I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Ro
bo
t
ics a
nd
Aut
o
m
a
t
io
n
(
I
J
RA)
Vo
l.
1
5
,
No
.
1
,
Ma
r
ch
20
2
6
,
p
p
.
7
1
~
79
I
SS
N:
2722
-
2
5
8
6
,
DOI
:
1
0
.
1
1
5
9
1
/i
jr
a
.
v
1
5
i
1
.
pp
7
1
-
79
71
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
r
a
.
ia
esco
r
e.
co
m
Sema
ntic
s
eg
men
tatio
n f
o
r da
ta v
a
lida
tion in unma
n
ned robo
tic
v
ehicles
I
v
a
n Sun
it
Ro
ut,
P
P
a
l
P
a
nd
ia
n
,
Anil Ra
j
,
Anil M
elwy
n
Reg
o
,
Sa
j
na
P
a
rim
it
a
P
a
nig
r
a
hi
D
e
p
a
r
t
me
n
t
o
f
M
e
c
h
a
n
i
c
a
l
,
A
u
t
o
m
o
b
i
l
e
,
R
o
b
o
t
i
c
s
a
n
d
M
e
c
h
a
t
r
o
n
i
c
s,
C
h
r
i
s
t
U
n
i
v
e
r
s
i
t
y
,
B
e
n
g
a
l
u
r
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
,
2
0
2
5
R
ev
is
ed
No
v
1
0
,
2
0
2
5
Acc
ep
ted
Feb
9
,
2
0
2
6
S
e
m
a
n
ti
c
se
g
m
e
n
tatio
n
is
a
v
i
tal
a
sp
e
c
t
o
f
c
o
m
p
u
ter
v
isio
n
,
wid
e
ly
u
se
d
i
n
field
s
su
c
h
a
s
a
u
t
o
n
o
m
o
u
s
d
r
iv
in
g
,
m
e
d
ica
l
ima
g
i
n
g
,
a
n
d
in
d
u
strial
a
u
to
m
a
ti
o
n
.
M
a
in
tain
i
n
g
h
ig
h
-
q
u
a
li
ty
d
a
tas
e
ts i
s c
ru
c
ial
fo
r
e
n
h
a
n
c
in
g
m
o
d
e
l
a
c
c
u
ra
c
y
a
n
d
m
in
im
izi
n
g
re
a
l
-
w
o
rl
d
e
rr
o
rs
.
T
h
i
s
p
a
p
e
r
f
o
c
u
se
s
o
n
d
e
v
e
l
o
p
i
n
g
a
c
o
m
p
re
h
e
n
siv
e
d
a
ta
v
a
li
d
a
t
io
n
p
ip
e
li
n
e
fo
r
se
m
a
n
ti
c
se
g
m
e
n
tatio
n
u
sin
g
Op
e
n
CV.
Th
e
p
ro
p
o
se
d
fra
m
e
wo
rk
i
n
teg
ra
tes
a
u
t
o
m
a
ted
in
teg
rit
y
c
h
e
c
k
s,
p
re
p
ro
c
e
ss
in
g
tec
h
n
i
q
u
e
s,
a
n
d
c
o
n
siste
n
c
y
v
e
rifi
c
a
ti
o
n
t
o
m
a
n
a
g
e
l
a
rg
e
-
sc
a
le
d
a
tas
e
ts
e
ffe
c
ti
v
e
ly
.
Ke
y
v
a
li
d
a
ti
o
n
p
ro
c
e
ss
e
s
in
c
l
u
d
e
ima
g
e
q
u
a
li
ty
a
ss
e
ss
m
e
n
t
(d
e
tec
ti
o
n
o
f
b
l
u
rri
n
e
ss
a
n
d
n
o
ise
),
v
e
rifi
c
a
ti
o
n
o
f
a
n
n
o
tati
o
n
a
c
c
u
ra
c
y
,
c
las
s
d
istri
b
u
ti
o
n
a
n
a
ly
sis,
a
n
d
i
d
e
n
ti
fica
ti
o
n
o
f
a
n
o
m
a
li
e
s.
Ad
d
it
i
o
n
a
ll
y
,
Op
e
n
CV
-
p
o
we
re
d
p
re
p
ro
c
e
ss
in
g
ste
p
s,
su
c
h
a
s
ima
g
e
re
siz
in
g
,
n
o
rm
a
li
z
a
ti
o
n
,
c
o
n
tras
t
o
p
ti
m
iza
ti
o
n
,
a
n
d
d
a
ta
a
u
g
m
e
n
tati
o
n
,
a
re
a
p
p
li
e
d
t
o
re
fin
e
d
a
tas
e
t
q
u
a
li
ty
fo
r
se
g
m
e
n
tatio
n
m
o
d
e
ls.
T
h
is
p
a
p
e
r
a
lso
a
d
d
re
ss
e
s
sc
a
lab
il
it
y
c
o
n
c
e
rn
s
a
ss
o
c
iate
d
wit
h
p
r
o
c
e
ss
in
g
e
x
ten
siv
e
d
a
tas
e
ts,
in
tro
d
u
c
i
n
g
o
p
ti
m
ize
d
b
a
tch
h
a
n
d
li
n
g
a
n
d
p
a
ra
ll
e
l
v
a
li
d
a
ti
o
n
tec
h
n
iq
u
e
s.
By
imp
lem
e
n
ti
n
g
a
str
u
c
tu
re
d
v
a
li
d
a
ti
o
n
w
o
rk
fl
o
w,
t
h
is
re
se
a
rc
h
e
n
h
a
n
c
e
s
th
e
re
li
a
b
il
it
y
,
r
o
b
u
stn
e
ss
,
a
n
d
o
v
e
ra
ll
e
ffe
c
ti
v
e
n
e
ss
o
f
se
m
a
n
ti
c
se
g
m
e
n
tatio
n
m
o
d
e
ls,
e
n
su
r
in
g
h
i
g
h
-
q
u
a
li
ty
train
in
g
d
a
ta f
o
r
d
e
e
p
lea
rn
in
g
a
p
p
li
c
a
ti
o
n
s.
K
ey
w
o
r
d
s
:
Data
s
et
Dee
p
L
ab
I
m
ag
in
g
s
em
a
n
tic
s
eg
m
en
tatio
n
Op
en
C
V
Un
m
an
n
ed
r
o
b
o
tic
v
eh
icles
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
:
I
v
an
Su
n
it R
o
u
t
Dep
ar
tm
en
t o
f
Me
ch
an
ical,
A
u
to
m
o
b
ile,
R
o
b
o
tics
an
d
Me
c
h
atr
o
n
ics
,
C
h
r
is
t U
n
iv
er
s
ity
B
en
g
alu
r
u
,
I
n
d
ia
E
m
ail: iv
an
.
r
o
u
t@
ch
r
is
tu
n
iv
er
s
ity
.
in
1.
I
NT
RO
D
UCT
I
O
N
Sem
an
tic
s
eg
m
en
tatio
n
is
a
c
r
u
cial
co
m
p
o
n
en
t
in
th
e
d
ev
e
lo
p
m
en
t
o
f
u
n
m
an
n
ed
r
o
b
o
tic
v
eh
icles,
en
ab
lin
g
p
r
ec
is
e
s
ce
n
e
u
n
d
e
r
s
tan
d
in
g
b
y
class
if
y
in
g
ea
ch
p
ix
el
o
f
an
im
ag
e
in
to
m
ea
n
i
n
g
f
u
l
ca
teg
o
r
ies
s
u
ch
as
r
o
ad
s
,
p
e
d
estrian
s
,
v
eh
icle
s
,
an
d
o
b
s
tacle
s
.
Ho
wev
er
,
th
e
ef
f
ec
tiv
en
ess
o
f
d
ee
p
lear
n
i
n
g
m
o
d
els
u
s
ed
f
o
r
s
eg
m
en
tatio
n
lar
g
ely
d
ep
e
n
d
s
o
n
th
e
q
u
ality
o
f
th
e
d
ataset
u
s
ed
f
o
r
tr
ain
in
g
an
d
v
alid
ati
o
n
.
Po
o
r
-
q
u
ality
d
ata,
in
co
r
r
ec
t
a
n
n
o
tatio
n
s
,
an
d
in
c
o
n
s
is
ten
cies
ca
n
lead
to
u
n
r
eliab
le
p
r
e
d
ictio
n
s
,
wh
ich
ar
e
p
a
r
ticu
lar
ly
cr
itical
in
au
to
n
o
m
o
u
s
d
r
iv
in
g
ap
p
licatio
n
s
wh
er
e
s
af
ety
is
p
ar
am
o
u
n
t.
T
h
is
r
esear
ch
f
o
cu
s
es
o
n
d
ata
v
alid
atio
n
f
o
r
s
em
an
tic
s
eg
m
en
tatio
n
u
s
in
g
Op
en
C
V,
aim
in
g
t
o
estab
lis
h
a
r
o
b
u
s
t
f
r
am
ew
o
r
k
f
o
r
en
s
u
r
i
n
g
d
ataset
ac
cu
r
ac
y
,
co
n
s
is
ten
cy
,
an
d
r
eliab
ilit
y
.
B
y
lev
er
a
g
in
g
Op
en
C
V’
s
im
ag
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
es,
aim
to
d
etec
t
e
r
r
o
r
s
[
1
]
,
v
er
if
y
an
n
o
tatio
n
s
,
an
d
o
p
tim
ize
d
ata
q
u
ality
f
o
r
tr
ain
in
g
d
ee
p
lear
n
in
g
m
o
d
els
in
a
u
to
n
o
m
o
u
s
v
eh
icle
s
y
s
tem
s
.
Un
m
an
n
ed
r
o
b
o
tic
v
eh
icles
r
ely
o
n
d
ee
p
lear
n
i
n
g
m
o
d
els
to
in
ter
p
r
et
th
ei
r
s
u
r
r
o
u
n
d
in
g
s
ac
c
u
r
ately
.
Sem
an
tic
s
eg
m
en
tatio
n
is
a
k
ey
tech
n
iq
u
e
u
s
ed
to
s
eg
m
en
t
an
im
ag
e
in
to
m
ea
n
in
g
f
u
l
r
e
g
io
n
s
,
b
u
t
er
r
o
r
s
in
d
ataset
q
u
ality
s
u
ch
as
m
is
lab
eled
p
ix
els,
class
im
b
alan
ce
,
an
d
im
ag
e
d
is
to
r
tio
n
s
ca
n
s
ev
er
ely
im
p
ac
t
m
o
d
el
p
er
f
o
r
m
an
ce
.
E
n
s
u
r
in
g
h
ig
h
-
q
u
ality
d
ata
th
r
o
u
g
h
a
s
tr
u
ct
u
r
ed
v
alid
atio
n
p
r
o
ce
s
s
is
ess
en
tial
to
im
p
r
o
v
e
s
eg
m
en
tatio
n
ac
cu
r
ac
y
an
d
r
el
iab
ilit
y
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
ch
20
2
6
:
7
1
-
79
72
T
h
e
p
r
o
b
lem
was
id
en
tifie
d
to
m
ak
e
a
d
etailed
an
aly
s
is
o
f
war
r
an
ty
o
f
p
ar
ts
to
d
eter
m
in
e
as
to
wh
eth
er
au
to
m
o
tiv
e
g
e
n
u
in
e
p
ar
ts
m
ee
t
th
eir
q
u
ality
r
eq
u
ir
em
en
ts
as
p
er
th
e
war
r
a
n
ty
p
o
licy
o
r
n
o
t;
wh
at
is
th
e
r
ate
an
d
f
r
eq
u
e
n
cy
o
f
war
r
an
ty
f
ailu
r
e/r
e
p
lace
m
en
t;
an
d
wh
eth
er
th
er
e
is
an
y
s
co
p
e
o
f
im
p
r
o
v
em
e
n
t
s
o
as
to
r
ed
u
ce
war
r
a
n
ty
f
ailu
r
e
ai
m
in
g
to
war
d
s
ze
r
o
war
r
an
ty
f
ailu
r
e
.
T
h
e
p
e
r
f
o
r
m
an
ce
an
d
d
ep
en
d
ab
ilit
y
o
f
s
em
an
tic
s
eg
m
en
tatio
n
m
o
d
el
s
in
u
n
m
an
n
ed
r
o
b
o
tic
v
eh
icl
es
ar
e
lar
g
ely
in
f
lu
en
ce
d
b
y
th
e
q
u
ality
o
f
th
e
tr
ain
in
g
d
ata
[
2
]
.
I
n
ad
e
q
u
ate
d
ataset
v
alid
atio
n
ca
n
lead
to
i
n
ac
cu
r
ate
s
ce
n
e
in
ter
p
r
etatio
n
,
in
cr
ea
s
in
g
t
h
e
r
is
k
o
f
s
af
ety
is
s
u
es.
T
h
er
ef
o
r
e,
a
n
ef
f
ec
tiv
e
d
ata
v
ali
d
atio
n
f
r
am
ewo
r
k
u
tili
zin
g
Op
en
C
V
is
es
s
en
tial
to
m
ain
tain
d
ataset
in
teg
r
ity
an
d
e
n
h
an
ce
m
o
d
el
r
o
b
u
s
tn
ess
.
L
o
n
g
et
a
l.
[
3
]
i
n
tr
o
d
u
ce
d
a
m
ajo
r
s
h
if
t
b
y
r
ep
lacin
g
t
r
ad
itio
n
al
f
u
lly
c
o
n
n
ec
te
d
la
y
er
s
with
I
n
f
o
r
m
atio
n
Sem
a
n
tic
s
eg
m
en
tatio
n
h
as
u
n
d
er
g
o
n
e
s
ig
n
if
ic
an
t
ad
v
a
n
ce
m
en
ts
,
p
ar
ticu
lar
l
y
in
d
ee
p
lear
n
in
g
-
b
ased
ap
p
r
o
ac
h
es.
W
h
ile
v
ar
io
u
s
m
o
d
els
h
av
e
b
ee
n
p
r
o
p
o
s
ed
to
en
h
an
ce
s
eg
m
en
tatio
n
a
cc
u
r
ac
y
,
ea
ch
co
m
es
with
its
o
wn
ad
v
a
n
tag
es
a
n
d
lim
itatio
n
s
.
T
h
is
s
ec
tio
n
cr
itically
ex
am
in
es
k
ey
w
o
r
k
s
in
t
h
e
f
ield
.
co
n
v
o
l
u
tio
n
al
o
n
es,
allo
win
g
f
o
r
p
i
x
el
-
wis
e
p
r
ed
ictio
n
s
.
Desp
ite
its
ef
f
ec
tiv
en
ess
in
d
en
s
e
class
if
icatio
n
,
FC
N
o
f
ten
s
tr
u
g
g
les
with
ca
p
tu
r
in
g
f
in
e
-
g
r
ai
n
ed
d
etails,
lead
in
g
t
o
s
u
b
o
p
tim
al
b
o
u
n
d
ar
y
d
e
f
in
itio
n
s
in
s
eg
m
en
tatio
n
task
s
.
T
h
e
m
o
d
e
l's
r
elian
ce
o
n
d
o
wn
s
am
p
lin
g
also
r
esu
lts
in
a
lo
s
s
o
f
s
p
atial
r
eso
lu
tio
n
,
wh
ic
h
ca
n
im
p
ac
t
th
e
ac
c
u
r
ac
y
o
f
s
e
g
m
en
tatio
n
in
co
m
p
le
x
s
ce
n
es
[
4
]
.
B
ad
r
i
n
ar
ay
an
a
n
et
a
l.
[
5
]
in
tr
o
d
u
ce
d
Seg
Net,
an
en
co
d
er
-
d
ec
o
d
er
ar
ch
itectu
r
e
aim
ed
at
im
p
r
o
v
in
g
c
o
m
p
u
tatio
n
al
ef
f
icien
cy
.
I
ts
s
tr
en
g
th
lies
in
its
ab
ilit
y
to
r
etain
s
p
atial
d
etails
u
s
in
g
an
ef
f
icien
t
u
p
s
am
p
lin
g
m
eth
o
d
.
Ho
wev
er
,
Seg
Net
lack
s
th
e
e
x
ten
s
iv
e
m
u
lti
-
s
ca
le
f
ea
tu
r
e
ex
tr
ac
tio
n
ca
p
a
b
ilit
ies
s
ee
n
in
m
o
r
e
r
ec
en
t
a
r
ch
itectu
r
es.
T
h
e
ab
s
en
ce
o
f
m
ec
h
an
is
m
s
s
u
ch
as
d
ilated
co
n
v
o
l
u
tio
n
s
o
r
atten
tio
n
-
b
ase
d
r
ef
in
em
e
n
t
lim
its
its
ab
ilit
y
to
p
er
f
o
r
m
well
in
h
ig
h
ly
d
et
ailed
s
eg
m
en
tatio
n
s
ce
n
ar
io
s
.
T
h
e
U
-
Net
ar
ch
itectu
r
e,
d
ev
elo
p
ed
b
y
R
o
n
n
eb
er
g
er
et
a
l.
[
6
]
,
s
ig
n
if
ican
tly
im
p
r
o
v
ed
s
eg
m
en
tatio
n
ac
cu
r
ac
y
b
y
in
co
r
p
o
r
atin
g
s
k
ip
co
n
n
ec
tio
n
s
,
w
h
ich
p
r
eser
v
e
f
in
e
-
s
ca
le
d
etails.
W
h
ile
U
-
Net
h
as
p
r
o
v
en
h
ig
h
ly
ef
f
ec
tiv
e
i
n
m
ed
ical
im
ag
in
g
an
d
o
th
er
s
tr
u
ctu
r
e
d
en
v
ir
o
n
m
e
n
ts
,
its
r
elian
ce
o
n
s
y
m
m
etr
ical
en
co
d
in
g
an
d
d
ec
o
d
in
g
m
ay
n
o
t
b
e
o
p
tim
al
f
o
r
lar
g
e
-
s
ca
le
au
to
n
o
m
o
u
s
v
eh
icle
d
atasets
th
at
r
eq
u
ir
e
m
o
r
e
co
m
p
lex
co
n
tex
tu
al
u
n
d
e
r
s
tan
d
in
g
.
Ad
d
itio
n
ally
,
t
h
e
co
m
p
u
tatio
n
al
co
s
t
o
f
tr
ain
in
g
U
-
Net
o
n
h
ig
h
-
r
eso
lu
tio
n
im
ag
es
r
em
ain
s
a
ch
allen
g
e
.
C
h
en
et
al
[
7
]
a
d
d
r
ess
ed
s
o
m
e
o
f
th
ese
lim
itatio
n
s
with
Dee
p
L
ab
,
wh
ich
in
tr
o
d
u
ce
d
atr
o
u
s
co
n
v
o
lu
tio
n
s
to
ex
p
a
n
d
th
e
r
ec
e
p
tiv
e
f
ield
with
o
u
t
in
cr
ea
s
in
g
co
m
p
u
tatio
n
al
co
s
t.
T
h
is
in
n
o
v
atio
n
en
h
an
ce
d
m
u
lti
-
s
ca
le
f
ea
tu
r
e
ex
tr
ac
tio
n
,
m
ak
in
g
Dee
p
L
ab
p
ar
ticu
lar
ly
e
f
f
ec
tiv
e
in
s
eg
m
en
tin
g
o
b
jects
at
v
ar
y
in
g
s
ca
les.
Fu
r
th
er
m
o
r
e,
th
e
in
teg
r
atio
n
o
f
f
u
lly
c
o
n
n
ec
ted
C
R
Fs
im
p
r
o
v
ed
b
o
u
n
d
a
r
y
p
r
ec
is
io
n
.
Ho
wev
er
,
Dee
p
L
ab
’
s
r
elian
c
e
o
n
co
m
p
lex
ar
c
h
itectu
r
al
m
o
d
if
icatio
n
s
in
cr
ea
s
es
co
m
p
u
tatio
n
al
o
v
er
h
ea
d
,
m
ak
in
g
it less
s
u
itab
le
f
o
r
r
ea
l
-
tim
e
ap
p
licatio
n
s
in
u
n
m
an
n
e
d
r
o
b
o
tic
v
eh
icles with
o
u
t f
u
r
t
h
er
o
p
tim
izatio
n
.
I
n
th
is
p
ap
er
,
th
e
au
th
o
r
s
h
a
v
e
wo
r
k
ed
o
n
th
e
s
em
a
n
tic
s
eg
m
en
tatio
n
o
f
th
e
d
ata
f
o
r
u
n
m
an
n
ed
r
o
b
o
tic
v
eh
icles
g
iv
in
g
clar
i
ty
o
n
th
e
ar
c
h
itectu
r
e
m
o
d
e
l
o
f
th
e
im
ag
e
u
s
in
g
Op
en
C
V
an
d
Dee
p
L
a
b
tech
n
iq
u
es.
Fu
r
th
e
r
,
s
ec
tio
n
2
talk
s
ab
o
u
t
th
e
d
etails
o
f
t
h
e
t
o
o
ls
ad
o
p
ted
in
id
en
tif
y
in
g
th
e
r
eg
io
n
s
,
s
ec
tio
n
3
d
is
cu
s
s
ed
ab
o
u
t
th
e
s
em
an
tic
s
eg
m
en
tatio
n
o
f
a
s
ce
n
e
clea
r
ly
h
ig
h
lig
h
ti
n
g
th
e
b
o
u
n
d
a
r
y
an
d
th
e
d
if
f
er
en
t
r
eg
io
n
s
to
id
e
n
tify
th
e
o
b
ject
s
an
d
f
in
ally
s
ec
tio
n
4
co
n
cl
u
d
es
with
th
e
v
er
if
icatio
n
p
r
o
ce
s
s
an
d
an
n
o
tate
d
lab
els o
f
th
e
s
ce
n
e.
2.
M
E
T
H
O
D
2
.
1
.
M
et
ho
do
lo
g
y
T
o
en
s
u
r
e
t
h
e
q
u
ality
,
co
n
s
is
t
en
cy
,
an
d
r
eliab
ilit
y
o
f
th
e
d
a
taset
p
r
io
r
to
m
o
d
el
tr
ai
n
in
g
,
s
y
s
tem
atic
d
ata
p
r
ep
a
r
atio
n
an
d
v
alid
atio
n
p
ip
elin
e
is
em
p
lo
y
ed
.
T
h
e
o
v
er
all
p
r
o
ce
s
s
co
n
s
is
ts
o
f
th
e
f
o
llo
win
g
s
tag
es:
−
Data
s
et
C
o
llect
io
n
:
Gath
er
r
a
w
im
ag
e
d
ata
an
d
an
n
o
tatio
n
s
.
−
Op
en
C
V
p
r
ep
r
o
ce
s
s
in
g
: c
o
n
tr
ast en
h
an
ce
m
en
t,
n
o
is
e
r
ed
u
cti
o
n
an
d
n
o
r
m
aliza
tio
n
.
−
An
n
o
tatio
n
C
o
n
s
is
ten
cy
C
h
ec
k
:
Dete
ct
an
d
co
r
r
ec
t m
is
lab
eled
r
eg
io
n
s
.
−
C
las
s
I
m
b
alan
ce
An
aly
s
is
:
I
d
e
n
tify
u
n
d
er
r
e
p
r
esen
ted
class
es
[
8
]
.
−
Data
Au
g
m
en
tatio
n
: A
p
p
ly
tr
a
n
s
f
o
r
m
atio
n
s
to
b
alan
ce
class
d
is
tr
ib
u
tio
n
.
−
Au
to
m
ated
Qu
ality
Ass
ess
m
e
n
t
:
E
v
alu
ate
d
ataset
in
teg
r
ity
u
s
in
g
p
r
ed
ef
i
n
ed
m
etr
ics.
−
Ma
n
u
al
R
ev
iew
(
if
n
ee
d
e
d
)
:
A
d
d
r
ess
u
n
r
eso
lv
e
d
in
co
n
s
is
ten
cies.
−
Fin
al
Data
s
et
Valid
atio
n
:
Ap
p
r
o
v
e
r
e
f
in
ed
d
ataset
f
o
r
m
o
d
el
tr
ain
in
g
s
h
o
wn
in
Fig
u
r
e
1
.
Un
m
an
n
ed
r
o
b
o
tic
v
e
h
icles
(
AVs)
an
d
ad
v
an
ce
d
d
r
iv
er
a
s
s
is
tan
ce
s
y
s
tem
s
(
A
DAS)
d
ep
en
d
o
n
s
em
an
tic
s
eg
m
en
tatio
n
f
o
r
ac
cu
r
ate
s
ce
n
e
in
ter
p
r
etatio
n
[
9
]
.
E
r
r
o
r
s
in
th
ese
m
ask
s
ca
n
r
esu
lt
in
m
is
al
ig
n
ed
lan
e
d
etec
tio
n
,
m
is
s
ed
o
b
jects,
o
r
in
co
r
r
ec
t
p
ath
p
lan
n
in
g
,
p
o
s
in
g
s
ig
n
if
ican
t
s
af
ety
co
n
ce
r
n
s
.
Au
to
m
atin
g
s
em
an
tic
s
eg
m
en
tatio
n
an
aly
s
is
im
p
r
o
v
es
AI
m
o
d
el
ac
cu
r
ac
y
,
en
h
an
ce
s
r
ea
l
-
wo
r
ld
p
er
f
o
r
m
an
ce
,
an
d
e
n
s
u
r
es
s
af
er
o
p
er
atio
n
.
T
h
is
p
r
o
ce
s
s
is
cr
u
cial
f
o
r
ass
es
s
in
g
m
o
d
el
p
er
f
o
r
m
an
ce
d
u
r
in
g
test
d
r
iv
e
s
,
wh
er
e
en
g
in
ee
r
s
ev
alu
ate
m
etr
ics
s
u
ch
as
I
n
te
r
s
ec
tio
n
o
v
er
Un
i
o
n
(
I
o
U)
,
m
is
alig
n
m
en
t
d
etec
tio
n
,
a
n
d
c
o
n
to
u
r
co
m
p
leten
ess
[
1
0
]
.
I
t
also
h
elp
s
id
en
tify
f
ailu
r
e
ca
s
es
in
ch
allen
g
in
g
r
ea
l
-
wo
r
ld
co
n
d
itio
n
s
,
in
clu
d
i
n
g
r
ain
,
f
o
g
,
a
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
S
ema
n
tic
s
eg
men
ta
tio
n
fo
r
d
a
ta
va
lid
a
tio
n
in
u
n
ma
n
n
e
d
r
o
b
o
tic
ve
h
icles
(
I
va
n
S
u
n
it R
o
u
t
)
73
n
ig
h
ttime
d
r
iv
in
g
.
Sem
an
tic
s
eg
m
en
tatio
n
an
aly
s
is
p
lay
s
a
cr
u
cial
r
o
le
in
en
h
an
cin
g
th
e
p
er
ce
p
tio
n
ca
p
ab
ilit
ies
o
f
u
n
m
an
n
e
d
r
o
b
o
tic
v
eh
icles
(
AVs).
T
h
ese
m
ask
s
h
elp
AV
s
y
s
tem
s
in
ter
p
r
et
t
h
e
en
v
ir
o
n
m
e
n
t
b
y
ac
cu
r
ately
id
en
tify
in
g
lan
es,
o
b
s
tacle
s
,
an
d
o
th
er
cr
itical
r
o
ad
elem
en
ts
.
Ho
wev
er
,
er
r
o
r
s
i
n
s
eg
m
en
tatio
n
ca
n
lead
to
m
is
alig
n
ed
lan
e
d
etec
tio
n
,
u
n
d
etec
ted
o
b
jects,
o
r
f
au
l
ty
p
ath
p
lan
n
in
g
,
wh
ich
m
ay
c
o
m
p
r
o
m
is
e
s
af
ety
.
Au
to
m
atin
g
m
ask
an
al
y
s
is
al
lo
ws
en
g
in
ee
r
s
to
ass
ess
AI
m
o
d
el
p
er
f
o
r
m
a
n
ce
ef
f
ec
tiv
el
y
,
en
s
u
r
in
g
h
ig
h
e
r
ac
cu
r
ac
y
an
d
r
eliab
ilit
y
in
r
ea
l
-
wo
r
ld
s
ce
n
ar
io
s
[
1
1
]
.
Key
ev
alu
atio
n
m
etr
ics
in
clu
d
e
I
o
U
,
c
o
n
to
u
r
co
m
p
l
eten
ess
,
an
d
m
is
alig
n
m
en
t
d
etec
tio
n
.
Ad
d
itio
n
ally
,
t
h
is
an
aly
s
is
h
el
p
s
id
en
tify
p
er
f
o
r
m
a
n
ce
is
s
u
es
in
ad
v
e
r
s
e
co
n
d
itio
n
s
s
u
ch
a
s
r
ain
,
f
o
g
,
o
r
lo
w
-
lig
h
t
en
v
ir
o
n
m
en
ts
,
u
ltima
tely
im
p
r
o
v
in
g
th
e
r
o
b
u
s
tn
ess
o
f
AV
p
er
ce
p
tio
n
s
y
s
tem
s
.
Data
C
o
llectio
n
:
Gath
er
r
ea
l
-
wo
r
ld
d
r
iv
in
g
d
ata,
in
clu
d
in
g
d
iv
e
r
s
e
en
v
ir
o
n
m
e
n
ts
s
u
ch
as
h
ig
h
way
s
,
u
r
b
a
n
s
tr
ee
ts
,
an
d
ad
v
e
r
s
e
wea
th
er
co
n
d
itio
n
s
.
Pre
p
r
o
ce
s
s
in
g
:
N
o
r
m
alize
an
d
en
h
an
ce
th
e
c
o
llected
d
ata
to
im
p
r
o
v
e
s
eg
m
en
tatio
n
ac
c
u
r
ac
y
,
en
s
u
r
in
g
co
n
s
is
ten
cy
ac
r
o
s
s
v
ar
io
u
s
s
ce
n
ar
i
o
s
.
Mo
d
el
I
n
f
er
en
ce
:
Ap
p
l
y
th
e
tr
ain
ed
AI
m
o
d
el
to
g
en
er
at
e
s
em
an
tic
s
eg
m
en
tatio
n
f
o
r
o
b
j
ec
ts
,
lan
es,
an
d
o
th
er
r
o
ad
ele
m
en
ts
.
Per
f
o
r
m
an
ce
E
v
alu
atio
n
: A
s
s
es
s
th
e
m
o
d
el
u
s
in
g
m
etr
ics
lik
e
I
o
U
,
b
o
u
n
d
ar
y
ac
cu
r
ac
y
,
an
d
m
is
alig
n
m
e
n
t
d
etec
tio
n
[
1
2
]
.
E
r
r
o
r
Dete
ct
io
n
:
I
d
en
tif
y
f
ailu
r
e
ca
s
es,
s
u
ch
as
in
co
r
r
ec
t
o
b
ject
class
if
icatio
n
s
,
in
co
m
p
lete
c
o
n
to
u
r
s
,
o
r
m
is
s
in
g
lan
e
m
ar
k
in
g
s
.
C
o
m
p
a
r
is
o
n
with
s
em
an
tic
s
eg
m
en
tatio
n
:
C
o
m
p
ar
e
g
e
n
er
ated
m
ask
s
with
m
an
u
ally
la
b
eled
s
em
an
tic
s
eg
m
en
tatio
n
d
ata
t
o
m
ea
s
u
r
e
ac
cu
r
ac
y
a
n
d
d
ete
ct
in
co
n
s
is
ten
cies.
I
ter
ativ
e
I
m
p
r
o
v
em
en
t:
R
ef
in
e
th
e
m
o
d
el
b
y
ad
j
u
s
tin
g
p
ar
am
eter
s
,
r
etr
ai
n
in
g
with
a
d
d
itio
n
al
d
ata,
an
d
ad
d
r
ess
in
g
d
etec
ted
e
r
r
o
r
s
.
R
ea
l
-
W
o
r
ld
T
esti
n
g
:
Valid
ate
p
er
f
o
r
m
an
ce
i
n
r
ea
l
-
w
o
r
ld
c
o
n
d
itio
n
s
,
i
n
clu
d
in
g
ch
allen
g
in
g
s
ce
n
ar
io
s
lik
e
f
o
g
,
r
ain
,
an
d
n
ig
h
t
d
r
iv
i
n
g
.
Dep
lo
y
m
en
t
an
d
Mo
n
ito
r
in
g
:
I
n
teg
r
ate
th
e
im
p
r
o
v
e
d
m
o
d
el
in
to
AV
s
y
s
tem
s
an
d
co
n
tin
u
o
u
s
ly
m
o
n
ito
r
its
r
ea
l
-
tim
e
p
er
f
o
r
m
an
ce
f
o
r
f
u
r
t
h
er
en
h
a
n
ce
m
en
ts
[
1
3
]
.
Fig
u
r
e
1
.
Flo
wch
ar
t
d
ep
ictin
g
th
e
s
tep
s
in
v
o
lv
ed
i
n
m
eth
o
d
o
l
o
g
y
2.
2
.
T
ec
hn
iqu
es
Me
th
o
d
u
s
ed
:
co
lo
r
m
a
p
p
in
g
&
alp
h
a
b
len
d
in
g
.
Fo
r
v
is
u
a
l
in
s
p
ec
tio
n
o
n
th
e
o
r
ig
in
al
i
m
ag
e,
th
e
s
em
an
tic
s
eg
m
en
tatio
n
o
v
er
la
y
s
th
e
p
r
ed
icted
im
ag
e.
c
v
2
.
ap
p
ly
C
o
lo
r
Ma
p
(
m
ask
,
c
v
2
.
C
OL
OR
MA
P_
J
E
T
)
:
C
o
n
v
er
ts
th
e
g
r
ay
s
ca
le
m
ask
in
to
a
co
l
o
r
r
e
p
r
esen
tatio
n
u
s
in
g
th
e
J
et
co
lo
r
m
a
p
.
T
h
is
e
n
h
an
ce
s
v
is
ib
ilit
y
b
y
m
ap
p
in
g
d
if
f
er
e
n
t
p
ix
el
i
n
ten
s
ities
to
co
lo
r
s
[
1
4
]
.
c
v
2
.
ad
d
W
eig
h
ted
(
im
ag
e,
0
.
7
,
0
,
co
l
o
r
_
m
ask
,
0
.
3
)
:
Usi
n
g
weig
h
ted
ad
d
itio
n
t
h
e
co
l
o
u
r
ed
m
ask
b
len
d
s
th
e
im
a
g
e
o
r
ig
in
ality
.
T
h
e
o
r
i
g
in
al
i
m
ag
e
is
g
iv
e
n
7
0
%
im
p
o
r
tan
ce
,
an
d
th
e
m
ask
3
0
%,
cr
ea
tin
g
a
s
em
i
-
tr
a
n
s
p
ar
en
t
o
v
e
r
lay
e
f
f
ec
t.
Pu
r
p
o
s
e:
T
h
is
h
elp
s
in
v
is
u
alizin
g
th
e
s
eg
m
en
tatio
n
r
esu
lts
b
y
s
u
p
er
im
p
o
s
in
g
th
em
o
n
th
e
o
r
ig
i
n
al
im
ag
e
s
h
o
wn
i
n
Fig
u
r
e
2
[
1
5
]
.
Me
th
o
d
Used
: Co
n
to
u
r
Dete
ctio
n
cv
2
.
f
in
d
C
o
n
to
u
r
s
(
m
ask
,
cv
2
.
R
E
T
R
_
T
R
E
E
,
cv
2
.
C
HAI
N_
APPR
OX_
SIM
PLE
)
:
cv
2
.
R
E
T
R
_
T
R
E
E
: Retr
iev
es a
ll c
o
n
to
u
r
s
an
d
r
ec
o
n
s
tr
u
cts th
e
h
ier
ar
ch
y
.
cv
2
.
C
HAI
N_
APPR
O
X_
SIM
P
L
E
: Co
m
p
r
ess
es c
o
n
to
u
r
p
o
in
t
s
to
s
av
e
m
em
o
r
y
.
cv
2
.
d
r
awCo
n
to
u
r
s
(
im
ag
e,
co
n
to
u
r
s
,
-
1
,
(
0
,
2
5
5
,
0
)
,
2
)
:
Dr
aws a
ll c
o
n
to
u
r
s
o
n
t
h
e
o
r
i
g
in
al
im
ag
e
u
s
in
g
g
r
ee
n
c
o
lo
r
(
(
0
,
2
5
5
,
0
)
)
with
a
th
ick
n
ess
o
f
2
p
ix
el
s
[
1
6
]
.
Pu
r
p
o
s
e:
I
d
en
tifie
s
th
e
b
o
u
n
d
a
r
ies o
f
s
eg
m
en
ted
r
eg
io
n
s
,
wh
i
ch
ca
n
b
e
u
s
ef
u
l f
o
r
f
u
r
th
er
o
b
ject
an
aly
s
is
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
ch
20
2
6
:
7
1
-
79
74
Me
th
o
d
Used
: H
is
to
g
r
am
C
alcu
latio
n
cv
2
.
ca
lcHist
(
[
m
ask
]
,
[
0
]
,
No
n
e,
[
2
5
6
]
,
[
0
,
2
5
6
]
)
:
C
o
m
p
u
tes th
e
h
is
to
g
r
am
o
f
th
e
g
r
ay
s
ca
le
m
ask
.
[
0
]
: Sp
ec
if
ies th
e
g
r
a
y
s
ca
le
ch
an
n
el
[
1
7
]
.
No
n
e:
No
m
ask
is
ap
p
lied
; e
n
t
ir
e
im
ag
e
is
an
aly
ze
d
.
[
2
5
6
]
:
2
5
6
b
in
s
(
o
n
e
f
o
r
ea
ch
i
n
ten
s
ity
lev
el)
.
[
0
,
2
5
6
]
:
T
h
e
v
alu
es o
f
in
ten
s
ity
v
ar
ies f
r
o
m
2
5
5
(
wh
ite)
t
o
0
(
b
lack
)
.
p
lt.p
lo
t(
h
is
t)
: Plo
ts
th
e
h
is
to
g
r
am
.
Pu
r
p
o
s
e:
Help
s
id
en
tify
class
im
b
alan
ce
in
s
eg
m
e
n
tatio
n
b
y
ch
ec
k
in
g
h
o
w
m
a
n
y
p
i
x
els b
elo
n
g
to
t
h
e
f
o
r
eg
r
o
u
n
d
an
d
b
ac
k
g
r
o
u
n
d
.
Me
th
o
d
Used
:
Mo
r
p
h
o
lo
g
ical
Op
er
atio
n
s
to
C
lean
No
is
y
Ma
s
k
Ker
n
el
= F
o
r
m
o
r
p
h
o
lo
g
ical
tr
an
s
f
o
r
m
atio
n
s
is
u
s
ed
b
y
3
×3
k
er
n
el
an
d
is
g
iv
en
b
y
n
p
.
o
n
es(n
p
.
u
i
n
t8
,
(
3
,
3
)
)
.
cv
2
.
m
o
r
p
h
o
lo
g
y
E
x
(
m
ask
,
cv
2
.
MO
R
PH_
OPE
N,
k
er
n
el)
: o
MO
R
PH_
O
PEN
=
E
r
o
s
io
n
+
Dilatio
n
:
▪
E
r
o
s
io
n
r
em
o
v
es sm
all
wh
ite
n
o
is
e.
Dilatio
n
r
esto
r
es th
e
o
v
er
all
s
h
ap
e.
Pu
r
p
o
s
e:
C
lean
s
u
p
s
m
all
n
o
is
e
o
r
is
o
lated
p
ix
els in
th
e
s
em
an
tic
s
eg
m
en
tatio
n
,
im
p
r
o
v
i
n
g
ac
cu
r
ac
y
.
Fig
u
r
e
2
.
Sem
an
tic
s
eg
r
e
g
atio
n
o
f
m
ask
o
v
er
la
y
2.
3
.
L
ibra
ries
L
ab
elM
e:
An
o
p
en
-
s
o
u
r
ce
to
o
l
d
esig
n
ed
f
o
r
m
an
u
ally
an
n
o
t
atin
g
an
d
v
e
r
if
y
in
g
im
ag
e
s
eg
m
en
tatio
n
d
atasets
.
I
t
allo
ws
u
s
er
s
to
d
r
aw
p
r
ec
is
e
p
o
l
y
g
o
n
al
an
n
o
tat
io
n
s
ar
o
u
n
d
o
b
jects
in
an
i
m
ag
e,
cr
ea
tin
g
h
i
g
h
-
q
u
ality
lab
eled
d
atasets
f
o
r
t
r
ain
in
g
A
I
m
o
d
els
[
1
8
]
.
L
ab
elM
e
s
u
p
p
o
r
ts
v
ar
io
u
s
an
n
o
tatio
n
f
o
r
m
ats
an
d
en
ab
les u
s
er
s
to
r
ev
iew
an
d
r
e
f
in
e
s
em
an
tic
s
eg
m
en
tatio
n
t
o
en
s
u
r
e
ac
cu
r
ac
y
.
C
o
m
p
u
ter
v
is
io
n
a
n
n
o
tatio
n
t
o
o
l
(
C
VAT
)
:
A
p
o
wer
f
u
l
an
n
o
tatio
n
p
latf
o
r
m
th
at
i
n
clu
d
es
b
u
ilt
-
in
v
alid
atio
n
f
ea
tu
r
es.
I
t
p
r
o
v
id
es
to
o
ls
f
o
r
an
n
o
tatin
g
o
b
jects,
v
er
i
f
y
in
g
s
eg
m
en
tatio
n
ac
cu
r
ac
y
,
a
n
d
e
n
s
u
r
in
g
c
o
n
s
is
ten
cy
in
lab
eled
d
atasets
.
C
VAT
s
u
p
p
o
r
ts
m
u
ltip
le
a
n
n
o
tatio
n
f
o
r
m
ats,
c
o
llab
o
r
atio
n
am
o
n
g
an
n
o
tato
r
s
,
a
n
d
in
teg
r
atio
n
with
d
ee
p
lear
n
i
n
g
wo
r
k
f
lo
ws f
o
r
en
h
a
n
ce
d
d
ataset
q
u
ality
co
n
tr
o
l.
Py
th
o
n
L
i
b
r
ar
ies
(
Nu
m
Py
,
Op
en
C
V,
Py
T
o
r
ch
)
:
T
h
ese
l
ib
r
ar
ies
ar
e
wid
ely
u
s
ed
to
au
to
m
ate
s
em
a
n
tic
s
eg
m
en
tatio
n
v
alid
atio
n
[
1
9
]
.
Nu
m
Py
en
ab
les
ef
f
icien
t
n
u
m
er
ical
o
p
er
atio
n
s
,
s
u
ch
as
ca
lcu
latin
g
p
ix
el
-
wis
e
d
if
f
er
en
ce
s
b
etwe
en
s
em
an
tic
s
eg
m
en
tatio
n
an
d
p
r
e
d
icted
s
eg
m
en
tatio
n
o
u
tp
u
ts
[
2
0
]
.
Op
en
C
V
p
r
o
v
i
d
es
im
ag
e
p
r
o
ce
s
s
in
g
f
u
n
ctio
n
s
f
o
r
ev
alu
at
in
g
co
n
to
u
r
alig
n
m
en
t,
d
etec
tin
g
m
is
class
if
ied
r
eg
io
n
s
,
an
d
p
er
f
o
r
m
i
n
g
m
o
r
p
h
o
lo
g
ical
o
p
er
atio
n
s
.
Py
T
o
r
ch
f
ac
ilit
ates
d
ee
p
lear
n
in
g
-
b
ased
v
alid
atio
n
,
allo
wi
n
g
m
o
d
els
to
ass
ess
s
eg
m
en
tatio
n
q
u
ality
u
s
in
g
p
er
f
o
r
m
an
ce
m
etr
ics lik
e
I
o
U
an
d
Dice
co
ef
f
icien
t
[
2
1
]
.
B
y
co
m
b
in
in
g
th
ese
to
o
ls
an
d
lib
r
ar
ies
s
h
o
wn
in
Fig
u
r
e
3
,
en
g
in
ee
r
s
ca
n
s
tr
ea
m
lin
e
th
e
an
n
o
tatio
n
,
v
er
if
icatio
n
,
an
d
v
alid
atio
n
p
r
o
ce
s
s
,
im
p
r
o
v
i
n
g
th
e
ac
cu
r
ac
y
o
f
im
ag
e
s
eg
m
e
n
tatio
n
d
atasets
u
s
ed
in
au
to
n
o
m
o
u
s
v
eh
icle
p
er
ce
p
tio
n
s
y
s
tem
s
[
2
2
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
S
ema
n
tic
s
eg
men
ta
tio
n
fo
r
d
a
ta
va
lid
a
tio
n
in
u
n
ma
n
n
e
d
r
o
b
o
tic
ve
h
icles
(
I
va
n
S
u
n
it R
o
u
t
)
75
Fig
u
r
e
3
.
Fo
ld
e
r
r
ep
r
esen
tatio
n
an
d
lib
r
ar
y
im
p
o
r
t
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
s
em
an
tic
s
eg
m
en
tatio
n
o
f
an
im
a
g
e
s
ee
n
in
Fig
u
r
e
4
h
as
b
ee
n
g
iv
en
to
th
e
ca
m
er
a
o
f
a
n
au
to
n
o
m
o
u
s
v
e
h
icle
wh
ich
w
ill
h
elp
to
m
ar
k
th
e
o
b
jects
with
its
d
is
tin
ct
o
u
tlin
e
f
ea
tu
r
es
o
f
c
o
lo
r
in
g
f
o
r
d
is
tin
g
u
is
h
in
g
m
o
v
in
g
o
b
jects,
s
tag
n
an
t
b
o
d
ies,
h
u
m
an
s
,
v
e
h
icles
an
d
th
e
wate
r
b
o
d
ies
[
2
3
]
.
T
h
e
ab
ilit
y
to
im
p
r
o
v
e
p
i
x
el
lev
el
p
r
ec
is
io
n
is
d
o
n
e.
T
h
e
o
b
ject
d
etec
tio
n
an
d
th
e
b
eh
av
io
r
o
f
th
e
o
b
je
cts
in
th
e
im
ag
e
is
tr
ac
k
ed
b
y
tr
ain
in
g
t
h
e
m
o
d
el
wh
ich
m
ak
es
th
e
o
b
ject
is
o
la
ted
.
T
h
e
p
r
e
p
r
o
ce
s
s
in
g
o
f
th
e
d
ata
is
s
tr
ea
m
lin
ed
u
s
in
g
th
ese
m
ask
s
an
d
r
e
d
u
ce
co
s
ts
o
f
co
m
p
u
tatio
n
b
y
ir
r
elev
an
t
p
ix
els
[
2
4
]
.
T
h
e
m
o
r
e
c
o
m
p
lex
m
ask
s
u
s
es
in
teg
er
s
s
ay
1
f
o
r
r
o
a
d
s
,
2
f
o
r
ca
r
s
,
3
f
o
r
lam
p
p
o
s
ts
an
d
4
f
o
r
wate
r
b
o
d
ies.
Fig
u
r
e
4
.
Sem
an
tic
s
eg
m
e
n
t
atio
n
o
f
a
s
ce
n
e
T
h
e
s
ig
n
if
ican
t
f
ea
t
u
r
e
o
f
u
s
in
g
o
v
e
r
lay
ed
s
eg
m
e
n
tatio
n
s
h
o
wn
in
Fig
u
r
e
5
ca
n
r
ed
u
ce
th
e
o
v
er
lap
o
f
th
e
b
o
u
n
d
a
r
ies
an
d
o
b
jects
wh
ich
m
ak
es
it
d
if
f
ic
u
lt
to
id
en
tif
y
an
d
d
etec
t
f
r
o
m
o
th
er
o
b
ject
s
[
2
5
]
.
Ma
k
in
g
u
s
e
o
f
th
is
f
ea
tu
r
e
ca
n
s
ig
n
if
ican
tl
y
in
cr
ea
s
e
th
e
p
ix
el
z
o
n
e
an
d
t
h
er
eb
y
r
ed
u
cin
g
th
e
b
lac
k
an
d
g
r
ay
s
ca
le
s
p
o
ts
in
th
e
im
ag
e
to
ea
s
ily
co
n
tr
ast
a
n
d
s
h
o
w
th
e
o
b
jects
to
id
e
n
tif
y
th
e
im
ag
es
i
n
th
e
s
ce
n
e
an
d
m
ak
e
k
n
o
wn
o
f
th
e
o
b
jects
f
o
r
b
etter
d
is
tin
ctio
n
a
n
d
clar
ity
g
iv
in
g
h
ig
h
v
alu
e
p
r
ec
is
io
n
an
d
tr
an
s
p
ar
en
c
y
to
th
e
im
ag
e
d
etec
tio
n
[
2
6
]
.
T
h
e
m
ask
p
ix
el
d
is
tr
ib
u
tio
n
s
ee
n
in
Fig
u
r
e
6
ca
n
in
f
o
r
m
th
e
u
s
er
o
f
th
e
clar
ity
o
f
t
h
e
o
b
ject
d
is
tin
ctio
n
in
ca
p
tu
r
in
g
th
e
co
lo
r
in
g
o
f
th
e
im
a
g
e
a
n
d
m
o
r
e
p
r
o
m
in
e
n
t
d
is
tr
ib
u
tio
n
o
f
t
h
e
zo
n
es
in
g
ettin
g
th
e
ar
ea
s
m
ar
k
ed
f
o
r
id
e
n
tify
in
g
th
e
o
b
jects
an
d
d
is
tin
ctiv
ely
u
tili
zin
g
th
e
b
o
u
n
d
ar
ies
f
o
r
p
r
o
m
in
en
t
m
ar
k
in
g
ar
ea
s
.
T
h
e
clea
n
ed
m
ask
f
ea
tu
r
e
s
ee
n
in
Fig
u
r
e
7
will
r
ef
in
e
th
e
ed
g
es
an
d
b
o
u
n
d
ar
ies
to
m
o
r
e
clar
ity
o
f
th
e
p
ix
els
zo
n
e
to
m
ak
e
th
e
o
b
ject
s
d
is
tin
ctly
id
en
tifie
d
in
th
e
s
c
en
e.
E
ac
h
elem
e
n
t
in
s
id
e
th
e
b
o
u
n
d
a
r
y
will
h
av
e
a
r
ef
in
e
z
o
n
e
to
clea
r
ly
id
e
n
tif
y
th
e
im
ag
e
an
d
id
en
tify
th
e
f
u
n
ctio
n
o
f
th
e
o
b
ject.
Hen
ce
,
th
e
f
ea
tu
r
e
e
n
ab
les
th
e
im
ag
e
ca
p
tu
r
ed
with
m
o
r
e
s
p
ec
if
ic
an
d
d
is
tin
ctly
id
en
tifi
ed
b
o
u
n
d
ar
ies.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
ch
20
2
6
:
7
1
-
79
76
Fig
u
r
e
5
.
Ov
e
r
lay
ed
Seg
m
en
ta
tio
n
o
f
a
s
ce
n
e
Fig
u
r
e
6
.
His
to
g
r
am
o
f
p
ix
el
d
is
tr
ib
u
tio
n
Fig
u
r
e
7
.
C
lean
ed
m
ask
o
f
a
s
ce
n
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
S
ema
n
tic
s
eg
men
ta
tio
n
fo
r
d
a
ta
va
lid
a
tio
n
in
u
n
ma
n
n
e
d
r
o
b
o
tic
ve
h
icles
(
I
va
n
S
u
n
it R
o
u
t
)
77
4.
CO
NCLU
SI
O
N
T
h
e
m
o
d
el
d
e
m
o
n
s
tr
ated
h
ig
h
ac
cu
r
ac
y
in
d
etec
tin
g
an
d
s
eg
m
en
tin
g
o
b
jects,
with
p
er
f
o
r
m
a
n
ce
m
etr
ics
s
u
ch
as
I
o
U
an
d
Di
ce
co
ef
f
icien
t
in
d
icatin
g
s
tr
o
n
g
alig
n
m
en
t
b
etwe
en
p
r
ed
i
cted
an
d
s
em
an
tic
s
eg
m
en
tatio
n
.
Misclass
if
icatio
n
an
d
s
eg
m
e
n
tatio
n
er
r
o
r
s
wer
e
id
en
tifie
d
in
c
h
allen
g
in
g
c
o
n
d
itio
n
s
lik
e
f
o
g
,
r
ain
,
an
d
lo
w
-
lig
h
t
en
v
ir
o
n
m
e
n
ts
.
So
m
e
in
s
tan
ce
s
o
f
in
co
m
p
lete
co
n
to
u
r
s
an
d
b
o
u
n
d
ar
y
m
is
alig
n
m
en
t
wer
e
n
o
ted
.
Pre
p
r
o
ce
s
s
in
g
tech
n
iq
u
es,
in
clu
d
in
g
d
ata
au
g
m
en
tatio
n
an
d
n
o
is
e
r
ed
u
ctio
n
,
s
ig
n
if
ican
tly
im
p
r
o
v
e
d
th
e
m
o
d
el’
s
ab
ilit
y
to
g
en
er
alize
ac
r
o
s
s
d
if
f
er
e
n
t
s
ce
n
ar
io
s
.
Pr
o
ce
s
s
ed
lab
els
co
n
tr
i
b
u
ted
to
b
etter
s
eg
m
en
tatio
n
co
n
s
is
ten
cy
.
T
h
e
m
o
d
el’
s
p
e
r
f
o
r
m
an
ce
was
ev
alu
ated
in
r
e
al
-
wo
r
ld
d
r
i
v
in
g
c
o
n
d
itio
n
s
.
W
h
ile
it
ef
f
ec
tiv
ely
d
etec
ted
r
o
ad
elem
e
n
ts
in
clea
r
wea
th
er
,
p
er
f
o
r
m
an
ce
d
ec
lin
ed
s
lig
h
tly
in
co
m
p
lex
en
v
ir
o
n
m
en
ts
,
h
ig
h
lig
h
tin
g
th
e
n
ee
d
f
o
r
f
u
r
t
h
er
r
ef
in
em
en
t.
A
m
an
u
al
v
e
r
if
icatio
n
p
r
o
ce
s
s
r
ev
ea
led
th
at
th
e
m
o
d
el
clo
s
ely
m
atch
ed
h
u
m
an
-
a
n
n
o
tated
la
b
els
in
m
o
s
t
ca
s
es,
b
u
t
o
cc
asio
n
al
d
is
cr
ep
an
cies
r
eq
u
ir
ed
ad
d
itio
n
al
f
in
e
-
tu
n
i
n
g
.
T
h
e
s
y
s
tem
was
o
p
tim
ized
f
o
r
f
aster
in
f
er
en
ce
tim
es,
en
s
u
r
in
g
r
ea
l
-
tim
e
ap
p
licab
ilit
y
in
a
u
to
n
o
m
o
u
s
v
eh
icle
s
y
s
tem
s
with
o
u
t c
o
m
p
r
o
m
is
in
g
ac
cu
r
ac
y
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
N
o
f
u
n
d
in
g
was
in
v
o
lv
ed
in
th
e
r
esear
ch
wo
r
k
ca
r
r
ied
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
I
v
an
Su
n
it R
o
u
t
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
P
Pal Pan
d
ian
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
An
il R
aj
✓
✓
✓
✓
✓
✓
✓
✓
✓
An
il M
elwy
n
R
eg
o
✓
✓
✓
✓
✓
✓
✓
✓
Sajn
a
Par
im
ita
Pan
ig
r
a
h
i
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
−
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
ailab
l
e
o
n
r
eq
u
est
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
,
[
I
SR
]
.
T
h
e
d
ata,
wh
ich
co
n
tain
in
f
o
r
m
atio
n
th
at
co
u
ld
co
m
p
r
o
m
is
e
th
e
p
r
iv
ac
y
o
f
r
esear
c
h
p
ar
ticip
an
ts
,
ar
e
n
o
t p
u
b
licly
av
aila
b
le
d
u
e
to
ce
r
tain
r
estrictio
n
s
.
−
Der
iv
ed
d
ata
s
u
p
p
o
r
tin
g
th
e
f
i
n
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
aila
b
le
f
r
o
m
t
h
e
c
o
r
r
esp
o
n
d
in
g
au
th
o
r
[
I
SR
]
o
n
r
eq
u
est.
−
T
h
e
d
ata
t
h
at
s
u
p
p
o
r
t
th
e
f
in
d
i
n
g
s
o
f
th
is
s
tu
d
y
a
r
e
av
ailab
le
f
r
o
m
t
h
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
,
[
I
SR
]
,
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
A
n
a
n
t
h
a
r
a
ma
n
,
M
.
V
e
l
a
z
q
u
e
z
,
a
n
d
Y
.
Le
e
,
“
U
t
i
l
i
z
i
n
g
M
a
s
k
R
-
C
N
N
f
o
r
D
e
t
e
c
t
i
o
n
a
n
d
S
e
g
m
e
n
t
a
t
i
o
n
o
f
O
r
a
l
D
i
s
e
a
s
e
s,
”
i
n
Pro
c
e
e
d
i
n
g
s
-
2
0
1
8
I
EE
E
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
B
i
o
i
n
f
o
rm
a
t
i
c
s
a
n
d
Bi
o
m
e
d
i
c
i
n
e
,
BI
BM
2
0
1
8
,
D
e
c
.
2
0
1
9
,
p
p
.
2
1
9
7
–
2
2
0
4
.
d
o
i
:
1
0
.
1
1
0
9
/
B
I
B
M
.
2
0
1
8
.
8
6
2
1
1
1
2
.
[
2
]
S
.
F
a
n
g
,
B
.
Zh
a
n
g
,
a
n
d
J.
H
u
,
“
I
mp
r
o
v
e
d
M
a
s
k
R
-
C
N
N
mu
l
t
i
-
t
a
r
g
e
t
d
e
t
e
c
t
i
o
n
a
n
d
se
g
me
n
t
a
t
i
o
n
f
o
r
a
u
t
o
n
o
mo
u
s
d
r
i
v
i
n
g
i
n
c
o
m
p
l
e
x
sc
e
n
e
s,”
S
e
n
so
rs
,
v
o
l
.
2
3
,
n
o
.
8
,
p
.
3
8
5
3
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
3
0
8
3
8
5
3
.
[
3
]
E.
S
h
e
l
h
a
mer,
J.
Lo
n
g
,
a
n
d
T.
D
a
r
r
e
l
l
,
“
F
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
s
f
o
r
sem
a
n
t
i
c
s
e
g
me
n
t
a
t
i
o
n
,
”
i
n
I
EEE
T
ra
n
sa
c
t
i
o
n
s
o
n
Pa
t
t
e
r
n
An
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
,
S
e
p
.
2
0
1
7
,
v
o
l
.
3
9
,
n
o
.
4
,
p
p
.
6
4
0
–
6
5
1
.
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
1
6
.
2
5
7
2
6
8
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
7
2
2
-
2
5
8
6
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
ch
20
2
6
:
7
1
-
79
78
[
4
]
H
.
H
e
e
t
a
l
.
,
“
M
a
s
k
R
-
C
N
N
b
a
s
e
d
a
u
t
o
ma
t
e
d
i
d
e
n
t
i
f
i
c
a
t
i
o
n
a
n
d
e
x
t
r
a
c
t
i
o
n
o
f
o
i
l
w
e
l
l
si
t
e
s,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
Ap
p
l
i
e
d
Ea
rt
h
O
b
ser
v
a
t
i
o
n
a
n
d
G
e
o
i
n
f
o
rm
a
t
i
o
n
,
v
o
l
.
1
1
2
,
p
.
1
0
2
8
7
5
,
A
u
g
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
a
g
.
2
0
2
2
.
1
0
2
8
7
5
.
[
5
]
V
.
B
a
d
r
i
n
a
r
a
y
a
n
a
n
,
A
.
K
e
n
d
a
l
l
,
a
n
d
R
.
C
i
p
o
l
l
a
,
“
S
e
g
N
e
t
:
A
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
e
n
c
o
d
e
r
-
d
e
c
o
d
e
r
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
i
m
a
g
e
seg
m
e
n
t
a
t
i
o
n
,
”
I
E
EE
T
r
a
n
sa
c
t
i
o
n
s o
n
Pa
t
t
e
rn
An
a
l
y
si
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
9
,
n
o
.
1
2
,
p
p
.
2
4
8
1
–
2
4
9
5
,
D
e
c
.
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
1
6
.
2
6
4
4
6
1
5
.
[
6
]
O
.
R
o
n
n
e
b
e
r
g
e
r
,
P
.
F
i
sc
h
e
r
,
a
n
d
T.
B
r
o
x
,
“
U
t
i
l
i
z
i
n
g
M
a
s
k
R
-
C
N
N
f
o
r
d
e
t
e
c
t
i
o
n
a
n
d
s
e
g
me
n
t
a
t
i
o
n
o
f
o
r
a
l
d
i
sea
ses
,
”
L
e
c
t
u
r
e
N
o
t
e
s
i
n
C
o
m
p
u
t
e
r S
c
i
e
n
c
e
(
i
n
c
l
u
d
i
n
g
s
u
b
se
ri
e
s
L
e
c
t
u
r
e
N
o
t
e
s i
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
r
e
N
o
t
e
s i
n
B
i
o
i
n
f
o
rm
a
t
i
c
s)
,
v
o
l
.
9
3
5
1
,
p
p
.
2
3
4
–
2
4
1
,
2
0
1
5
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
3
1
9
-
2
4
5
7
4
-
4
_
2
8
.
[
7
]
L.
C
.
C
h
e
n
,
G
.
P
a
p
a
n
d
r
e
o
u
,
I
.
K
o
k
k
i
n
o
s,
K
.
M
u
r
p
h
y
,
a
n
d
A
.
L.
Y
u
i
l
l
e
,
“
D
e
e
p
La
b
:
S
e
m
a
n
t
i
c
i
ma
g
e
se
g
me
n
t
a
t
i
o
n
w
i
t
h
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
s
,
a
t
r
o
u
s
c
o
n
v
o
l
u
t
i
o
n
,
a
n
d
f
u
l
l
y
c
o
n
n
e
c
t
e
d
C
R
F
s
,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
Pa
t
t
e
rn
A
n
a
l
y
s
i
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
0
,
n
o
.
4
,
p
p
.
8
3
4
–
8
4
8
,
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
1
7
.
2
6
9
9
1
8
4
.
[
8
]
J.
C
h
e
n
g
,
S
.
T
i
a
n
,
L
.
Y
u
,
H
.
L
u
,
a
n
d
X
.
L
v
,
“
F
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
f
o
r
b
i
o
m
e
d
i
c
a
l
i
m
a
g
e
se
g
me
n
t
a
t
i
o
n
,
”
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
i
n
Me
d
i
c
i
n
e
,
v
o
l
.
1
0
7
,
p
.
1
0
1
8
9
9
,
Ju
l
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
r
t
med
.
2
0
2
0
.
1
0
1
8
9
9
.
[
9
]
M
.
C
o
r
d
t
s
e
t
a
l
.
,
“
T
h
e
C
i
t
y
s
c
a
p
e
s
d
a
t
a
set
f
o
r
se
ma
n
t
i
c
u
r
b
a
n
sce
n
e
u
n
d
e
r
st
a
n
d
i
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
I
E
EE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
si
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
J
u
n
.
2
0
1
6
,
v
o
l
.
2
0
1
6
-
D
e
c
e
m
,
p
p
.
3
2
1
3
–
3
2
2
3
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
3
5
0
.
[
1
0
]
D
.
Tr
i
p
h
e
n
a
D
e
l
i
g
h
t
a
n
d
V
.
K
a
r
u
n
a
k
a
r
a
n
,
“
D
e
e
p
l
e
a
r
n
i
n
g
b
a
s
e
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
u
si
n
g
M
a
s
k
R
C
N
N
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
6
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
El
e
c
t
ro
n
i
c
s
S
y
s
t
e
m
s,
I
C
C
E
S
2
0
2
1
,
A
u
g
.
2
0
2
1
,
p
p
.
1
6
8
4
–
1
6
9
0
.
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
ES
5
1
3
5
0
.
2
0
2
1
.
9
4
8
9
1
5
2
.
[
1
1
]
P
.
W
.
K
o
h
a
n
d
P
.
L
i
a
n
g
,
“
U
n
d
e
r
s
t
a
n
d
i
n
g
b
l
a
c
k
-
b
o
x
p
r
e
d
i
c
t
i
o
n
s
v
i
a
i
n
f
l
u
e
n
c
e
f
u
n
c
t
i
o
n
s,
”
i
n
3
4
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
,
I
C
ML
2
0
1
7
,
M
a
r
.
2
0
1
7
,
v
o
l
.
4
,
p
p
.
2
9
7
6
–
2
9
8
7
.
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
1
7
0
3
.
0
4
7
3
0
.
[
1
2
]
K
.
H
e
,
G
.
G
k
i
o
x
a
r
i
,
P
.
D
o
l
l
á
r
,
a
n
d
R
.
G
i
r
sh
i
c
k
,
“
M
a
sk
R
-
C
N
N
,
”
i
n
P
ro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
EEE
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
,
M
a
r
.
2
0
1
7
,
p
p
.
2
9
6
1
–
2
9
6
9
.
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
1
7
0
3
.
0
6
8
7
0
.
[
1
3
]
T.
Y
o
o
n
a
n
d
H
.
K
i
m,
“
U
n
c
e
r
t
a
i
n
t
y
e
st
i
m
a
t
i
o
n
b
y
d
e
n
si
t
y
a
w
a
r
e
e
v
i
d
e
n
t
i
a
l
d
e
e
p
l
e
a
r
n
i
n
g
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
Re
se
a
rc
h
,
S
e
p
.
2
0
2
4
,
v
o
l
.
2
3
5
,
p
p
.
5
7
2
1
7
–
5
7
2
4
3
.
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
2
4
0
9
.
0
8
7
5
4
.
[
1
4
]
S
.
Y
.
H
u
a
n
g
,
W
.
L
.
H
s
u
,
R
.
J.
H
s
u
,
a
n
d
D
.
W
.
L
i
u
,
“
F
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
f
o
r
t
h
e
sem
a
n
t
i
c
s
e
g
m
e
n
t
a
t
i
o
n
o
f
me
d
i
c
a
l
i
ma
g
e
s
:
A
s
u
r
v
e
y
,
”
D
i
a
g
n
o
st
i
c
s
,
v
o
l
.
1
2
,
n
o
.
1
1
,
p
.
2
7
6
5
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
d
i
a
g
n
o
st
i
c
s
1
2
1
1
2
7
6
5
.
[
1
5
]
C
.
G
.
N
o
r
t
h
c
u
t
t
,
A
.
A
t
h
a
l
y
e
,
a
n
d
J.
M
u
e
l
l
e
r
,
“
P
e
r
v
a
si
v
e
l
a
b
e
l
e
r
r
o
r
s
i
n
t
e
st
s
e
t
s
d
e
s
t
a
b
i
l
i
z
e
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
b
e
n
c
h
m
a
r
k
s,
”
Ad
v
a
n
c
e
s i
n
N
e
u
r
a
l
I
n
f
o
rm
a
t
i
o
n
Pro
c
e
ssi
n
g
S
y
st
e
m
s
,
2
0
2
1
.
[
1
6
]
Z.
T
o
n
g
,
P
.
X
u
,
a
n
d
T.
D
e
n
œ
u
x
,
“
Ev
i
d
e
n
t
i
a
l
f
u
l
l
y
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
f
o
r
sem
a
n
t
i
c
se
g
me
n
t
a
t
i
o
n
,
”
A
p
p
l
i
e
d
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
5
1
,
n
o
.
9
,
p
p
.
6
3
7
6
–
6
3
9
9
,
M
a
r
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
8
9
-
021
-
0
2
3
2
7
-
0.
[
1
7
]
W
.
W
e
n
g
a
n
d
X
.
Z
h
u
,
“
I
N
e
t
:
C
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
s
f
o
r
b
i
o
me
d
i
c
a
l
i
mag
e
s
e
g
m
e
n
t
a
t
i
o
n
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
1
6
5
9
1
–
1
6
6
0
3
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
0
5
3
4
0
8
.
[
1
8
]
H
.
Zh
a
n
g
e
t
a
l
.
,
“
R
e
sN
e
S
t
:
S
p
l
i
t
-
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
s,”
i
n
A
d
v
a
n
c
e
s
i
n
N
e
u
ra
l
I
n
f
o
rm
a
t
i
o
n
Pr
o
c
e
ssi
n
g
S
y
st
e
m
s
,
D
e
c
.
2
0
2
0
.
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
2
0
0
4
.
0
8
9
5
5
.
[
1
9
]
H
.
Zh
a
n
g
e
t
a
l
.
,
“
R
e
sN
e
S
t
:
S
p
l
i
t
-
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
s,
”
i
n
I
E
EE
C
o
m
p
u
t
e
r
S
o
c
i
e
t
y
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
s
i
o
n
a
n
d
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
Wo
r
k
sh
o
p
s
,
Ju
n
.
2
0
2
2
,
v
o
l
.
2
0
2
2
-
J
u
n
e
,
p
p
.
2
7
3
5
–
2
7
4
5
.
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
W
5
6
3
4
7
.
2
0
2
2
.
0
0
3
0
9
.
[
2
0
]
H
.
Zu
n
a
i
r
a
n
d
A
.
B
e
n
H
a
mza
,
“
S
h
a
r
p
U
-
N
e
t
:
D
e
p
t
h
w
i
se
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
f
o
r
b
i
o
me
d
i
c
a
l
i
m
a
g
e
s
e
g
m
e
n
t
a
t
i
o
n
,
”
C
o
m
p
u
t
e
rs
i
n
Bi
o
l
o
g
y
a
n
d
M
e
d
i
c
i
n
e
,
v
o
l
.
1
3
6
,
p
.
1
0
4
6
9
9
,
S
e
p
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mp
b
i
o
me
d
.
2
0
2
1
.
1
0
4
6
9
9
.
[
2
1
]
A
.
I
q
b
a
l
,
M
.
S
h
a
r
i
f
,
M
.
A
.
K
h
a
n
,
W
.
N
i
sar,
a
n
d
M
.
A
l
h
a
i
so
n
i
,
“
F
F
-
U
N
e
t
:
A
U
-
sh
a
p
e
d
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
mu
l
t
i
m
o
d
a
l
b
i
o
me
d
i
c
a
l
i
ma
g
e
se
g
men
t
a
t
i
o
n
,
”
C
o
g
n
i
t
i
v
e
C
o
m
p
u
t
a
t
i
o
n
,
v
o
l
.
1
4
,
n
o
.
4
,
p
p
.
1
2
8
7
–
1
3
0
2
,
Ju
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
2
5
5
9
-
0
2
2
-
1
0
0
3
8
-
y.
[
2
2
]
S
.
F
u
j
i
t
a
a
n
d
X
.
H
.
H
a
n
,
“
C
e
l
l
d
e
t
e
c
t
i
o
n
a
n
d
s
e
g
me
n
t
a
t
i
o
n
i
n
m
i
c
r
o
sc
o
p
y
i
mag
e
s
w
i
t
h
i
m
p
r
o
v
e
d
M
a
sk
R
-
C
N
N
,
”
i
n
L
e
c
t
u
r
e
N
o
t
e
s
i
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
(
i
n
c
l
u
d
i
n
g
s
u
b
s
e
ri
e
s L
e
c
t
u
r
e
N
o
t
e
s
i
n
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
L
e
c
t
u
re
N
o
t
e
s i
n
Bi
o
i
n
f
o
rm
a
t
i
c
s)
,
F
e
b
.
2
0
2
1
,
v
o
l
.
1
2
6
2
8
LN
C
S
,
p
p
.
5
8
–
7
0
.
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
0
-
6
9
7
5
6
-
3
_
5
.
[
2
3
]
X
.
W
a
n
g
,
W
.
H
e
,
D
.
M
e
t
a
x
a
s
,
R
.
M
a
t
h
e
w
,
a
n
d
E.
W
h
i
t
e
,
“
C
e
l
l
s
e
g
m
e
n
t
a
t
i
o
n
a
n
d
t
r
a
c
k
i
n
g
u
si
n
g
t
e
x
t
u
r
e
-
a
d
a
p
t
i
v
e
sn
a
k
e
s
,
”
i
n
2
0
0
7
4
t
h
I
E
EE
I
n
t
e
r
n
a
t
i
o
n
a
l
S
y
m
p
o
si
u
m
o
n
Bi
o
m
e
d
i
c
a
l
I
m
a
g
i
n
g
:
Fro
m
N
a
n
o
t
o
Ma
c
r
o
-
Pro
c
e
e
d
i
n
g
s
,
M
a
y
2
0
0
7
,
p
p
.
1
0
1
–
1
0
4
.
d
o
i
:
1
0
.
1
1
0
9
/
I
S
B
I
.
2
0
0
7
.
3
5
6
7
9
8
.
[
2
4
]
J.
Ja
n
a
i
,
F
.
G
ü
n
e
y
,
A
.
B
e
h
l
,
a
n
d
A
.
G
e
i
g
e
r
,
“
C
o
m
p
u
t
e
r
v
i
si
o
n
f
o
r
a
u
t
o
n
o
mo
u
s
v
e
h
i
c
l
e
s,”
F
o
u
n
d
a
t
i
o
n
s
a
n
d
T
r
e
n
d
s
i
n
C
o
m
p
u
t
e
r
G
ra
p
h
i
c
s
a
n
d
Vi
s
i
o
n
,
v
o
l
.
1
2
,
n
o
.
1
–
3
,
p
p
.
1
–
3
0
8
,
J
u
l
.
2
0
2
0
,
d
o
i
:
1
0
.
1
5
6
1
/
0
6
0
0
0
0
0
0
7
9
.
[
2
5
]
S
.
G
r
i
g
o
r
e
s
c
u
,
B
.
Tr
a
s
n
e
a
,
T
.
C
o
c
i
a
s,
a
n
d
G
.
M
a
c
e
sa
n
u
,
“
A
su
r
v
e
y
o
f
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
f
o
r
a
u
t
o
n
o
mo
u
s
d
r
i
v
i
n
g
,
”
J
o
u
rn
a
l
o
f
Fi
e
l
d
R
o
b
o
t
i
c
s
,
v
o
l
.
3
7
,
n
o
.
3
,
p
p
.
3
6
2
–
3
8
6
,
M
a
r
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
2
/
r
o
b
.
2
1
9
1
8
.
[
2
6
]
F
.
I
.
D
i
a
k
o
g
i
a
n
n
i
s,
F
.
W
a
l
d
n
e
r
,
P
.
C
a
c
c
e
t
t
a
,
a
n
d
C
.
W
u
,
“
R
e
sU
N
e
t
-
a
:
A
d
e
e
p
l
e
a
r
n
i
n
g
f
r
a
mew
o
r
k
f
o
r
sem
a
n
t
i
c
se
g
me
n
t
a
t
i
o
n
o
f
r
e
mo
t
e
l
y
se
n
se
d
d
a
t
a
,
”
I
S
PRS
J
o
u
r
n
a
l
o
f
P
h
o
t
o
g
ra
m
m
e
t
ry
a
n
d
Re
m
o
t
e
S
e
n
si
n
g
,
v
o
l
.
1
6
2
,
p
p
.
9
4
–
1
1
4
,
A
p
r
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
s
p
r
sj
p
r
s.
2
0
2
0
.
0
1
.
0
1
3
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Iv
a
n
S
u
n
it
R
o
u
t
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
th
e
S
c
h
o
o
l
o
f
En
g
in
e
e
rin
g
a
n
d
Tec
h
n
o
l
o
g
y
,
C
h
rist
Un
i
v
e
rsity
,
B
a
n
g
a
lo
re
.
He
h
a
s p
u
b
li
sh
e
d
5
tex
t
b
o
o
k
s
a
n
d
5
p
a
ten
ts.
He
is
a
g
o
l
d
m
e
d
a
li
st
in
h
is
M
.
Tec
h
d
e
g
r
e
e
a
n
d
wo
n
t
h
e
Be
st
Tea
c
h
e
r
Aw
a
rd
in
2
0
2
4
.
He
h
a
s
3
se
e
d
m
o
n
e
y
p
r
o
jec
ts
a
n
d
o
n
e
e
x
tern
a
ll
y
fu
n
d
e
d
p
r
o
jec
t
fro
m
IS
RO
a
n
d
NIT
Tri
c
h
y
.
His
m
a
i
n
re
se
a
rc
h
d
irec
ti
o
n
s
in
c
l
u
d
e
a
u
t
o
m
a
ti
o
n
,
m
o
b
il
e
r
o
b
o
ti
c
s,
u
n
d
e
rwa
ter
ro
b
o
ti
c
s,
a
u
t
o
n
o
m
o
u
s
a
n
d
u
n
m
a
n
n
e
d
ro
b
o
ti
c
v
e
h
icle
s,
sm
a
rt
fa
c
to
ry
,
o
p
ti
m
iza
ti
o
n
tec
h
n
i
q
u
e
s,
a
d
d
it
iv
e
m
a
n
u
fa
c
tu
r
in
g
,
a
n
d
In
d
u
str
y
4
.
0
.
He
c
a
n
b
e
c
o
n
ta
c
ted
a
t
iv
a
n
.
ro
u
t@c
h
rist
u
n
i
v
e
rsity
.
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
AE
S
I
n
t
J
R
o
b
&
A
u
to
m
I
SS
N:
2722
-
2
5
8
6
S
ema
n
tic
s
eg
men
ta
tio
n
fo
r
d
a
ta
va
lid
a
tio
n
in
u
n
ma
n
n
e
d
r
o
b
o
tic
ve
h
icles
(
I
va
n
S
u
n
it R
o
u
t
)
79
P
Pa
l
Pa
n
d
i
a
n
is
a
p
ro
fe
ss
o
r
in
th
e
S
c
h
o
o
l
o
f
E
n
g
i
n
e
e
rin
g
a
n
d
Tec
h
n
o
lo
g
y
,
Ch
rist
Un
iv
e
rsity
,
Ba
n
g
a
lo
re
.
He
h
a
s
p
u
b
li
s
h
e
d
1
0
tex
t
b
o
o
k
s
a
n
d
2
5
p
a
te
n
ts
a
n
d
h
a
s
g
u
id
e
d
2
re
se
a
rc
h
sc
h
o
lars
.
He
h
a
s
wo
n
Be
st
Tea
c
h
e
r
Aw
a
rd
s
o
n
v
a
rio
u
s
o
c
c
a
sio
n
s
a
n
d
h
a
s
d
e
li
v
e
re
d
talk
s
o
n
a
d
v
a
n
c
e
d
to
p
ics
a
t
re
n
o
wn
e
d
i
n
stit
u
ti
o
n
s
a
c
ro
ss
t
h
e
c
o
u
n
try
.
He
is
c
u
rre
n
tl
y
wo
r
k
in
g
o
n
h
u
m
a
n
o
id
r
o
b
o
ts
a
n
d
d
ro
n
e
s.
His
m
a
in
re
se
a
rc
h
d
irec
ti
o
n
s
i
n
c
lu
d
e
traje
c
to
r
y
p
la
n
n
i
n
g
,
k
in
e
m
a
ti
c
s
a
n
d
d
y
n
a
m
ics
o
f
r
o
b
o
t
m
e
c
h
a
n
ism
s,
d
r
o
n
e
n
a
v
i
g
a
ti
o
n
a
n
d
c
o
n
tro
l
,
h
u
m
a
n
o
id
ro
b
o
ts,
a
u
to
n
o
m
o
u
s
a
n
d
u
n
m
a
n
n
e
d
ro
b
o
t
ic
v
e
h
icle
s,
a
n
d
I
n
d
u
stry
4
.
0
.
He
a
lso
r
u
n
s
h
is
o
w
n
c
o
n
su
lt
i
n
g
firm,
P
latin
u
m
Co
n
su
l
ti
n
g
P
v
t.
Lt
d
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
p
a
l.
p
a
n
d
ian
@c
h
rist
u
n
iv
e
rsity
.
in
.
Ani
l
Ra
j
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
th
e
S
c
h
o
o
l
o
f
E
n
g
i
n
e
e
rin
g
a
n
d
Tec
h
n
o
lo
g
y
,
Ch
rist
Un
iv
e
rsit
y
,
Ba
n
g
a
l
o
re
.
He
h
a
s p
u
b
l
ish
e
d
1
te
x
tb
o
o
k
a
n
d
5
p
a
ten
ts.
He
h
a
s 2
se
e
d
m
o
n
e
y
p
ro
jec
ts
a
n
d
h
a
s g
u
i
d
e
d
o
n
e
re
se
a
rc
h
sc
h
o
lar.
His
m
a
in
re
se
a
rc
h
d
ir
e
c
ti
o
n
s
i
n
c
lu
d
e
a
u
t
o
m
a
ti
o
n
,
in
d
u
strial
r
o
b
o
ti
c
s,
sm
a
rt
m
a
teria
ls,
a
d
d
it
iv
e
m
a
n
u
fa
c
tu
rin
g
,
m
e
tal
m
a
c
h
in
in
g
,
a
n
d
o
p
ti
m
iza
ti
o
n
tec
h
n
iq
u
e
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
a
n
i
l.
ra
j@c
h
ristu
n
i
v
e
rsity
.
in
.
Ani
l
Melwy
n
Re
g
o
is
a
n
a
ss
istan
t
p
r
o
fe
ss
o
r
i
n
t
h
e
S
c
h
o
o
l
o
f
En
g
in
e
e
rin
g
a
n
d
Tec
h
n
o
l
o
g
y
,
C
h
rist
Un
i
v
e
rsity
,
Ba
n
g
a
lo
re
.
His
m
a
in
re
se
a
rc
h
d
i
re
c
ti
o
n
s
in
c
l
u
d
e
S
ix
S
ig
m
a
,
p
ro
jec
t
m
a
n
a
g
e
m
e
n
t,
o
p
e
ra
ti
o
n
s
m
a
n
a
g
e
m
e
n
t,
in
d
u
strial
ro
b
o
t
ics
,
a
n
d
ro
b
o
t
ic
n
a
v
ig
a
ti
o
n
a
n
d
c
o
n
tro
l.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
a
n
il
.
re
g
o
@c
h
r
istu
n
iv
e
rsit
y
.
i
n
.
S
a
jn
a
P
a
r
im
it
a
P
a
n
i
g
r
a
h
i
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
th
e
S
c
h
o
o
l
o
f
En
g
in
e
e
rin
g
a
n
d
Tec
h
n
o
l
o
g
y
,
C
h
rist
Un
iv
e
rs
it
y
,
Ba
n
g
a
lo
re
.
He
r
m
a
in
re
se
a
rc
h
d
irec
ti
o
n
s
i
n
c
lu
d
e
lea
n
m
a
n
u
fa
c
tu
rin
g
,
a
d
d
i
ti
v
e
m
a
n
u
fa
c
tu
rin
g
,
r
o
b
o
t
k
i
n
e
m
a
ti
c
s
a
n
d
d
y
n
a
m
ics
o
f
c
o
n
tr
o
l,
a
n
d
e
n
g
in
e
e
rin
g
g
ra
p
h
ics
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
sa
jn
a
.
p
a
n
i
g
ra
h
i@c
h
ri
stu
n
i
v
e
rsity
.
i
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.