely
m
u
ltip
le
h
y
p
o
th
esi
s
d
etec
tio
n
an
d
tr
ac
k
in
g
(
MH
DT
)
to
d
ea
l
with
th
e
m
u
lti
-
o
b
ject
tr
ac
k
in
g
in
v
id
eo
s
.
T
h
e
r
esear
ch
m
eth
o
d
j
o
in
tly
in
teg
r
ates
Kalm
an
f
ilter
[
2
8
]
a
n
d
d
ata
ass
o
ciatio
n
-
b
ased
tr
ac
k
in
g
u
s
in
g
YOL
O
d
etec
tio
n
[
2
9
]
to
r
o
b
u
s
tly
tr
ac
k
v
eh
icu
lar
o
b
je
cts
in
th
e
co
m
p
lex
v
id
eo
s
ce
n
es.
On
ce
th
e
v
eh
icle
o
b
jects
ar
e
d
etec
ted
th
en
th
e
s
y
s
tem
em
p
lo
y
s
Kalm
an
f
ilter
b
ased
tr
ac
k
in
g
m
o
d
el.
T
h
is
ap
p
lies
a
tem
p
o
r
al
co
r
r
elatio
n
-
b
ased
th
eo
r
y
to
tr
ac
k
v
e
h
icles
am
o
n
g
o
n
e
f
r
a
m
e
to
an
o
th
er
.
T
h
e
d
esig
n
o
f
Kalm
an
f
ilter
[
2
8
]
is
co
n
s
tr
u
cted
in
s
u
ch
a
way
wh
er
e
f
o
r
ea
ch
tim
e
in
s
tan
ce
o
f
t,
it
p
r
o
v
id
es
th
e
f
ir
s
t
p
r
ed
ictio
n
́
.
Her
e
y
t c
o
r
r
esp
o
n
d
to
th
e
s
tate.
́
=
×
−
1
(
1
)
T
h
e
Kalm
an
f
ilter
also
esti
m
ates
th
e
s
tate
p
r
ed
ictio
n
s
t
ep
s
co
n
s
id
er
in
g
a
co
v
a
r
ian
ce
esti
m
atio
n
ca
lcu
latio
n
.
T
h
e
s
tu
d
y
also
a
n
aly
s
es
v
ar
io
u
s
r
elate
d
wo
r
k
s
an
d
o
b
s
er
v
e
d
th
at
m
o
s
t
o
f
t
h
e
s
tu
d
ies
an
d
th
eir
in
co
r
p
o
r
ated
al
g
o
r
ith
m
s
co
n
s
id
er
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
et
wo
r
k
(
C
NN
)
as
class
if
ier
s
an
d
it
y
ield
s
b
etter
ac
cu
r
ac
y
wh
ich
lies
b
etwe
en
9
3
%
to
9
7
%.
T
h
e
co
m
p
u
tatio
n
al
co
m
p
lex
ity
is
ev
alu
ated
with
r
esp
ec
t
to
th
e
esti
m
atio
n
o
f
b
o
u
n
d
in
g
b
o
x
c
o
o
r
d
in
ates
(
b
)
wh
ich
s
tates
th
at
th
e
o
v
er
all
co
m
p
u
tatio
n
al
co
s
t
o
f
th
e
m
o
d
el
s
tan
d
s
as
(
b
3
+
b
2
+
b
)
.
I
t
h
as
b
ee
n
o
b
s
er
v
ed
th
at
th
e
v
ar
iatio
n
f
ac
to
r
in
illu
m
in
atio
n
ca
u
s
es
s
ig
n
if
ican
t
ch
allen
g
es
in
v
id
eo
s
u
r
v
eillan
ce
s
y
s
tem
s
to
war
d
s
m
u
ltip
le
o
b
ject
d
etec
tio
n
an
d
tr
ac
k
in
g
in
t
h
e
p
r
esen
ce
o
f
m
o
tio
n
f
ac
to
r
s
.
E
v
en
th
o
u
g
h
v
ar
io
u
s
s
ch
em
es
b
ein
g
ev
o
lv
ed
an
d
s
tu
d
ied
f
o
r
s
ev
e
r
al
d
ec
ad
es
f
o
r
d
if
f
e
r
en
t
task
s
,
d
u
e
to
illu
m
in
atio
n
v
ar
iatio
n
f
ac
to
r
s
,
th
er
e
r
e
m
ai
n
co
n
s
tr
ain
ts
o
f
d
ef
o
r
m
atio
n
o
f
m
o
b
ile
o
b
jects,
p
au
s
e
m
o
ti
o
n
b
lu
r
,
o
cc
lu
s
io
n
s
(
f
u
ll/p
ar
tial
)
a
n
d
ca
m
er
a
v
iew
an
g
le.
T
h
ese
cr
u
cial
asp
ec
ts
a
r
e
y
et
u
n
s
o
lv
e
d
p
r
o
b
lem
s
ass
o
ciate
d
with
m
o
b
ile
o
b
ject
d
etec
tio
n
a
n
d
tr
ac
k
in
g
f
r
o
m
d
y
n
am
ic
v
id
eo
s
ce
n
es.
Als
o
,
th
e
ch
allen
g
es
with
th
e
tr
ad
itio
n
al
tr
ac
k
in
g
s
y
s
tem
s
ar
e
lack
o
f
ef
f
ec
tiv
e
n
ess
in
lo
ca
lizin
g
th
e
o
b
ject
o
f
in
ter
est
p
r
o
p
er
ly
in
t
h
e
p
r
esen
ce
o
f
d
y
n
am
ic
tr
an
s
itio
n
o
f
b
ac
k
g
r
o
u
n
d
,
lack
o
f
h
a
n
d
lin
g
t
h
e
p
r
esen
ce
o
f
v
ar
iatio
n
in
asp
ec
t
r
ati
o
s
,
v
ar
i
atio
n
o
f
in
t
r
a
-
class
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2
2
5
2
-
8
9
3
8
Desig
en
o
f a
n
effec
tive
mu
ltip
l
e
o
b
ject
tr
a
ck
in
g
fr
a
mewo
r
k
fo
r
d
yn
a
mic
vid
eo
s
ce
n
es
(
K
a
r
a
n
a
m
S
u
n
il K
u
ma
r
)
3881
o
b
jects,
ap
p
r
o
p
r
iate
co
n
te
x
tu
a
l
in
f
o
r
m
atio
n
an
d
p
r
esen
ce
o
f
co
m
p
lex
b
ac
k
g
r
o
u
n
d
[
3
0
]
,
[
3
1
]
.
Ap
ar
t
f
r
o
m
th
is
,
th
e
m
o
s
t
s
ig
n
if
ican
t
ch
allen
g
e
ar
is
es
with
h
ig
h
er
ac
cu
r
ac
y
o
f
m
u
ltip
le
o
b
ject
d
etec
tio
n
an
d
tr
ac
k
in
g
wh
ile
b
alan
cin
g
c
o
n
s
id
er
ab
le
co
s
t
-
ef
f
ec
tiv
e
co
m
p
u
tatio
n
al
p
er
f
o
r
m
an
ce
,
wh
ich
is
less
lik
e
ly
ex
p
lo
r
ed
i
n
th
e
ex
is
tin
g
s
y
s
tem
s
o
f
MO
T
m
o
d
els.
Af
ter
r
ev
iewin
g
t
h
e
ex
is
tin
g
s
tu
d
ies
o
n
MO
T
,
t
h
e
id
en
tifie
d
r
esear
ch
p
r
o
b
lem
s
o
u
tlin
e
th
e
f
ac
t
th
at
ev
en
th
o
u
g
h
th
er
e
e
x
is
t
v
ar
io
u
s
f
o
r
m
o
f
wo
r
k
o
n
MO
T
b
u
t
th
e
m
ajo
r
ity
o
f
th
e
tr
ac
k
in
g
m
o
d
els
ac
co
m
p
lis
h
h
ig
h
er
ac
c
u
r
ac
y
o
f
d
etec
tio
n
a
n
d
tr
ac
k
i
n
g
at
t
h
e
co
s
t
o
f
co
m
p
u
tatio
n
al
co
m
p
lex
ity
,
wh
ich
is
th
e
s
im
ilar
ca
s
e
with
th
e
ex
is
tin
g
m
ac
h
in
e
lear
n
in
g
(
ML
)
b
ased
a
p
p
r
o
ac
h
es
as
well.
Seco
n
d
ly
,
m
o
s
t
s
tu
d
ies
d
o
n
o
t
co
n
s
id
er
co
n
tex
tu
al
c
o
n
n
ec
tiv
ity
f
ac
to
r
s
o
f
an
o
b
ject
with
its
b
ac
k
g
r
o
u
n
d
,
wh
ich
r
em
ain
s
a
ch
alle
n
g
e
in
th
e
e
x
is
tin
g
wo
r
k
s
.
T
h
e
ap
p
r
o
p
r
iate
in
clu
s
io
n
o
f
f
ea
tu
r
e
en
g
in
ee
r
in
g
i
s
also
m
i
s
s
in
g
in
th
e
ex
is
tin
g
ML
-
b
ased
MO
T
tech
n
iq
u
es
f
o
r
tr
ac
k
i
n
g
d
y
n
a
m
ic
m
o
b
ile
o
b
jects
in
th
e
c
o
m
p
lex
v
id
eo
s
ce
n
es,
w
h
er
e
co
n
tex
tu
al
s
ce
n
e
in
f
o
r
m
atio
n
also
p
lay
s
a
cr
u
cial
r
o
le.
T
h
e
s
tu
d
y
'
s
p
r
o
b
lem
s
tatem
en
t
is
"
T
o
d
esig
n
a
c
o
s
t
-
ef
f
ec
tiv
e
an
d
h
i
g
h
ly
ac
cu
r
ate
MO
T
f
r
am
ewo
r
k
to
p
er
f
o
r
m
o
b
ject
d
etec
tio
n
an
d
tr
ac
k
in
g
f
r
o
m
co
m
p
lex
v
id
e
o
s
ce
n
es
co
n
s
id
er
in
g
co
n
tex
tu
al
in
f
o
r
m
atio
n
is
a
h
ig
h
ly
c
h
allen
g
in
g
task
".
T
h
is
p
r
o
p
o
s
ed
s
tu
d
y
a
d
d
r
ess
es
th
is
p
r
o
b
lem
,
an
d
a
n
o
v
el
co
m
p
u
tatio
n
al
co
n
tex
tu
al
f
r
am
ewo
r
k
is
in
tr
o
d
u
ce
d
f
o
r
e
f
f
ec
tiv
e
MO
T
.
T
h
e
n
o
v
elty
o
f
th
is
f
r
am
ewo
r
k
is
t
h
at
it
ca
n
id
en
tify
n
u
m
er
o
u
s
m
o
b
ile
o
b
jects
f
r
o
m
t
h
e
d
y
n
a
m
ic
s
ce
n
es
an
d
also
r
ed
u
ce
s
t
h
e
co
s
t
o
f
co
m
p
u
tatio
n
al
e
f
f
o
r
t
with
a
s
im
p
lifie
d
tr
ac
k
in
g
m
o
d
u
le.
T
h
e
co
n
tr
i
b
u
tio
n
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
is
it
ap
p
lies
co
s
t
-
ef
f
ec
tiv
e
m
o
d
ellin
g
o
f
ass
ig
n
in
g
o
b
ject
d
etec
tio
n
in
th
e
c
u
r
r
en
t
f
r
am
e
to
ex
is
tin
g
t
r
ac
k
s
with
a
n
o
p
tim
al
esti
m
ato
r
.
I
t
also
ex
p
lo
r
es
th
e
s
co
p
e
o
f
im
p
r
o
v
em
e
n
t
in
m
o
b
ile
o
b
je
ct
d
etec
tio
n
co
n
s
id
er
in
g
th
e
m
eth
o
d
o
f
Gau
s
s
ian
m
ix
tu
r
e
m
o
d
el
(
GM
M)
an
d
im
p
r
o
v
es
th
e
tr
ac
k
in
g
p
e
r
f
o
r
m
an
ce
u
s
in
g
Kalm
an
f
ilter
-
b
a
s
ed
ap
p
r
o
ac
h
.
Her
e
th
e
s
tr
ate
g
y
also
ex
p
lo
r
es
th
e
ass
o
ciatio
n
am
o
n
g
th
e
d
etec
t
ed
m
o
b
ile
o
b
jects
f
r
o
m
o
n
e
f
r
am
e
to
th
e
n
ex
t
an
d
o
v
e
r
co
m
es
th
e
ass
o
ciatio
n
p
r
o
b
lem
.
Her
e
t
h
e
in
clu
s
io
n
o
f
th
e
Kalm
an
f
ilter
m
eth
o
d
p
r
ed
icts
th
e
s
tate
v
ar
iab
les
ef
f
ec
tiv
ely
,
wh
ich
en
h
an
ce
s
th
e
tr
ac
k
in
g
p
er
f
o
r
m
an
ce
with
c
o
s
t
-
ef
f
ec
tiv
e
tr
ajec
to
r
y
f
o
r
m
u
latio
n
f
o
r
th
e
m
o
b
ile
o
b
jects
ev
e
n
in
th
e
p
r
esen
ce
o
f
co
m
p
lex
an
d
d
y
n
am
ic
s
ce
n
es.
I
t h
as to
b
e
n
o
t
ed
th
at
th
e
id
en
tific
atio
n
o
f
m
o
b
ile
o
b
jects in
th
e
p
r
o
p
o
s
ed
s
tu
d
y
co
n
s
id
er
s
th
e
co
n
tex
t
u
al
asp
ec
t
o
f
th
e
o
b
ject,
wh
ich
is
also
r
e
f
er
r
ed
to
as
th
e
lin
e
o
f
m
o
v
em
en
t
(
L
o
M)
.
An
o
th
er
n
o
v
elty
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
im
p
lied
d
esig
n
ex
ec
u
ti
o
n
wh
ich
m
a
k
es
th
e
en
tire
s
y
s
tem
co
m
p
u
tatio
n
ally
ef
f
icien
t w
h
en
c
o
m
p
ar
e
d
with
th
e
ex
is
tin
g
b
aselin
e
ap
p
r
o
ac
h
es.
T
h
is
n
ew
co
n
ce
p
t
o
f
d
y
n
am
ic
tr
ac
k
in
g
o
f
n
u
m
er
o
u
s
m
o
b
ile
o
b
jects
tak
es
a
d
v
an
tag
e
o
f
G
MM
in
th
e
s
eg
m
en
tatio
n
o
f
o
b
jects.
I
t
also
h
an
d
les
th
e
co
n
s
tr
ain
ts
o
f
tr
ad
itio
n
al
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
m
eth
o
d
s
to
war
d
s
th
e
ap
p
r
o
p
r
iate
d
etec
tio
n
o
f
m
o
v
in
g
o
b
jects.
T
h
e
s
tu
d
y
also
f
u
r
th
er
im
p
r
o
v
is
es
th
e
tr
ac
k
in
g
m
o
d
el
co
n
s
id
er
in
g
th
e
p
o
ten
tial
f
ea
t
u
r
es
o
f
th
e
Kalm
an
f
ilter
to
war
d
s
p
r
ed
ictin
g
th
e
ce
n
tr
o
id
o
f
ea
ch
tr
ac
k
f
o
r
m
o
tio
n
-
b
ased
tr
ac
k
in
g
,
th
r
o
u
g
h
wh
ich
it
h
as
also
h
an
d
led
th
e
tr
ac
k
ass
ig
n
m
en
t
p
r
o
b
lem
.
T
h
e
ex
p
e
r
im
en
tal
o
u
tco
m
e
f
u
r
th
er
ju
s
tifie
s
h
o
w
th
e
f
o
r
m
u
lated
co
n
ce
p
t
o
f
L
o
M
co
n
s
id
er
s
d
ir
ec
tio
n
al
ity
m
o
v
em
en
t
th
at
co
s
t
-
ef
f
ec
tiv
ely
p
er
f
o
r
m
s
ass
o
ciatio
n
am
o
n
g
id
e
n
tifie
d
m
o
v
in
g
o
b
jects
an
d
p
er
f
o
r
m
s
tr
ac
k
in
g
co
n
s
id
er
in
g
tr
ajec
to
r
y
f
o
r
m
u
latio
n
.
I
t
al
s
o
s
h
o
ws
b
etter
id
en
tific
atio
n
p
er
f
o
r
m
a
n
ce
b
y
t
h
e
tr
ac
k
in
g
m
o
d
u
le
with
co
s
t
-
ef
f
ec
tiv
en
ess
wh
en
co
m
p
ar
ed
with
th
e
b
aselin
e
a
p
p
r
o
ac
h
es.
Un
lik
e
b
aselin
e
s
tu
d
ies,
th
e
p
r
o
p
o
s
ed
s
tr
ateg
y
o
f
f
er
s
a
m
u
ch
l
o
wer
r
esp
o
n
s
e
tim
e
with
co
n
s
id
er
ab
l
e
p
r
o
ce
s
s
in
g
ex
ec
u
tio
n
an
d
ite
r
atio
n
s
.
2.
M
E
T
H
O
D
T
h
is
p
ar
t
o
f
th
e
s
tu
d
y
f
o
r
m
u
lates
th
e
an
aly
tical
d
esig
n
m
o
d
elin
g
o
f
th
e
p
r
o
p
o
s
ed
c
o
s
t
-
ef
f
icien
t
d
y
n
am
ic
tr
ac
k
i
n
g
m
o
d
el
wh
ich
is
ca
p
ab
le
o
f
tr
ac
k
in
g
m
u
ltip
le
v
id
eo
o
b
jects
with
h
i
g
h
er
ac
cu
r
ac
y
an
d
co
m
p
u
tatio
n
al
ef
f
icien
cy
.
T
h
e
s
tu
d
y
f
o
r
m
u
lates
th
e
f
lo
w
o
f
th
e
d
esig
n
with
an
aly
tical
r
e
s
ea
r
ch
m
o
d
elin
g
t
o
r
ea
lis
e
th
e
wo
r
k
in
g
s
ce
n
a
r
io
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
.
I
t
also
in
v
o
lv
es
a
s
et
o
f
f
u
n
ctio
n
al
m
o
d
u
les
wh
ich
o
p
er
ates o
n
f
u
lf
illi
n
g
th
e
d
esig
n
r
eq
u
i
r
em
en
ts
o
f
t
h
e
p
r
o
p
o
s
e
d
s
y
s
tem
.
T
h
e
b
lo
c
k
-
b
ased
a
r
ch
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
in
Fig
u
r
e
1
ex
h
ib
its
th
at
it
c
o
n
s
id
er
s
o
f
a
s
et
o
f
o
p
er
atio
n
al
m
o
d
u
les
wh
er
e
th
e
f
ir
s
t
m
o
d
u
le
is
ass
o
ciate
d
with
v
id
eo
I
/O
i
n
itializatio
n
wh
er
e
it
co
n
s
tr
u
cts
a
v
id
eo
r
ea
d
er
o
b
ject
an
d
r
ea
d
t
h
e
v
id
eo
f
ile.
Her
e
th
e
f
u
n
cti
o
n
ality
co
n
s
tr
u
cts
a
r
ef
er
e
n
ce
o
b
ject
(
Ov
)
wh
ich
b
asically
co
m
p
u
tes
d
if
f
er
en
t
a
ttrib
u
tes
wh
ich
is
f
u
r
th
er
d
is
c
u
s
s
ed
in
th
e
co
n
s
ec
u
tiv
e
s
ec
ti
o
n
s
.
Fu
r
th
er
it
also
in
itializes
two
p
lay
er
s
wh
ich
ar
e
P1
an
d
P2
r
esp
ec
tiv
ely
to
v
is
u
alize
th
e
co
m
p
u
tatio
n
o
f
f
o
r
eg
r
o
u
n
d
m
as
an
d
th
e
v
id
eo
f
ile
s
eq
u
e
n
ce
o
f
(
Vf
)
.
Fu
r
th
e
r
th
e
s
y
s
tem
also
co
n
s
tr
u
cts
ex
p
licit
f
u
n
ctio
n
ali
ties
to
in
itialize
th
e
o
p
er
atio
n
s
c
o
r
r
esp
o
n
d
in
g
to
G
au
s
s
ian
b
ased
d
etec
to
r
f
o
r
f
o
r
eg
r
o
u
n
d
a
n
d
b
in
ar
y
lar
g
e
o
b
je
cts
(
B
lo
B
)
an
aly
ze
r
wh
ich
also
co
n
s
id
er
s
th
e
r
ef
e
r
en
ce
o
b
ject
f
r
o
m
th
e
v
id
eo
s
e
q
u
en
ce
.
Fu
r
th
er
th
e
s
tu
d
y
also
em
p
lo
y
s
a
d
y
n
am
ic
m
o
b
ile
o
b
ject
d
etec
tio
n
m
o
d
u
le
wh
ich
b
asically
co
n
s
tr
u
c
ts
s
y
s
tem
o
b
jects
to
r
ea
d
th
e
v
id
eo
f
ile
in
p
u
t
s
eq
u
en
ce
an
d
also
d
etec
t
th
e
f
o
r
eg
r
o
u
n
d
o
b
ject.
He
r
e
th
e
s
tu
d
y
also
en
h
an
ce
s
th
e
o
p
er
atio
n
s
o
f
p
r
ec
is
e
o
b
ject
d
etec
tio
n
b
y
in
c
o
r
p
o
r
atin
g
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
wh
ic
h
p
er
f
o
r
m
s
p
r
e
-
p
r
o
ce
s
s
in
g
o
v
er
th
e
d
ata
a
n
d
m
ak
e
it
s
u
itab
le
f
o
r
v
id
e
o
a
n
aly
s
is
f
o
r
B
lo
b
A
n
aly
ze
r
.
T
h
e
p
r
o
p
o
s
ed
s
tr
ateg
y
f
u
r
th
er
ap
p
lie
s
GM
M
to
p
e
r
f
o
r
m
p
r
ec
is
e
o
b
ject
s
eg
m
en
tatio
n
f
r
o
m
th
e
co
m
p
lex
v
id
e
o
s
ce
n
es.
T
h
e
ap
p
r
o
ac
h
also
co
n
s
id
er
s
in
itializatio
n
o
f
tr
ac
k
in
g
m
o
d
u
le
wh
er
e
it
co
n
s
tr
u
cts
s
tr
u
ctu
r
e
ar
r
a
y
f
ield
s
.
Fin
ally
,
th
e
s
tu
d
y
a
p
p
lies
a
Kalm
an
f
il
t
er
t
o
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
3
8
7
9
-
3
8
9
1
3882
en
h
an
ce
th
e
p
r
e
d
ictio
n
o
f
n
e
w
lo
ca
tio
n
o
f
tr
ac
k
wh
er
e
th
e
co
m
p
u
tatio
n
o
f
ce
n
tr
o
id
ca
lc
u
latio
n
an
d
u
p
d
atin
g
b
o
u
n
d
in
g
b
o
x
also
e
v
o
lv
es.
Fin
ally
,
th
e
p
r
o
p
o
s
ed
s
y
s
tem
s
tr
ateg
y
also
h
an
d
les
th
e
tr
ac
k
a
s
s
ig
n
m
en
t
p
r
o
b
lem
f
o
r
d
etec
ted
m
o
b
ile
o
b
ject
an
d
h
er
e
also
u
s
e
Kalm
an
f
ilter
ap
p
r
o
ac
h
to
p
e
r
f
o
r
m
d
etec
tio
n
s
to
tr
ac
k
ass
ig
n
m
en
t.
I
t
h
as
to
b
e
n
o
te
d
th
at
th
e
en
tire
p
r
o
ce
s
s
also
m
in
im
izes
th
e
co
s
t
o
f
tr
ac
k
a
llo
ca
tio
n
wh
er
e
th
e
tr
ac
k
d
ep
icts
th
e
c
o
n
tex
tu
al
L
o
M
asp
ec
t
f
o
r
th
e
m
o
b
ile
o
b
ject.
Fu
r
th
er
t
h
e
p
r
o
p
o
s
ed
s
t
r
ateg
y
p
e
r
f
o
r
m
s
th
e
u
p
d
atin
g
o
p
er
atio
n
s
with
r
esp
ec
t
to
u
p
d
atin
g
attr
ib
u
tes
an
d
ex
h
ib
its
th
e
f
in
al
tr
ac
k
ed
m
o
b
ile
o
b
jects
f
r
o
m
th
e
co
m
p
lex
v
id
e
o
s
ce
n
es.
I
t
h
as
to
b
e
n
o
ted
th
at
th
e
co
r
e
s
tr
ateg
y
o
f
th
e
p
r
o
p
o
s
ed
tr
ac
k
in
g
m
o
d
u
le
is
to
ef
f
ec
tiv
ely
lo
ca
te
th
e
m
o
v
i
n
g
o
b
ject
o
r
m
u
ltip
le
o
b
jects
o
v
e
r
p
r
o
g
r
ess
iv
e
tim
e
f
o
r
a
g
iv
en
V
f
.
Her
e
th
e
in
th
e
co
r
e
s
tr
ateg
y
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
id
en
tifie
s
th
e
ass
o
ciatio
n
p
r
o
b
lem
an
d
d
etec
ts
an
o
b
j
ec
t
ac
r
o
s
s
m
u
ltip
le
f
r
am
es
o
f
a
v
id
eo
s
tr
ea
m
.
T
h
e
co
r
e
s
tr
ateg
y
o
f
t
h
e
p
r
o
p
o
s
ed
s
y
s
tem
also
co
n
s
id
er
s
t
h
e
f
u
n
d
am
en
tal
p
r
in
cip
le
o
f
b
aselin
e
m
o
d
els
o
f
tr
ac
k
i
n
g
wh
er
e
th
e
c
o
r
e
p
h
ilo
s
o
p
h
y
is
to
in
itially
d
etec
t
th
e
o
b
jec
ts
o
f
in
ter
est
in
th
e
v
id
eo
f
r
am
e
an
d
f
u
r
th
er
p
er
f
o
r
m
in
g
p
r
e
d
ictio
n
t
o
co
n
s
tr
u
ct
th
e
L
o
M
o
f
o
b
ject
tr
ajec
t
o
r
ies
o
v
er
th
e
n
ex
t
co
n
s
ec
u
tiv
e
f
r
am
es
o
f
a
v
id
e
o
s
eq
u
en
ce
.
T
h
e
p
r
o
p
o
s
ed
s
tu
d
y
h
a
n
d
les
th
e
p
r
o
b
lem
o
f
d
ata
ass
o
ciatio
n
b
y
esti
m
atin
g
th
e
p
r
ed
icted
lo
ca
tio
n
s
an
d
f
u
r
th
er
ass
o
ciate
th
e
d
etec
tio
n
s
ac
r
o
s
s
th
e
f
r
a
m
e
s
to
f
o
r
m
u
late
t
h
e
tr
ajec
to
r
ies f
o
r
th
e
L
o
M
f
o
r
r
e
s
p
ec
tiv
e
o
b
jects.
Fig
u
r
e
1
.
Ar
c
h
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
MO
T
f
r
am
ewo
r
k
2
.
1
.
Video
inp
ut
-
o
utput
ini
t
ia
liza
t
io
n
T
h
e
co
m
p
u
tin
g
p
r
o
ce
s
s
in
v
o
l
v
ed
in
th
e
p
r
o
p
o
s
ed
in
th
e
p
r
o
p
o
s
ed
co
s
t
-
ef
f
ec
tiv
e
d
y
n
am
i
c
tr
ac
k
in
g
m
o
d
el
in
itially
e
m
p
lo
y
s
a
f
u
n
ctio
n
ality
f
o
r
v
id
e
o
in
p
u
t
-
o
u
tp
u
t
i
n
itializatio
n
.
Her
e
t
h
e
s
y
s
tem
in
itially
co
n
s
id
er
s
th
e
in
p
u
t
v
id
eo
(
V
f
)
f
r
o
m
th
e
s
u
r
v
eillan
ce
s
y
s
tem
.
T
h
e
in
f
o
r
m
atio
n
r
elate
d
to
O
v
is
h
an
d
led
wh
ile
co
n
s
tr
u
ctin
g
a
r
ef
er
e
n
ce
o
b
je
ct
(
O
v
)
.
Her
e
th
e
s
y
s
tem
em
p
lo
y
s
a
f
u
n
ctio
n
ality
o
f
f
VR
(V
f
)
→
O
v
wh
ich
h
elp
s
co
n
s
tr
u
ctin
g
th
is
o
b
ject.
T
h
is
p
h
ase
o
f
co
m
p
u
tatio
n
also
co
m
es
u
n
d
er
d
ata
ex
p
lo
r
atio
n
co
r
r
esp
o
n
d
in
g
to
th
e
in
p
u
t
V
f
.
T
h
e
c
o
m
p
u
tatio
n
o
f
th
e
r
ef
e
r
en
ce
in
f
o
r
m
atio
n
c
o
r
r
esp
o
n
d
s
to
V
f
.
T
h
e
ex
p
lo
r
ati
o
n
o
f
th
e
r
ef
er
e
n
ce
co
n
s
tr
u
cted
o
b
ject
o
f
Ov
s
h
o
ws
th
at
th
e
cu
r
r
en
t
tim
e
(
C
t
)
r
ef
er
s
to
tim
e
s
tam
p
r
eq
u
ir
ed
to
r
ea
d
th
e
f
r
am
e
co
r
r
esp
o
n
d
to
V
f
.
Her
e
t
h
e
tag
attr
ib
u
te
b
asically
r
ef
er
s
to
as
a
r
e
f
er
en
ce
to
id
e
n
tif
y
th
e
Ov
s
u
ch
as
[
tag
➔
O
v
]
.
T
h
is
is
an
o
p
tio
n
al
n
am
e
-
v
alu
e
p
air
ar
g
u
m
en
t
f
o
r
th
e
c
o
m
p
u
tatio
n
o
f
th
e
r
ef
er
en
ce
o
b
ject
f
r
o
m
th
e
v
id
eo
f
ile.
Her
e
t
h
e
u
s
er
d
ata
(
UD)
is
also
co
n
s
tr
u
cted
as
an
o
p
tio
n
al
n
a
m
e
-
v
alu
e
p
ai
r
attr
ib
u
te
wh
er
e
it
r
ef
er
s
to
a
g
en
er
ic
f
ield
to
h
o
ld
an
y
n
ew
in
f
o
r
m
atio
n
wh
ic
h
ca
n
b
e
ad
d
ed
t
o
th
e
r
e
f
er
en
ce
o
b
ject
Ov
.
T
h
e
p
r
o
ce
s
s
in
g
an
d
t
h
e
co
m
p
u
tatio
n
o
f
t
h
e
V
f
with
th
e
f
u
n
ctio
n
ality
o
f
f
VR
(
x
)
co
n
s
tr
u
cts
a
r
ef
er
en
ce
o
b
ject
Ov
wh
ich
h
o
l
d
s
th
e
f
o
llo
win
g
p
r
o
p
er
ties
as
s
h
o
wn
in
th
e
f
o
llo
win
g
Fig
u
r
e
2
.
T
h
e
lo
ca
tio
n
a
ttrib
u
te
o
f
p
ath
(
P)
co
n
tain
s
th
e
r
ef
er
en
ce
p
ath
to
lo
ca
te
th
e
v
id
eo
f
ile.
T
h
e
g
en
e
r
al
p
r
o
p
e
r
ty
o
f
th
e
r
e
f
er
en
ce
o
b
ject
also
in
clu
d
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2
2
5
2
-
8
9
3
8
Desig
en
o
f a
n
effec
tive
mu
ltip
l
e
o
b
ject
tr
a
ck
in
g
fr
a
mewo
r
k
fo
r
d
yn
a
mic
vid
eo
s
ce
n
es
(
K
a
r
a
n
a
m
S
u
n
il K
u
ma
r
)
3883
th
e
n
am
e
o
f
th
e
v
id
e
o
f
ile
nV
f
wh
ich
is
as
s
o
ciate
d
with
th
e
o
b
ject
O
v
.
Her
e
th
e
d
u
r
atio
n
(
t
)
co
n
s
id
er
s
th
e
to
tal
len
g
th
o
f
th
e
V
f
.
T
h
e
co
m
p
u
te
d
r
ef
er
en
ce
o
b
ject
o
f
th
e
V
f
al
s
o
co
n
s
is
ts
o
f
o
th
er
im
p
o
r
tan
t in
f
o
r
m
atio
n
r
elate
d
to
v
id
eo
p
r
o
p
er
ties
.
Her
e
in
t
h
e
T
ab
le
1
th
e
attr
ib
u
te
o
f
b
p
r
e
f
er
s
to
th
e
b
its
am
o
u
n
t
c
o
r
r
esp
o
n
d
t
o
u
n
it
o
f
p
ix
el
in
th
e
r
esp
ec
tiv
e
V
f
.
T
h
e
attr
ib
u
te
(
Fr
)
also
r
ef
er
s
to
th
e
f
r
am
e
r
ate
o
f
th
e
V
f
co
m
p
u
ted
in
f
r
am
e/s.
I
t
also
co
m
p
u
tes
th
e
h
eig
h
t
(
h
)
o
f
t
h
e
i
th
f
r
am
e
(
fr
a
me
i
)
o
f
V
f
in
p
ix
els
alo
n
g
with
wid
th
(
w
)
o
f
th
e
(
fr
a
me
i
)
in
p
ix
els.
I
t a
ls
o
co
m
p
u
tes th
e
n
u
m
b
e
r
o
f
f
r
am
es (
fr
a
me
n
)
alo
n
g
with
t
h
e
v
id
eo
f
o
r
m
at
ty
p
e.
T
h
e
s
tr
u
ctu
r
e
o
f
Ov
is
f
in
ally
co
n
s
tr
u
cted
co
n
s
id
er
in
g
its
ess
en
tial
p
r
o
p
er
ties
to
u
n
d
er
s
tan
d
th
e
in
p
u
t
v
id
eo
d
ata.
T
h
e
ch
allen
g
es
ar
i
s
e
in
th
e
co
n
v
en
tio
n
al
s
y
s
tem
s
in
d
etec
tio
n
o
f
m
o
v
in
g
o
b
ject
s
f
r
o
m
th
e
d
y
n
am
ic
v
id
eo
s
ce
n
es.
I
n
th
e
p
r
o
b
lem
o
f
tr
ac
k
in
g
th
e
m
o
v
in
g
o
b
ject
s
f
r
o
m
th
e
v
id
e
o
s
eq
u
en
ce
s
,
s
eg
m
en
tatio
n
o
f
th
e
d
y
n
am
ic
r
eg
i
o
n
in
th
e
r
ea
l
-
tim
e
s
y
n
ch
r
o
n
izatio
n
is
a
q
u
ite
ch
allen
g
in
g
task
b
ec
au
s
e
o
f
v
ar
i
o
u
s
r
ea
s
o
n
s
wh
ich
in
clu
d
e
co
m
p
lex
an
d
m
o
v
in
g
b
ac
k
g
r
o
u
n
d
,
o
cc
lu
s
io
n
,
m
o
ti
o
n
b
lu
r
,
illu
m
in
atio
n
v
a
r
iatio
n
s
an
d
m
an
y
m
o
r
e
o
th
er
f
ac
to
r
s
.
T
h
er
ef
o
r
e,
t
o
h
an
d
le
in
d
iv
id
u
al
ch
allen
g
es
m
an
y
c
u
s
to
m
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
m
eth
o
d
s
is
b
ein
g
e
v
o
lv
ed
.
T
h
e
T
ab
le
1
f
u
r
th
er
p
r
o
v
i
d
es
s
o
m
e
o
f
th
e
im
p
o
r
tan
t
i
n
f
o
r
m
atio
n
a
b
o
u
t
th
e
p
r
o
p
er
ties
o
f
th
e
V
f
th
r
o
u
g
h
Ov
.
T
h
e
in
f
er
e
n
ce
o
f
T
ab
le
1
s
h
o
ws th
e
im
p
o
r
tan
t p
r
o
p
er
ties
o
f
V
f
e
x
p
lo
r
e
d
th
r
o
u
g
h
th
e
o
b
ject
an
d
its
ass
o
ciate
d
m
eth
o
d
s
o
f
Ov
.
Fig
u
r
e
2
.
Gen
e
r
al
p
r
o
p
er
ties
: O
v
T
ab
le
1
.
I
m
p
o
r
ta
n
t p
r
o
p
er
ties
o
f
Vf
S
l
.
N
o
P
r
o
p
e
r
t
y
N
a
m
e
1
B
i
t
s /
P
i
x
e
l
(
b
p
)
2
F
r
a
me
R
a
t
e
(Fr)
3
H
e
i
g
h
t
(
h
)
4
W
i
d
t
h
(
w
)
5
N
u
mb
e
r
o
f
F
r
a
mes
(
n
)
6
V
i
d
e
o
F
o
r
ma
t
I
n
t
h
es
e
m
et
h
o
d
s
t
h
e
f
ast
le
ar
n
in
g
i
n
t
h
e
d
en
s
e
en
v
i
r
o
n
m
en
t i
s
th
e
m
a
in
f
o
cu
s
o
f
r
esea
r
ch
.
T
h
e
e
x
p
lic
it
alg
o
r
it
h
m
f
o
r
t
h
e
v
i
d
e
o
in
p
u
t
-
o
u
tp
u
t
i
n
i
tia
liz
ati
o
n
as
i
n
A
lg
o
r
it
h
m
1
.
T
h
e
n
u
m
er
ic
al
alg
o
r
it
h
m
m
o
d
eli
n
g
in
i
tial
ly
c
o
n
s
id
er
s
th
e
v
i
d
e
o
s
e
q
u
e
n
c
es
th
r
o
u
g
h
t
h
e
v
i
d
eo
f
i
le
(
V
f
)
a
n
d
i
n
iti
all
y
cr
ea
t
es
tw
o
p
la
y
er
o
b
je
cts
as
P
1
an
d
P
2
f
o
r
f
o
r
e
g
r
o
u
n
d
m
as
k
a
n
d
o
r
i
g
i
n
al
v
i
d
eo
s
e
q
u
e
n
c
es
r
es
p
ec
ti
v
el
y
.
T
h
e
s
t
u
d
y
f
u
r
t
h
e
r
em
p
l
o
y
s
in
i
tial
iz
ati
o
n
an
d
c
r
e
ati
o
n
o
f
a
n
e
x
p
lici
t
f
u
n
cti
o
n
:
f
u
n
c
ti
o
n
f
o
r
t
h
e
f
o
r
e
g
r
o
u
n
d
d
e
te
ct
o
r
(
f
fd
)
ta
k
es
i
n
p
u
t
p
a
r
a
m
e
te
r
s
et
as
{
Nu
m
b
e
r
o
f
Ga
u
s
s
i
a
n
s
(
N
g
)
,
n
u
m
b
e
r
o
f
f
r
am
es
f
o
r
th
e
tr
ai
n
i
n
g
(
N
T
f
)
,
p
er
ce
n
ta
g
e
o
f
t
h
e
m
i
n
im
u
m
b
ac
k
g
r
o
u
n
d
r
ati
o
(
M
B
r
)
}
t
o
c
o
n
s
t
r
u
ct
t
h
e
d
ete
ct
o
r
(
D)
to
g
e
t
ad
v
a
n
t
a
g
es
o
f
t
h
e
G
MM
[
3
2
]
,
[
3
3
]
.
Alg
o
r
ith
m
1
:
Fo
r
v
id
e
o
in
p
u
t
-
o
u
tp
u
t in
itializatio
n
1.
Input: V
f
2.
Output: D,B
3.
Begin
4.
Initialization of players
a.
P1
foreground Mask
b.
P2
V
f
5.
D
ffd
(
Ng,NTf,MBr)
6.
B
fba(BOp, AOp, COp, MBa)
7.
End
2
.
2
.
Co
m
pu
t
a
t
io
n
m
ea
s
ures o
f
bin
a
ry
la
rg
e
o
bje
ct
T
h
e
id
ea
o
f
GM
M
p
lay
s
a
cr
u
cial
r
o
le
to
in
f
lu
en
ce
th
e
o
u
tco
m
e
o
f
b
ac
k
g
r
o
u
n
d
s
u
b
tr
a
ctio
n
f
o
r
th
e
d
etec
tio
n
o
f
m
o
v
in
g
o
b
jects.
T
h
e
id
ea
o
f
b
ac
k
g
r
o
u
n
d
s
u
b
tr
a
ctio
n
allo
ws
in
d
etec
tin
g
th
e
m
o
v
in
g
o
b
jects
f
r
o
m
d
y
n
am
ic
v
id
eo
s
ce
n
es
. W
h
ich
is
ap
p
lied
in
th
is
p
r
o
p
o
s
ed
s
tu
d
y
co
n
s
id
er
i
n
g
GM
M.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
3
8
7
9
-
3
8
9
1
3884
I
d
ea
o
f
GM
M:
I
t
h
as
b
ee
n
o
b
s
er
v
ed
th
at
d
if
f
er
e
n
t
b
ac
k
g
r
o
u
n
d
o
b
jects
co
u
ld
m
o
r
e
li
k
ely
ap
p
ea
r
at
th
e
s
am
e
p
ix
el
lo
ca
tio
n
o
f
o
v
e
r
a
s
p
ec
if
ic
p
er
io
d
o
f
tim
e.
T
h
is
ar
is
es
a
ch
allen
g
e
o
f
s
in
g
le
-
v
alu
ed
b
ac
k
g
r
o
u
n
d
m
o
d
el.
Sev
er
al
r
esear
ch
e
r
s
talk
s
ab
o
u
t th
e
d
esig
n
an
d
m
o
d
elin
g
o
f
m
u
lti
-
v
alu
ed
b
ac
k
g
r
o
u
n
d
m
o
d
el
wh
ich
ca
n
ea
s
ily
co
p
e
with
th
e
m
u
ltip
le
b
ac
k
g
r
o
u
n
d
o
b
jects
ap
p
ea
r
i
n
g
in
v
i
d
eo
s
ce
n
es
[
3
4
]
,
[
3
5
]
.
T
h
e
m
o
d
el
p
r
o
v
id
es
b
etter
d
escr
ip
tio
n
o
f
b
o
th
f
o
r
eg
r
o
u
n
d
a
n
d
b
ac
k
g
r
o
u
n
d
v
al
u
es
b
y
d
escr
ib
in
g
th
e
p
r
o
b
ab
ilit
y
o
f
o
b
s
er
v
in
g
a
ce
r
tain
p
ix
el
v
alu
e
(
x
t
)
at
a
s
p
ec
if
ic
tim
e
o
f
(
t)
.
T
h
e
m
et
h
o
d
GM
M
co
m
p
u
tes
ea
ch
p
ix
el
with
in
a
tem
p
o
r
al
win
d
o
w
(
w
)
co
n
s
id
er
in
g
k
n
u
m
b
er
o
f
m
ix
tu
r
es
o
f
eith
e
r
s
in
g
le
o
r
m
u
lti
-
d
im
e
n
s
io
n
al
Ga
u
s
s
ian
d
is
tr
ib
u
tio
n
.
Her
e
if
th
e
v
alu
e
o
f
k
is
lar
g
er
th
at
ten
d
s
to
s
tr
o
n
g
er
ab
ilit
y
to
d
ea
l
with
th
e
d
is
tu
r
b
an
ce
b
ac
k
g
r
o
u
n
d
.
I
f
th
e
s
eq
u
en
ce
is
o
b
s
er
v
ed
with
=
{
1
,
2
…
.
}
f
o
r
a
g
iv
en
p
ix
el.
T
h
e
n
th
e
p
r
o
b
ab
ilit
y
co
m
p
u
tatio
n
f
o
r
o
b
s
er
v
in
g
a
cu
r
r
en
t p
ix
el
v
alu
e
at
tim
e
t c
an
b
e
r
ep
r
esen
ted
with
th
e
f
o
llo
win
g
m
ath
em
ati
ca
l
(
1
)
.
(
)
=
∑
,
=
1
(
,
,
,
,
)
(
2
)
Her
e
k
r
ep
r
esen
ts
th
e
n
u
m
b
e
r
o
f
g
au
s
s
ian
d
is
tr
ib
u
tio
n
s
wh
ich
r
ep
r
esen
ts
d
escr
ip
tio
n
f
o
r
o
n
e
o
f
t
h
e
o
b
s
er
v
ab
le
f
o
r
e
g
r
o
u
n
d
o
r
b
ac
k
g
r
o
u
n
d
o
b
jects.
I
n
p
r
ac
tical
i
n
s
tan
ce
s
k
v
alu
e
is
lik
ely
to
b
e
r
esid
e
with
in
t
h
e
r
an
g
e
o
f
3
≤
≤
5
.
T
h
e
co
m
p
u
tatio
n
o
f
Gau
s
s
ian
s
r
em
ain
s
m
u
lti
-
v
a
r
iate
f
o
r
th
e
p
u
r
p
o
s
e
o
f
d
escr
i
b
in
g
th
e
r
ed
,
g
r
ee
n
,
an
d
b
l
u
e
v
al
u
es.
H
er
e
μ
i
,
t
r
ef
er
s
to
t
h
e
co
m
p
u
tatio
n
c
o
r
r
esp
o
n
d
to
th
e
m
ea
n
v
alu
e
o
f
ith
g
au
s
s
ian
in
th
e
m
ix
tu
r
e
o
f
m
o
d
els
at
th
e
in
s
tan
ce
o
f
t.
Als
o
Σ
i
,
t
co
m
p
u
ta
tio
n
d
en
o
tes
th
e
co
v
a
r
ian
ce
m
atr
ix
o
f
th
e
ith
g
au
s
s
ian
at
th
e
tim
e
t.
I
t
h
as
t
o
b
e
n
o
ted
th
at
h
er
e
k
is
d
eter
m
in
ed
co
n
s
id
er
in
g
th
e
co
m
p
u
tin
g
asp
ec
ts
o
f
b
o
th
m
em
o
r
y
a
n
d
co
m
p
u
tatio
n
al
p
o
wer
.
Her
e
th
e
esti
m
atio
n
o
f
ω
i
,
t
also
d
en
o
tes
th
e
f
ac
to
r
o
f
w
eig
h
t
ass
o
ciate
d
with
ith
Gau
s
s
ian
in
th
e
tim
e
in
s
tan
ce
o
f
t.
T
h
e
p
r
in
cip
l
e
h
er
e
f
o
llo
ws
th
at
th
e
f
ac
to
r
∑
ω
i
,
t
k
i
=
1
=
1
an
d
η
(
x
t
,
μ
i
,
t
,
Σ
i
,
t
)
co
n
s
id
er
ed
t
o
b
e
Gau
s
s
ian
p
r
o
b
ab
ilit
y
d
e
n
s
ity
f
u
n
ctio
n
.
η
(
x
t
,
μ
i
,
t
,
Σ
i
,
t
)
=
1
2π
n
2
⁄
|
Σ
|
1
/
2
e
−
1
2
⁄
(
x
t
−
μ
t
)
T
Σ
−
1
(
x
t
−
μ
t
)
(
3
)
T
h
e
s
y
s
tem
m
o
d
elin
g
also
co
n
s
id
er
s
th
e
b
en
ef
icial
f
ea
tu
r
es
ass
o
ciate
d
with
GM
M.
T
h
e
b
ac
k
g
r
o
u
n
d
m
o
d
elin
g
o
f
a
g
r
ay
s
ca
le
im
ag
e
co
n
s
id
er
s
th
e
v
al
u
e
o
f
n
=1
a
n
d
Σ
i
,
t
=
2
,
.
Ho
wev
er
also
wh
e
n
th
e
m
o
d
elin
g
is
ap
p
lied
o
n
an
R
GB
co
m
p
o
n
en
ts
th
en
,
it
u
p
d
ates
th
e
v
alu
es
o
f
n
=3
an
d
Σ
i
,
t
=
2
,
.
T
h
is
co
m
p
u
tatio
n
o
f
Σ
i
,
t
=
2
,
b
asically
ass
u
m
es
th
e
f
o
r
m
o
f
co
v
ar
ian
ce
m
at
r
ix
.
Ad
d
itio
n
ally
,
th
e
s
y
s
tem
ev
alu
ates
th
e
in
co
m
in
g
f
r
am
es
in
r
ea
l
tim
e,
an
d
GM
M
m
o
d
if
ies
its
p
ar
a
m
eter
s
in
s
tep
-
by
-
s
tep
r
esp
o
n
s
e
to
th
e
ch
an
g
in
g
p
ix
el
v
alu
e.
Ad
d
itio
n
ally
,
th
e
p
ix
els
ar
e
m
ap
p
ed
u
s
in
g
a
th
r
esh
o
ld
in
g
ap
p
r
o
ac
h
an
d
th
e
Gau
s
s
ian
m
o
d
el.
T
h
e
s
y
s
tem
f
u
r
th
er
m
o
d
if
ies
th
e
weig
h
ts
o
f
t
h
e
Gau
s
s
ian
co
m
p
o
n
en
ts
if
a
m
atch
is
id
e
n
tifi
ed
.
T
h
is
is
h
o
w
t
h
e
b
ac
k
g
r
o
u
n
d
m
o
d
el
esti
m
atio
n
ac
co
r
d
in
g
t
o
th
e
d
is
tr
i
b
u
tio
n
s
is
ca
r
r
ied
o
u
t,
an
d
b
ac
k
g
r
o
u
n
d
p
ix
el
ca
teg
o
r
izatio
n
is
p
o
s
s
ib
le.
T
h
e
f
u
n
ctio
n
alities
d
ef
in
e
d
in
th
e
m
o
d
elin
g
o
f
ffd
(
N
g
,
NT
f
,
MBr
)
b
asically
aim
s
to
f
o
r
m
th
e
f
o
r
eg
r
o
u
n
d
d
etec
t
o
r
co
n
s
id
er
in
g
ef
f
ec
tiv
e
s
eg
m
en
tatio
n
o
f
b
a
ck
g
r
o
u
n
d
s
u
b
t
r
a
ctio
n
.
T
h
e
f
o
r
m
atio
n
o
f
th
e
f
o
r
e
g
r
o
u
n
d
d
etec
tio
n
o
b
ject
b
asically
en
ab
les
th
e
p
o
t
en
tial
f
ea
tu
r
es
o
f
GM
M
in
wh
ich
it
co
m
p
a
r
es
th
e
co
lo
r
o
r
g
r
a
y
s
ca
le
v
id
eo
f
r
am
e
with
a
b
ac
k
g
r
o
u
n
d
m
o
d
el
as d
is
cu
s
s
ed
in
th
e
(
2
)
an
d
(
3
)
.
T
h
is
co
m
p
u
tatio
n
al
p
r
o
ce
s
s
en
ab
les
a
class
if
icatio
n
cr
iter
io
n
to
u
n
d
e
r
s
tan
d
wh
eth
er
a
ce
r
tain
p
ix
el
b
elo
n
g
s
to
a
p
ar
t
o
f
b
ac
k
g
r
o
u
n
d
o
r
f
o
r
e
g
r
o
u
n
d
.
T
h
is
co
m
p
u
tatio
n
al
p
r
o
ce
s
s
is
ess
en
tial
f
o
r
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
alg
o
r
ith
m
s
as
th
is
d
ata
ex
p
lo
r
atio
n
a
n
d
p
r
e
-
p
r
o
c
ess
in
g
s
tag
e
also
h
elp
s
elim
in
atin
g
th
e
r
ed
u
n
d
an
t
attr
ib
u
te
f
r
o
m
th
e
d
ata
a
n
d
m
ak
e
it
s
u
itab
le
f
o
r
f
u
r
th
e
r
c
o
m
p
u
tatio
n
al
an
aly
s
is
with
tr
u
th
f
u
l,
ac
cu
r
ate
a
n
d
co
m
p
lete
in
f
o
r
m
atio
n
ab
o
u
t
t
h
e
f
o
r
eg
r
o
u
n
d
o
b
ject.
Her
e
th
e
f
o
r
eg
r
o
u
n
d
m
ask
(
M
f
)
is
co
m
p
u
ted
wh
ich
is
ass
o
ciate
d
with
th
e
D.
An
d
th
e
alg
o
r
ith
m
c
o
r
r
esp
o
n
d
to
b
ac
k
g
r
o
u
n
d
s
u
b
tr
ac
tio
n
h
er
e
e
f
f
ic
ien
tly
co
m
p
u
tes
th
e
f
o
r
eg
r
o
u
n
d
o
b
jects
(
O
f
)
f
r
o
m
th
e
f
r
am
e
s
eq
u
en
ce
o
f
th
e
V
f.
a
n
o
th
e
r
ex
p
licit
f
u
n
ctio
n
f
o
r
th
e
p
u
r
p
o
s
e
o
f
an
aly
zin
g
th
e
p
r
o
p
er
ties
o
f
c
o
n
n
ec
ted
r
eg
io
n
s
is
b
ein
g
u
s
ed
as
f
u
n
ctio
n
f
o
r
B
lo
b
An
aly
s
er
(
f
b
a)
t
h
at
tak
es
p
ar
am
eter
s
as
in
s
et
{Po
r
t
f
o
r
th
e
b
o
u
n
d
in
g
b
o
x
(
B
o
p
)
,
p
o
r
t
f
o
r
o
u
tp
u
t
ar
ea
(
AOp
)
,
Po
r
t
f
o
r
o
u
tp
u
t
ce
n
t
r
o
id
(
C
Op
,
Min
im
u
m
b
lo
b
a
r
ea
(
MBa)
}
th
at
y
ield
th
e
b
lo
b
(
B
)
.
T
h
e
u
n
d
er
ly
in
g
id
ea
b
e
h
in
d
B
lo
b
an
aly
s
is
is
to
ex
p
lo
r
e
th
e
s
tatis
tics
f
o
r
lab
elled
r
eg
io
n
i
n
th
e
b
i
n
ar
y
f
r
am
e
o
f
th
e
v
id
eo
s
eq
u
en
ce
.
I
t
b
asically
h
elp
s
s
eg
m
en
tin
g
th
e
o
b
jects f
r
o
m
th
e
v
id
eo
s
eq
u
en
ce
.
T
h
e
d
escr
i
p
tio
n
o
f
th
e
B
lo
b
an
al
y
s
is
ca
n
b
e
s
ee
n
in
Fig
u
r
e
3
.
T
h
e
m
eth
o
d
o
f
B
o
b
a
n
aly
s
is
b
asically
r
ef
er
s
to
an
aly
zin
g
th
e
s
h
ap
e
f
ea
tu
r
es
ass
o
ciate
d
wi
th
o
b
jects.
Her
e
th
e
im
p
licatio
n
s
o
f
th
e
m
eth
o
d
B
o
b
an
aly
s
is
b
asically
id
en
tify
th
e
g
r
o
u
p
o
f
c
o
n
n
e
cted
p
ix
els
wh
ich
ar
e
m
o
r
e
lik
ely
r
elate
d
with
th
e
m
o
v
in
g
o
b
ject.
T
h
e
id
ea
o
f
B
o
b
a
n
aly
s
is
is
to
ex
p
l
o
r
es
th
e
p
ix
els
co
n
n
ec
tiv
ity
an
d
co
n
s
tr
u
ct
th
e
B
lo
b
th
r
o
u
g
h
th
e
f
u
n
cti
o
n
f
b
a
(
x
)
.
T
h
e
co
n
n
ec
tiv
ity
am
o
n
g
th
e
p
i
x
els
is
r
ep
r
esen
ted
with
B
lo
b
.
Firstl
y
,
th
e
p
r
o
ce
s
s
co
m
p
u
tes
th
e
s
tatis
tics
as
s
o
ciate
d
with
b
lo
b
an
d
f
u
r
th
e
r
a
n
aly
s
e
th
e
in
f
o
r
m
atio
n
o
f
B
lo
b
wh
ich
co
r
r
esp
o
n
d
to
g
e
o
m
etr
ic
ch
ar
ac
ter
is
tics
wh
ich
i
n
clu
d
e
p
o
in
ts
o
f
b
o
r
d
er
lin
e
,
a
n
d
p
e
r
im
eter
.
T
h
ese
id
ea
s
an
d
th
e
s
tan
d
ar
d
m
eth
o
d
s
ar
e
f
u
r
th
e
r
in
co
r
p
o
r
ated
in
d
esig
n
in
g
th
e
o
b
ject
d
etec
tio
n
an
d
tr
ac
k
i
n
g
m
eth
o
d
o
l
o
g
ies in
th
e
p
r
o
p
o
s
ed
s
y
s
tem
’
s
co
n
tex
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2
2
5
2
-
8
9
3
8
Desig
en
o
f a
n
effec
tive
mu
ltip
l
e
o
b
ject
tr
a
ck
in
g
fr
a
mewo
r
k
fo
r
d
yn
a
mic
vid
eo
s
ce
n
es
(
K
a
r
a
n
a
m
S
u
n
il K
u
ma
r
)
3885
Fig
u
r
e
3
.
B
lo
b
a
n
aly
s
is
d
escr
ip
tio
n
I
n
th
e
co
m
p
u
tatio
n
o
f
s
tatis
tics
b
lo
b
,
th
e
s
y
s
tem
an
aly
s
es
th
e
o
u
tp
u
t
o
f
AOp
wh
ic
h
r
e
p
r
esen
ts
a
v
ec
to
r
o
f
p
ix
els
in
th
e
lab
ele
d
r
e
g
io
n
s
.
Her
e
C
Op
r
ef
er
s
t
o
an
N
-
by
-
2
m
atr
ix
o
f
ce
n
tr
o
id
co
o
r
d
in
ates
c
(
x
,
y
)
wh
ich
co
u
ld
b
e
r
ep
r
esen
ted
with
th
e
f
o
llo
win
g
m
atr
ix
(
3
)
.
Her
e
N
r
ep
r
esen
ts
th
e
n
u
m
b
er
o
f
B
lo
b
s
.
Her
e
[
x
,
y
]
r
ep
r
esen
ts
th
e
ce
n
tr
o
id
co
o
r
d
i
n
ates.
Her
e
[
x
1
,y
1
]
➔
[x
N
y
N
]
im
p
lies
th
at
th
er
e
ar
e
two
b
lo
b
s
th
en
t
h
e
r
o
w
an
d
co
lu
m
n
c
o
o
r
d
i
n
ates o
f
th
eir
ce
n
tr
o
id
s
ar
e
[
x
1
,y
1
]
a
n
d
[
x
N
y
N
]
r
esp
ec
tiv
ely
.
C
Op
=
[
x
1
y
1
x
N
y
N
]
(
4
)
T
h
e
p
r
o
ce
s
s
o
f
co
m
p
u
tatio
n
f
o
r
th
e
m
ea
s
u
r
e
o
f
B
lo
b
(
B
)
also
an
aly
s
e
th
e
p
ar
am
eter
M
B
a
wh
ich
r
ef
er
s
to
an
o
th
er
N
-
by
-
4
m
atr
ix
wh
ich
is
o
f
[
x
,
y
]
d
im
e
n
s
io
n
.
Her
e
also
N
r
ep
r
esen
ts
th
e
n
u
m
b
er
o
f
b
lo
b
s
wh
er
ea
s
[
x
,
y
]
d
en
o
tes
th
e
u
p
p
er
lef
t
c
o
r
n
er
o
f
t
h
e
b
o
u
n
d
in
g
b
o
x
.
T
h
e
an
al
y
s
is
o
f
th
e
b
lo
b
c
o
n
s
id
er
in
g
s
tatis
t
ics r
etu
r
n
s
a
b
lo
b
an
aly
s
is
s
y
s
tem
o
b
ject
(
B
)
.
T
h
e
an
aly
s
is
o
f
B
also
co
n
s
tr
u
cts th
e
s
i
g
n
if
ican
t p
r
o
p
er
ties
o
f
ce
n
tr
o
id
,
b
o
u
n
d
in
g
b
o
x
,
la
b
el
m
atr
ix
a
n
d
b
lo
b
co
u
n
t
in
th
e
o
u
tp
u
t
wh
ich
ar
e
r
ef
er
en
c
ed
with
B
.
Fin
ally
,
th
is
co
m
p
u
tatio
n
p
r
o
ce
s
s
ex
tr
a
cts th
e
s
h
ap
e
f
ea
tu
r
es o
f
th
e
o
b
jects o
f
in
ter
est f
r
o
m
th
e
v
id
e
o
s
eq
u
en
ce
.
2
.
3
.
I
nitia
liza
t
io
n o
f
t
he
t
ra
c
k
ing
mo
du
le
T
h
e
f
o
r
m
u
late
d
d
esig
n
o
f
th
e
d
y
n
am
ic
tr
ac
k
in
g
m
o
d
el
f
u
r
th
er
co
n
s
tr
u
cts
an
em
p
ty
s
tr
u
ctu
r
e
ar
r
ay
o
f
tr
ac
k
in
g
m
o
d
u
le
with
s
ix
d
if
f
er
en
t
f
ield
s
.
W
h
ich
co
u
ld
b
e
s
h
o
wn
with
th
e
Fig
u
r
e
4
.
T
h
e
s
tr
u
ctu
r
e
ar
r
ay
b
asically
in
itializes
s
ix
d
if
f
er
en
t
f
ield
s
s
u
ch
as
(
ID
)
,
Kalm
ar
f
ilter
(
KF
)
,
Ag
e
(
a)
,
b
o
u
n
d
in
g
b
o
x
(
Bx
)
,
to
tal
v
is
ib
le
co
u
n
t m
ea
s
u
r
e
(
tVC
)
,
a
n
d
co
n
s
ec
u
tiv
e
in
v
is
ib
le
co
u
n
t
m
ea
s
u
r
e
(
cI
C
)
.
Fig
u
r
e
4
.
Stru
ctu
r
e
ar
r
a
y
f
ield
s
o
f
T
h
e
s
y
s
tem
also
f
o
r
m
u
lates
a
f
u
n
ctio
n
ality
to
in
itiate
th
e
s
tr
u
ctu
r
e
f
o
r
in
itializatio
n
o
f
ar
r
ay
o
f
tr
ac
k
s
.
Her
e
ea
ch
in
d
iv
id
u
al
t
r
ac
k
∈
.
Her
e
ea
ch
tr
ac
k
r
ep
r
es
en
ts
th
e
s
tr
u
ctu
r
e
co
r
r
esp
o
n
d
in
g
to
th
e
m
o
v
in
g
o
b
ject
ap
p
ea
r
in
g
in
th
e
V
f
.
T
h
e
d
esig
n
r
eq
u
i
r
em
en
t
f
o
r
th
e
tr
ac
k
in
g
m
o
d
u
le
in
th
e
p
r
o
p
o
s
ed
m
o
v
in
g
o
b
ject
d
etec
tio
n
a
n
d
t
r
ac
k
in
g
s
tr
ateg
y
is
to
f
o
r
m
u
late
th
e
s
tr
u
ctu
r
e
f
ield
s
in
s
u
ch
a
wa
y
s
o
th
at
th
e
s
tate
o
f
th
e
tr
ac
k
ed
o
b
ject
(
)
ca
n
b
e
m
ai
n
tain
ed
ap
p
r
o
p
r
iately
.
Her
e
r
ef
er
s
to
th
e
in
teg
er
I
D
o
f
th
e
tr
ac
k
,
r
ep
r
esen
ts
th
e
cu
r
r
en
t
b
o
u
n
d
in
g
b
o
x
ass
o
ciate
d
with
th
e
o
b
je
ct.
r
ep
r
esen
ts
a
Kalm
an
f
ilter
o
b
ject
wh
ich
is
u
s
ed
f
o
r
m
o
tio
n
-
b
ased
tr
ac
k
i
n
g
.
r
ef
er
s
t
o
th
e
f
r
am
e
co
u
n
t
s
in
ce
th
e
f
i
r
s
t
d
etec
tio
n
o
f
.
T
h
e
co
n
s
ec
u
tiv
e
v
is
ib
le
co
u
n
t
m
ea
s
u
r
e
r
ef
er
s
to
th
e
n
u
m
b
er
o
f
f
r
a
m
es
in
wh
ich
th
e
tr
ac
k
was
d
etec
ted
.
r
ep
r
esen
ts
th
e
n
u
m
b
er
o
f
co
u
n
ts
o
f
co
n
s
ec
u
ti
v
e
f
r
am
es
f
o
r
wh
ic
h
th
e
tr
ac
k
was
n
o
t
d
etec
ted
.
T
h
e
p
r
o
ce
s
s
o
f
co
m
p
u
tatio
n
o
f
s
tate
co
r
r
esp
o
n
d
t
o
th
e
in
f
o
r
m
atio
n
u
tili
ze
d
f
o
r
d
etec
tio
n
o
f
t
r
ac
k
allo
ca
tio
n
,
t
r
ac
k
ex
p
ir
y
a
n
d
d
is
p
lay
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
3
8
7
9
-
3
8
9
1
3886
2
.
4
.
O
bje
c
t
det
ec
t
io
n
m
o
du
le
T
h
e
co
m
p
u
tin
g
p
r
o
ce
s
s
f
u
r
th
e
r
co
n
s
id
er
s
id
e
n
tific
atio
n
n
u
m
b
er
o
f
th
e
n
e
x
t
tr
ac
k
(
)
an
d
in
i
tiates
th
e
p
r
o
ce
s
s
o
f
d
etec
tin
g
m
o
v
in
g
o
b
jects
co
n
s
id
er
in
g
a
lo
g
ical
f
u
n
ctio
n
hF
(
x
)
:
∀
x
∈
V
f
.
Her
e
th
e
f
u
n
ctio
n
ℎ
(
)
is
a
lo
g
ical
f
u
n
ctio
n
wh
ich
c
o
n
s
id
er
s
a
s
et
o
f
o
b
jects
ass
o
ciate
d
with
th
e
v
id
e
o
f
ile
(
)
to
r
ea
d
.
T
h
e
f
u
n
ctio
n
b
asically
r
etu
r
n
s
a
lo
g
ical
v
alu
e
f
r
o
m
th
e
s
et
o
f
→
{
1
,
0
}
.
I
f
th
e
f
u
n
ctio
n
hF
(
x
)
r
et
u
r
n
s
t
h
e
v
alu
e
1
th
at
im
p
lies
th
at
th
er
e
is
a
v
id
eo
f
r
am
e
av
ailab
le
to
r
ea
d
.
T
h
e
p
r
o
ce
s
s
f
u
r
th
er
also
ap
p
lies
an
o
th
er
f
u
n
ctio
n
o
f
rF
(
x
)
wh
ich
r
ea
d
s
th
e
v
id
eo
f
r
a
m
e
f
r
o
m
t
h
e
f
ile
th
en
th
e
p
r
o
ce
s
s
f
u
r
th
er
d
etec
ts
th
e
b
in
ar
y
m
ask
(
)
f
r
o
m
th
e
.
T
h
e
b
in
ar
y
m
ask
is
o
f
s
am
e
s
ize
o
f
th
e
in
p
u
t
.
Her
e
th
e
r
ea
d
in
g
o
f
th
e
f
r
am
e
co
n
s
id
er
s
co
n
s
tr
u
ctin
g
o
f
s
y
s
tem
o
b
ject
(
ob
j
)
.
T
h
e
p
r
o
ce
s
s
o
f
d
etec
tin
g
o
b
jects
f
r
o
m
th
e
en
a
b
les
an
o
t
h
er
ex
p
licit
f
u
n
ctio
n
o
f
(
)
,
h
er
e
th
e
f
u
n
ct
io
n
co
n
s
id
er
s
th
e
i
n
p
u
t
o
f
an
d
p
r
o
ce
s
s
it
to
g
en
e
r
ate
th
r
ee
d
is
tin
ct
attr
ib
u
tes
wh
ich
ar
e
{
,
,
}
.
Her
e
c
r
ef
er
s
to
th
e
ce
n
tr
o
id
ca
lcu
lati
o
n
co
n
s
id
er
i
n
g
th
e
d
etec
ted
o
b
jects,
B
x
is
b
o
u
n
d
in
g
b
o
x
o
f
co
m
p
u
tat
io
n
o
f
t
h
e
d
etec
ted
o
b
ject
f
o
l
lo
wed
b
y
th
e
m
ea
s
u
r
e
o
f
m
a
s
k
(
m
)
.
T
h
e
in
itial
co
m
p
u
tatio
n
o
f
t
h
e
f
u
n
ctio
n
(
)
co
n
s
id
er
s
th
e
v
id
e
o
f
r
a
m
e
s
eq
u
en
ce
o
f
an
d
id
en
tif
y
th
e
m
a
s
k
an
d
co
m
p
u
tes
a
lo
g
ical
m
atr
ix
(
,
)
.
Her
e
th
e
c
o
m
p
u
tin
g
f
u
n
ctio
n
o
f
b
in
a
r
y
m
ask
c
o
m
p
u
tatio
n
b
asically
p
er
f
o
r
m
s
m
o
tio
n
s
eg
m
en
tatio
n
co
n
s
id
er
in
g
a
n
ex
p
licit
m
eth
o
d
o
f
ffd
(
)
[
3
2
]
.
T
h
e
f
o
llo
wi
n
g
an
aly
tical
alg
o
r
ith
m
,
Alg
o
r
ith
m
2
,
b
as
ically
m
o
d
eled
to
p
r
esen
t
t
h
e
p
r
o
p
o
s
ed
wo
r
k
-
f
lo
w
ass
o
ciate
d
with
o
b
ject
d
etec
tio
n
f
r
o
m
v
i
d
eo
wh
er
e
th
e
ad
v
an
tag
eo
u
s
f
ac
to
r
s
o
f
th
e
m
eth
o
d
GM
M
u
s
u
tili
ze
d
to
p
er
f
o
r
m
b
lo
b
an
aly
s
is
.
T
h
e
co
m
p
u
ted
m
ask
f
u
r
th
er
u
n
d
e
r
g
o
es
th
r
o
u
g
h
p
r
e
-
p
r
o
ce
s
s
in
g
o
p
e
r
atio
n
s
as
d
ef
in
ed
b
y
m
o
r
p
h
o
lo
g
ical
o
p
e
r
atio
n
s
.
T
h
e
m
o
r
p
h
o
lo
g
ical
o
p
e
r
atio
n
h
e
r
e
s
u
b
jecte
d
to
elim
in
ate
r
ed
u
n
d
an
t
attr
ib
u
tes
o
f
p
ix
els
an
d
also
f
ill
t
h
e
m
is
s
in
g
g
a
p
s
in
t
h
e
b
lo
b
s
f
o
r
th
e
r
e
s
u
ltin
g
m
ask
.
T
h
e
p
r
o
ce
s
s
f
u
r
th
er
p
er
f
o
r
m
s
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
(
)
o
v
er
(
,
)
.
I
t
ap
p
lies
two
f
u
n
ctio
n
s
s
u
ch
as
1
an
d
2
to
p
e
r
f
o
r
m
th
e
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
wh
er
e
1
o
p
en
s
th
e
[
]
an
d
p
e
r
f
o
r
m
s
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
o
v
e
r
it
with
r
esp
ec
t
to
s
tr
u
ctu
r
in
g
elem
en
t
o
f
s
ize
[
×
]
an
d
u
p
d
ate
th
e
v
alu
e
s
o
f
.
T
h
e
p
r
o
ce
s
s
also
f
u
r
th
er
ap
p
lies
an
o
th
er
f
u
n
ctio
n
o
f
2
f
o
r
m
o
r
p
h
o
lo
g
ical
clo
s
e
o
p
er
atio
n
o
v
e
r
co
n
s
id
er
in
g
d
ilatio
n
f
o
llo
w
ed
b
y
e
r
o
s
io
n
[
3
3
]
.
Fin
ally
,
an
o
th
er
f
u
n
ctio
n
o
f
3
h
elp
s
f
illi
n
g
th
e
im
ag
e
r
eg
io
n
s
a
n
d
g
ap
s
an
d
m
a
k
e
th
e
u
p
d
ated
s
u
itab
le
f
o
r
ef
f
ec
tiv
e
b
lo
b
a
n
aly
s
is
.
T
h
e
cu
s
to
m
ized
f
u
n
c
tio
n
o
f
dO
(
x
)
f
in
ally
r
etu
r
n
s
th
r
ee
attr
ib
u
tes
o
f
{
c
,
Bx
,
m
}
an
d
ter
m
in
ates th
e
p
r
o
ce
s
s
o
f
ex
ec
u
tio
n
.
Alg
o
r
ith
m
2
: Fo
r
o
b
ject
d
etec
tio
n
f
r
o
m
v
id
e
o
Input:
V
f
Output:
{
c
,
Bx
,
m
}
Begin
1. Define:
dO
(
x
), construct system object
(
o
b
j
)
2. Define:
hF
(
x
)
:
∀
x
∈
V
f
3. While (
F
i
=
1
)
4.
rF
(
x
)
→
F
i
5. End
6. Return:
l
→
{
1
,
0
}
7.
Bm
[
Lm
]
←
ffd
(
x
)
:
∀
F
i
,
Lm
(
r
,
c
)
7.
MO
→
Bm
[
Lm
(
r
,
c
)
]
, for
{
I
1
,
I
2
,
I
3
}
8. Apply
f
b
a
(
x
)
:
∀
x
∈
Bm
(1), (2) for GMM
9. Return
{
c
,
Bx
,
m
}
End
2
.
5
.
P
re
dict
io
n
m
o
du
le
f
o
r
n
ew
po
s
it
io
n o
f
lin
e
o
f
mo
v
ement
T
h
e
co
r
e
s
tr
ateg
y
d
ev
elo
p
e
d
in
th
e
p
r
o
p
o
s
ed
s
y
s
tem
tar
g
ets
ap
p
r
o
p
r
iate
id
en
tific
atio
n
an
d
tr
ac
k
in
g
o
f
m
o
b
ile
o
b
jects
f
r
o
m
a
co
m
p
lex
s
et
o
f
s
ce
n
es.
Her
e
th
e
s
ce
n
es
ar
e
ca
p
tu
r
ed
f
r
o
m
a
ca
m
er
a
wh
ich
is
m
o
u
n
ted
i
n
s
tatic
p
o
s
itio
n
.
T
h
e
f
o
r
m
u
lated
tr
ac
k
in
g
m
o
d
u
le
f
u
r
th
er
c
o
n
s
id
er
s
∈
an
d
ap
p
ly
a
f
u
n
ctio
n
o
f
(
)
o
v
er
th
e
tr
ac
k
s
with
th
e
in
clu
s
io
n
o
f
Kalm
an
f
ilter
ap
p
r
o
ac
h
to
p
r
e
d
ict
th
e
n
ew
lo
ca
tio
n
o
f
th
e
L
o
M
.
Her
e
th
e
s
y
s
tem
co
n
s
id
er
s
th
e
co
m
p
u
tatio
n
o
f
B
x
c
o
n
s
id
er
i
n
g
th
e
u
p
d
ates
o
n
f
o
r
L
o
M
a
n
d
f
u
r
th
er
in
itially
p
r
ed
ict
an
d
esti
m
ate
th
e
cu
r
r
e
n
t
lo
ca
tio
n
o
f
th
e
tr
ac
k
o
f
L
o
M
co
n
s
id
er
in
g
th
e
f
u
n
ctio
n
o
f
(
)
it
o
p
tim
izes
th
e
p
r
o
ce
s
s
o
f
p
r
ed
ictio
n
o
f
ce
n
tr
o
id
co
n
s
id
er
in
g
th
e
ap
p
r
o
ac
h
Kalm
an
Fil
ter
(
)
.
T
h
e
c
o
m
p
u
tatio
n
p
r
o
ce
s
s
ca
n
b
e
r
ep
r
esen
ted
in
(
5
)
.
←
(
)
:
∀
∈
,
(
5
)
Her
e
th
e
co
m
p
u
tatio
n
o
f
p
r
e
d
i
ctio
n
o
f
ce
n
tr
o
id
b
asically
d
eter
m
in
es
th
e
cu
r
r
e
n
t
lo
ca
tio
n
attr
ib
u
tes
o
f
th
e
co
n
s
id
er
in
g
Kalm
a
n
f
ilter
o
b
ject.
T
h
e
f
u
r
th
er
co
m
p
u
tatio
n
co
n
s
id
er
s
s
h
if
tin
g
o
f
th
e
B
x
in
s
u
ch
a
s
o
th
at
its
ce
n
ter
lies
in
th
e
.
I
t is ac
h
i
ev
ed
with
th
e
(
6
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2
2
5
2
-
8
9
3
8
Desig
en
o
f a
n
effec
tive
mu
ltip
l
e
o
b
ject
tr
a
ck
in
g
fr
a
mewo
r
k
fo
r
d
yn
a
mic
vid
eo
s
ce
n
es
(
K
a
r
a
n
a
m
S
u
n
il K
u
ma
r
)
3887
=
−
(
)
2
⁄
(
6
)
T
h
e
f
u
n
ctio
n
f
u
r
t
h
er
u
p
d
ates
th
e
n
ew
lo
ca
tio
n
o
f
t
h
e
w
ith
r
esp
ec
t
to
th
e
L
o
M
f
o
r
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
also
ex
p
lo
r
e
s
th
e
s
h
ap
e
-
b
ased
f
ea
tu
r
es
o
f
th
e
tar
g
et
o
b
ject
wh
ich
f
u
r
th
e
r
ass
is
t
in
o
p
tim
al
esti
m
atio
n
o
f
m
o
tio
n
ass
o
ciate
d
with
th
e
id
en
tifie
d
o
b
jec
t
o
n
its
L
o
M.
T
h
e
n
ex
t
co
m
p
u
tatio
n
al
p
r
o
ce
s
s
p
er
f
o
r
m
s
L
o
M
allo
ca
tio
n
t
o
th
e
id
en
tifie
d
o
b
jects o
f
in
ter
est.
2
.
6
.
L
ine o
f
m
o
v
e
m
ent
a
llo
c
a
t
io
n t
o
t
he
i
dentif
ied o
bje
ct
s
I
n
t
h
e
f
u
n
cti
o
n
a
l
m
o
d
u
le
o
f
t
h
e
p
r
o
p
o
s
ed
s
y
s
te
m
t
h
e
esti
m
a
ti
o
n
o
f
th
e
n
ew
p
o
s
i
ti
o
n
o
f
t
r
a
c
k
(
L
o
M
)
is
p
r
e
d
ic
te
d
c
o
n
s
i
d
e
r
i
n
g
t
h
e
a
p
p
r
o
a
ch
o
f
K
al
m
a
n
f
ilt
e
r
o
v
er
t
h
e
p
r
o
g
r
ess
i
v
e
∈
.
I
n
t
h
is
s
ta
g
e
o
f
c
o
m
p
u
t
ati
o
n
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
th
e
a
p
p
r
o
p
r
iat
e
a
ll
o
ca
ti
o
n
o
f
L
o
M
t
o
t
h
e
i
d
e
n
ti
f
i
ed
m
o
v
i
n
g
o
b
j
ec
ts
ta
k
e
p
l
ac
e
al
o
n
g
wi
th
th
e
co
s
t
e
v
al
u
ati
o
n
.
T
h
e
s
y
s
te
m
h
er
e
e
m
p
lo
y
s
a
n
o
t
h
e
r
f
u
n
c
tio
n
o
f
(
)
w
h
i
ch
c
o
m
p
u
tes
t
h
e
n
u
m
b
e
r
o
f
id
e
n
t
if
ie
d
o
b
j
ec
ts
f
r
o
m
t
h
e
an
d
c
o
m
p
u
te
t
h
e
c
o
s
t
o
f
ass
ig
n
m
en
t
c
o
n
s
i
d
e
r
i
n
g
t
h
e
(
7
)
.
=
(
)
:
∀
1
→
,
2
→
,
3
→
(
7
)
Fin
ally
,
th
e
o
p
tim
ized
esti
m
ato
r
o
f
th
is
f
u
n
ctio
n
s
o
lv
es
th
e
allo
ca
tio
n
p
r
o
b
lem
o
f
id
e
n
tifi
ed
o
b
jects
to
th
e
tr
ac
k
o
r
L
o
M
f
o
r
m
u
lti
o
b
ject
tr
ac
k
in
g
.
A
ls
o
co
m
p
u
te
f
o
u
r
d
if
f
er
e
n
t
attr
ib
u
tes
s
u
ch
as
allo
ca
ted
L
o
M,
non
-
allo
ca
ted
L
o
M
an
d
n
o
n
-
a
llo
ca
ted
id
en
tife
d
o
b
jects.
T
h
e
Alg
o
r
ith
m
3
s
h
o
ws
th
e
d
es
ig
n
s
tr
ateg
y
o
f
t
h
e
tr
ac
k
in
g
m
o
d
u
le
wh
ic
h
h
as
g
o
t
in
f
lu
en
ce
d
f
r
o
m
th
e
[
3
6
]
,
[
3
7
]
f
o
r
s
o
lv
in
g
th
e
p
r
o
b
le
m
o
f
allo
ca
tio
n
o
f
d
etec
tio
n
s
to
tr
ac
k
s
d
u
r
in
g
m
u
ltio
b
ject
tr
ac
k
in
g
.
Alg
o
r
ith
m
3:
Fo
r
m
u
lti
-
o
b
ject
tr
ac
k
in
g
Input:
∈
Output:
Begin
1. Init
. Bx
2. Update
←
(
)
3. Compute current location of LoM
←
(
)
:
∀
∈
,
(5)
4. Predict the new position of LoM
=
−
(
)
2
⁄
(6)
5. Update
with respect to the LoM for
6. LoM
Allocation to identified objects
7. Evalutate Cost
=
(
)
:
∀
1
→
,
2
→
,
3
→
(7)
8. Update
allocated LoM, Non
-
Allocated LoM
9. Eliminate Missed LoM, Construct New LoM
10. Exibit
Final Tracked Objects (
)
End
On
ce
th
e
co
s
t
ev
alu
atio
n
m
etr
i
c
is
co
m
p
u
ted
f
o
r
s
o
lv
in
g
th
e
ass
ig
n
m
en
t
p
r
o
b
lem
,
f
u
r
th
e
r
th
e
p
r
o
ce
s
s
ex
ec
u
tes
u
p
d
atin
g
o
f
allo
ca
tio
n
o
f
L
o
M.
Her
e
t
h
e
alg
o
r
ith
m
s
tr
ateg
y
esti
m
ates
th
e
lo
ca
tio
n
o
f
t
h
e
d
etec
ted
o
b
jects
co
n
s
id
er
in
g
an
o
th
er
a
p
p
r
o
ac
h
b
ased
KF.
Her
e
th
e
KF
b
ased
m
eth
o
d
b
asically
p
e
r
f
o
r
m
s
co
r
r
ec
tio
n
o
f
th
e
m
o
v
in
g
o
b
ject’
s
lo
ca
tio
n
co
n
s
id
er
in
g
L
o
M
.
Her
e
th
e
f
i
n
etu
n
in
g
o
f
L
o
M
f
o
r
a
d
etec
t
ed
o
b
ject
also
tak
es
p
lace
wh
er
e
p
r
ed
icted
B
x
is
r
e
p
lace
d
with
th
e
d
etec
ted
B
x
.
Fin
ally
,
th
e
ag
e
co
r
r
esp
o
n
d
s
to
is
u
p
d
ated
with
v
is
ib
ilit
y
.
Fin
ally
,
th
e
p
r
o
p
o
s
e
d
alg
o
r
ith
m
s
tr
ateg
y
co
m
p
u
tes
th
e
u
p
d
ated
allo
ca
ted
L
o
M,
n
o
n
-
allo
ca
ted
L
o
M,
elim
in
ate
th
e
m
is
s
ed
L
o
M
an
d
co
n
s
tr
u
ct
n
ew
L
o
M
p
r
io
r
ex
h
ib
itin
g
th
e
attr
ib
u
te.
I
t
ca
n
b
e
s
ee
n
th
at
th
e
d
esig
n
s
tr
ateg
y
o
f
th
e
p
r
o
p
o
s
ed
MO
T
m
o
d
u
le
is
q
u
ite
s
im
p
lis
tic
an
d
less
-
iter
ativ
e
wh
ic
h
h
as
also
e
n
h
an
ce
d
th
e
co
m
p
u
tin
g
s
p
ee
d
o
f
an
a
ly
tical
o
p
er
atio
n
o
f
th
e
alg
o
r
ith
m
.
T
h
e
m
et
h
o
d
s
ar
e
c
o
m
p
u
tatio
n
ally
less
er
co
m
p
lex
wh
ic
h
p
er
f
o
r
m
t
h
e
tr
ac
k
in
g
o
p
er
atio
n
s
f
o
r
th
e
im
p
l
em
en
ted
id
ea
a
n
d
also
o
f
f
er
s
c
o
s
t
ef
f
ec
tiv
e
MO
T
.
T
h
e
n
e
x
t
s
ec
tio
n
f
u
r
th
er
d
is
cu
s
s
es
ex
p
er
im
en
tal
o
u
tco
m
e
o
b
tain
ed
f
r
o
m
th
e
s
im
u
latio
n
o
f
th
e
p
r
o
p
o
s
ed
s
tr
ateg
y
f
o
r
m
u
lti
-
o
b
ject
tr
ac
k
in
g
o
v
e
r
co
m
p
lex
v
i
d
eo
s
eq
u
e
n
ce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell,
Vo
l.
1
3
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
3
8
7
9
-
3
8
9
1
3888
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
ec
tio
n
d
is
cu
s
s
es
ab
o
u
t
th
e
s
im
u
latio
n
s
tu
d
y
o
u
tc
o
m
e
o
b
tain
e
d
f
r
o
m
im
p
lem
en
tin
g
th
e
p
r
o
p
o
s
ed
m
u
ltip
le
o
b
jects
tr
ac
k
in
g
f
r
am
ew
o
r
k
f
o
r
d
y
n
am
i
c
v
id
eo
s
ce
n
es.
T
h
e
s
tu
d
y
im
p
lem
en
tatio
n
o
f
th
e
an
aly
tical
alg
o
r
ith
m
s
is
s
cr
ip
ted
o
v
er
MA
T
L
AB
n
u
m
er
ic
al
co
m
p
u
tin
g
e
n
v
ir
o
n
m
en
t
s
u
p
p
o
r
ted
b
y
6
4
-
b
it
co
n
v
en
tio
n
al
win
d
o
ws
s
y
s
tem
.
T
h
e
s
tu
d
y
also
co
n
s
id
er
s
d
if
f
er
en
t
s
et
o
f
m
u
ltip
le
m
o
b
ile
o
b
ject
-
o
r
ien
te
d
d
atasets
as
r
ef
er
r
ed
f
r
o
m
[
3
8
]
.
I
t
h
as
to
b
e
n
o
ted
th
at
th
is
p
r
o
p
o
s
ed
s
tu
d
y
is
th
e
co
n
tin
u
ati
o
n
o
f
o
u
r
p
r
ev
io
u
s
r
esear
ch
wo
r
k
s
[
3
9
]
,
[
4
0
]
.
T
h
is
p
h
ase
o
f
th
e
s
tu
d
y
b
asi
ca
lly
ju
d
g
es
th
e
o
u
tco
m
e
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
an
d
e
x
h
ib
its
its
ef
f
ec
tiv
en
ess
in
ter
m
s
o
f
v
is
u
al
an
d
co
m
p
ar
ativ
e
p
e
r
f
o
r
m
a
n
ce
an
aly
s
is
f
r
o
m
b
o
th
ac
c
u
r
ac
y
o
f
tr
ac
k
in
g
a
n
d
co
s
t
p
o
in
t
o
f
v
iew.
T
h
e
in
itial
ex
p
er
im
e
n
tal
an
al
y
s
is
co
n
s
id
er
s
m
o
v
in
g
o
b
ject
d
etec
tio
n
an
d
tr
ac
k
i
n
g
f
o
r
a
s
in
g
le
test
o
b
ject.
I
n
th
is
r
eg
ar
d
th
e
s
y
s
tem
co
n
s
id
er
s
th
e
ca
s
e
o
f
two
-
lan
e
s
y
s
tem
o
f
r
o
a
d
way
wh
er
e
th
e
id
ea
is
to
tr
ac
k
a
s
in
g
le
m
o
v
in
g
v
e
h
icle
attem
p
tin
g
to
c
h
an
g
e
t
h
e
lan
e.
T
h
e
s
tu
d
y
co
n
s
id
er
s
tr
ac
k
in
g
o
f
a
wh
ite
an
d
a
b
lack
v
e
h
icle
wh
ich
ar
e
m
o
v
in
g
an
d
attem
p
ted
to
c
h
an
g
e
th
e
lan
e
wh
ich
is
f
u
r
th
er
s
h
o
wn
in
th
e
Fig
u
r
e
5
.
T
h
e
an
aly
s
is
an
d
in
ter
p
r
etatio
n
o
f
th
e
v
is
u
al
o
u
tco
m
e
o
f
Fig
u
r
e
5
h
ig
h
lig
h
ts
th
at
th
e
wh
ite
v
eh
icle
was
in
itially
m
o
v
in
g
o
v
er
its
ass
ig
n
ed
lef
t
lan
e
wh
er
e
it
h
as
b
ee
n
d
etec
ted
co
n
s
id
er
in
g
th
e
p
r
o
p
o
s
ed
tr
ac
k
in
g
m
o
d
u
le
Fig
u
r
e
5
(
a
)
.
Ho
wev
er
,
it h
as su
d
d
en
ly
s
h
if
ted
to
th
e
r
ig
h
t la
n
e
an
d
co
n
tin
u
ed
its
jo
u
r
n
ey
o
v
er
t
h
e
r
ig
h
t
lan
e
as
tr
ac
k
ed
b
y
th
e
p
r
o
p
o
s
e
d
tr
ac
k
in
g
m
o
d
u
le
Fig
u
r
es
5
(
b
)
-
5
(
c)
.
A
s
im
ilar
tr
ac
k
in
g
o
u
tco
m
e
is
also
f
o
u
n
d
in
th
e
ca
s
e
o
f
b
lack
v
eh
icle
w
h
ich
h
as
ch
a
n
g
ed
its
lan
e
f
r
o
m
r
ig
h
t
to
lef
t
an
d
co
n
tin
u
ed
i
ts
jo
u
r
n
ey
o
n
th
e
lef
t
lan
e
o
f
th
e
r
o
ad
wa
y
Fig
u
r
es
5
(
d
)
to
5
(
f
)
.
I
t
h
as
to
b
e
n
o
ted
t
h
at
th
e
tr
ac
k
in
g
o
f
th
e
tar
g
et
m
o
b
ile
o
b
ject
f
r
o
m
a
v
er
y
co
m
p
lex
d
y
n
a
m
ic
s
ce
n
e
i
s
ac
h
iev
ed
ef
f
ec
tiv
el
y
b
y
th
e
p
r
o
p
o
s
ed
tr
ac
k
in
g
m
o
d
u
le
ev
en
in
th
e
p
r
esen
ce
o
f
p
ar
tial
o
cc
lu
s
io
n
b
etwe
en
th
e
tar
g
et
v
e
h
icle
an
d
o
th
e
r
s
im
ila
r
v
eh
icles
o
v
er
th
e
f
r
am
e
s
eq
u
en
ce
.
T
h
e
o
u
tco
m
e
clea
r
ly
s
h
o
ws
th
at
f
o
r
a
s
in
g
le
m
o
b
ile
test
o
b
ject
th
e
p
r
o
p
o
s
ed
tr
ac
k
in
g
m
o
d
u
le
h
as
ac
h
iev
ed
h
ig
h
er
ac
cu
r
ac
y
in
tr
ac
k
in
g
th
e
f
ast
-
m
o
v
in
g
o
b
ject.
Ho
wev
er
,
th
e
p
er
f
o
r
m
a
n
ce
ass
ess
m
en
t
i
s
f
u
r
th
er
ex
ten
d
ed
f
o
r
m
u
ltip
l
e
m
o
v
in
g
o
b
jects a
s
well
wh
ich
is
f
u
r
th
er
s
h
o
wn
in
th
e
Fig
u
r
e
6
.
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
Fig
u
r
e
5
.
T
r
ac
k
in
g
o
f
a
s
in
g
le
test
o
b
ject
:
(
a)
n
o
tr
ac
k
in
g
o
f
wh
ite
v
eh
icle,
(
b
)
tr
ac
k
in
g
o
f
wh
ite
v
eh
icle
in
th
e
m
id
d
le
o
f
r
o
ad
wa
y
,
(
c)
tr
ac
k
in
g
o
f
wh
ite
v
eh
icle
in
th
e
r
ig
h
t
lan
e,
(
d
)
tr
ac
k
in
g
o
f
b
lack
v
e
h
icle
in
th
e
r
ig
h
t
lan
e,
(
e)
in
th
e
lef
t la
n
e,
an
d
(
f
)
co
n
tin
u
e
d
its
jo
u
r
n
e
y
o
n
th
e
l
ef
t la
n
e
An
o
th
er
test
in
s
tan
ce
in
th
e
p
r
o
p
o
s
ed
s
tu
d
y
m
o
d
el
is
co
n
s
id
er
ed
wh
er
e
id
en
tific
atio
n
an
d
tr
ac
k
in
g
o
f
m
u
ltip
le
m
o
b
ile
o
b
jects
ar
e
p
er
f
o
r
m
ed
co
n
s
id
er
in
g
th
e
p
r
o
p
o
s
ed
MO
T
f
r
am
ewo
r
k
.
T
h
e
Fig
u
r
e
6
clea
r
ly
s
h
o
ws
th
at
th
e
m
u
ltip
le
m
o
b
ile
o
b
jects
ar
e
d
is
tin
ctly
in
d
ex
e
d
in
itially
in
Fig
u
r
e
6
(
a)
wh
e
r
ea
s
in
th
e
s
eq
u
en
ce
o
f
o
th
e
r
f
r
a
m
es
th
e
d
etec
tio
n
a
n
d
tr
ac
k
i
n
g
is
s
lig
h
tly
af
f
ec
te
d
d
u
e
to
o
cc
lu
s
io
n
.
Ho
wev
er
,
in
Fig
u
r
es
6
(
b
)
-
6
(
d
)
m
ajo
r
ly
f
ea
tu
r
es
ar
e
p
o
s
itiv
ely
d
eter
m
in
ed
a
n
d
in
th
e
en
d
t
h
e
ac
cu
r
ac
y
o
f
tr
ac
k
in
g
also
im
p
r
o
v
ed
ir
r
esp
ec
tiv
e
o
f
th
e
p
r
esen
ce
o
f
p
ar
tial
o
c
clu
s
is
o
n
.
I
t
ca
n
also
b
e
s
ee
n
th
at
th
e
p
r
o
p
o
s
ed
s
tu
d
y
m
o
d
el
r
etain
s
a
p
r
o
p
er
b
alan
ce
b
etwe
en
th
e
p
er
f
o
r
m
an
ce
ac
cu
r
ac
y
o
f
t
r
ac
k
in
g
an
d
co
m
p
u
tatio
n
al
c
o
m
p
lex
ity
wh
ich
is
f
u
r
th
er
illu
s
tr
ated
in
th
e
f
o
llo
win
g
co
m
p
ar
ativ
e
T
ab
le
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.