I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
1
6
,
No
.
2
,
A
p
r
il
20
2
6
,
p
p
.
863
~
872
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
1
6
i
2
.
pp
8
6
3
-
8
7
2
863
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Integ
ra
ted
d
eep l
ea
rning
appro
a
ch f
o
r re
a
l
-
time ob
j
ect
detec
tion a
nd col
o
r ana
ly
sis
Srini
v
a
s
D
ib
bu
r
B
y
ra
pp
a
1
,
K
us
ha
l G
a
j
endra
1
,
Ro
hith H
o
lena
ra
s
ipu
ra
P
utt
a
ra
j
u
1
,
T
um
a
k
a
la
ha
lli
N
a
g
a
ra
j
M
a
li
ni
2
1
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
o
n
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
N
i
t
t
e
M
e
e
n
a
k
sh
i
I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
,
B
a
n
g
a
l
u
r
u
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
M
a
n
a
g
e
me
n
t
S
t
u
d
i
e
s,
N
i
t
t
e
M
e
e
n
a
k
s
h
i
I
n
st
i
t
u
t
e
o
f
Te
c
h
n
o
l
o
g
y
,
B
a
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
May
17
,
2
0
2
5
R
ev
is
ed
Dec
3
,
2
0
2
5
Acc
ep
ted
J
an
15
,
2
0
2
6
Ob
jec
t
id
e
n
ti
fica
ti
o
n
is
o
n
e
o
f
th
e
m
a
jo
r
a
p
p
li
c
a
ti
o
n
a
re
a
s
o
f
d
e
e
p
lea
rn
i
n
g
th
a
t
p
r
o
v
id
e
s
sig
n
ifi
c
a
n
tl
y
b
e
tt
e
r
fe
a
tu
re
e
x
trac
ti
o
n
a
n
d
re
p
re
se
n
t
a
ti
o
n
th
a
n
m
o
re
c
o
n
v
e
n
t
io
n
a
l
m
e
th
o
d
s
o
f
re
c
o
g
n
it
i
o
n
.
Driv
e
n
b
y
t
h
e
g
ro
win
g
sig
n
ifi
c
a
n
c
e
o
f
c
o
n
ju
n
c
ti
o
n
o
f
o
b
jec
ts
d
e
tec
ti
o
n
a
n
d
c
o
lo
r
i
n
terp
re
tatio
n
in
c
o
n
tem
p
o
ra
ry
c
o
m
p
u
ter
v
isio
n
sy
ste
m
s,
th
e
c
u
rre
n
t
w
o
rk
p
r
o
p
o
se
s
a
n
in
teg
ra
ted
,
re
a
l
-
ti
m
e
d
e
e
p
lea
rn
i
n
g
sy
ste
m
t
h
a
t
c
o
m
p
lete
s
t
h
e
tas
k
o
f
o
b
jec
t
lo
c
a
li
z
a
ti
o
n
a
n
d
c
o
l
o
r
a
n
a
ly
sis.
It
is
s
u
g
g
e
ste
d
t
h
a
t
t
h
e
p
ro
p
o
se
d
sy
ste
m
e
m
p
lo
y
s
a
fa
ste
r
re
g
i
o
n
-
b
a
se
d
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
r
k
(
F
a
ste
r
R
-
CNN
)
with
b
a
c
k
b
o
n
e
o
f
Re
sN
e
t
-
5
0
a
n
d
su
p
p
lem
e
n
ted
wit
h
a
fe
a
tu
r
e
p
y
ra
m
id
n
e
two
rk
t
o
p
e
rfo
rm
m
u
lt
i
-
sc
a
le
fe
a
tu
re
a
g
g
re
g
a
ti
o
n
.
Th
e
m
o
d
e
l
w
a
s
train
e
d
a
n
d
tes
ted
u
sin
g
t
h
e
P
a
sc
a
l
VO
C
2
0
1
2
d
a
tas
e
t
a
n
d
i
t
sh
o
we
d
g
o
o
d
re
su
lt
s
with
t
h
e
a
v
e
ra
g
e
p
re
c
isio
n
o
f
0
.
8
1
1
4
,
F
1
o
f
0
.
6
2
3
2
a
n
d
I
o
U
o
f
0
.
7
0
9
6
.
Th
e
larg
e
se
t
o
f
e
x
p
e
rime
n
ts
o
n
d
if
fe
re
n
t
lea
rn
i
n
g
ra
tes
a
n
d
trai
n
in
g
e
p
o
c
h
s
a
ll
o
we
d
o
p
ti
m
izin
g
t
h
e
d
e
tec
to
r
t
o
wo
rk
we
ll
in
a
v
a
riet
y
o
f
c
o
n
d
it
io
n
s
.
To
e
n
h
a
n
c
e
e
v
e
n
m
o
re
,
v
isu
a
li
z
a
ti
o
n
h
isto
g
ra
m
o
f
o
rien
ted
g
ra
d
ien
ts
(
HOG
)
a
n
d
g
ra
d
ie
n
t
-
we
ig
h
ted
c
las
s
a
c
ti
v
a
ti
o
n
m
a
p
p
in
g
(
G
ra
d
-
CAM)
w
a
s
u
se
d
to
g
a
in
a
m
o
re
p
ro
f
o
u
n
d
u
n
d
e
rsta
n
d
in
g
o
f
th
e
si
g
n
if
ica
n
c
e
o
f
fe
a
tu
r
e
s
a
n
d
th
e
lo
g
ic
b
e
h
in
d
a
m
o
d
e
l
.
Th
is
st
u
d
y
c
o
m
p
lem
e
n
ts
ima
g
e
p
e
rc
e
p
ti
o
n
with
c
o
l
o
r
b
y
c
o
m
b
in
in
g
o
b
jec
t
re
c
o
g
n
it
i
o
n
a
n
d
c
o
lo
r
in
a
sin
g
le
a
rc
h
it
e
c
tu
re
,
wh
ic
h
c
a
n
re
su
lt
in
fr
u
it
f
u
l
a
p
p
li
c
a
ti
o
n
s
in
a
re
a
s
o
f
a
u
to
n
o
m
o
u
s
v
e
h
icle
s,
in
d
u
strial
a
u
to
m
a
ti
o
n
,
a
n
d
m
e
d
ica
l
ima
g
i
n
g
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
Featu
r
e
p
y
r
am
i
d
n
etwo
r
k
Gr
ad
ien
t
class
ac
tiv
atio
n
His
to
g
r
am
o
f
o
r
ien
ted
g
r
ad
ien
ts
Ob
ject
d
etec
tio
n
Pas
ca
l V
O
C
R
eg
io
n
p
r
o
p
o
s
al
n
etwo
r
k
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
:
Srin
iv
as Dib
b
u
r
B
y
r
ap
p
a
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atio
n
Scien
ce
an
d
E
n
g
in
ee
r
i
n
g
,
Nitte
Me
en
ak
s
h
i I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
6
4
2
9
,
NI
T
T
E
Me
en
a
k
s
h
i Co
lleg
e
R
d
,
B
SF
C
am
p
u
s
,
Yela
h
an
k
a,
B
en
g
alu
r
u
,
Go
v
in
d
ap
u
r
a,
Kar
n
atak
a
5
6
0
0
6
4
,
I
n
d
ia
E
m
ail:
s
r
in
iv
as.d
b
@
n
m
it.a
c.
in
1.
I
NT
RO
D
UCT
I
O
N
Ob
ject
d
etec
tio
n
is
o
n
e
o
f
th
e
m
o
s
t
im
p
o
r
tan
t
an
d
o
n
e
o
f
th
e
m
o
s
t
p
o
p
u
lar
task
s
o
f
c
o
m
p
u
ter
v
is
io
n
an
d
a
f
u
n
d
am
e
n
tal
f
ield
o
f
ar
tific
ial
in
tellig
en
ce
.
I
t
p
r
o
v
id
es
a
b
asis
f
o
r
a
v
ar
iety
o
f
p
r
ac
tical
ap
p
licatio
n
s
,
s
u
ch
as
au
to
n
o
m
o
u
s
n
av
ig
ati
o
n
,
in
d
u
s
tr
ial
q
u
ality
co
n
tr
o
l,
r
o
b
o
tics
,
an
d
m
e
d
ical
d
iag
n
o
s
tics
.
Ov
er
th
e
last
f
ew
y
ea
r
s
,
d
ee
p
lear
n
i
n
g
,
p
a
r
ticu
lar
ly
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NNs)
h
av
e
h
ad
m
u
ltip
le
p
r
o
f
o
u
n
d
im
p
licatio
n
s
o
n
o
b
ject
d
etec
tio
n
,
in
clu
d
in
g
au
t
o
m
ated
le
ar
n
in
g
o
f
f
ea
tu
r
es
an
d
p
r
o
v
id
in
g
v
er
y
r
o
b
u
s
t
class
if
icatio
n
an
d
lo
ca
lizatio
n
p
er
f
o
r
m
a
n
ce
.
C
o
n
v
en
tio
n
a
l
co
m
p
u
te
r
v
is
io
n
m
eth
o
d
s
wer
e
in
ten
s
iv
e
o
n
s
tatis
t
ical
m
o
d
els
lik
ely
to
f
ail
in
th
e
p
r
esen
ce
o
f
v
ar
ia
b
ilit
y
in
th
e
ap
p
ea
r
an
ce
o
f
o
b
jects,
lig
h
tin
g
,
an
d
co
m
p
lex
ity
o
f
th
e
b
ac
k
g
r
o
u
n
d
.
W
ith
th
e
em
er
g
en
ce
o
f
d
ee
p
lear
n
in
g
,
C
NNs
[
1
]
,
[
2
]
h
a
v
e
s
o
f
ar
tr
an
s
f
o
r
m
ed
o
b
ject
d
etec
tio
n
b
y
a
u
to
m
atic
ally
ac
q
u
ir
in
g
h
ier
a
r
ch
ical
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
u
s
in
g
r
aw
im
ag
e
in
p
u
ts
.
T
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
6
3
-
872
864
s
tate
-
of
-
th
e
-
ar
t
o
b
ject
d
etec
tio
n
m
o
d
els
lik
e
th
e
Fas
t
er
R
-
C
NN
m
o
d
el
[
3
]
,
y
o
u
o
n
ly
lo
o
k
o
n
ce
(
YOL
O
)
[
4
]
,
[
5
]
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
in
[
6
]
is
ca
r
r
ied
o
u
t
u
s
in
g
less
m
em
o
r
y
an
d
less
p
r
o
ce
s
s
in
g
p
o
wer
an
d
p
r
o
ce
s
s
in
g
m
ass
d
ata
u
s
in
g
Pas
ca
l
VOC
an
d
Mic
r
o
s
o
f
t
C
OC
O
d
atasets
.
Qu
ick
d
etec
tio
n
alg
o
r
ith
m
c
an
d
is
cr
im
in
ate
th
e
o
b
ject
in
0
.
8
2
a
n
d
ab
o
v
e
s
ec
o
n
d
s
.
Ho
wev
er
,
th
er
e
is
a
cr
itical
g
ap
in
ex
is
tin
g
liter
atu
r
e
th
at
h
as
n
o
t
b
ee
n
f
u
lly
u
s
ed
,
i.e
.
,
th
e
u
s
e
o
f
in
tr
in
s
ic
o
b
ject
attr
ib
u
tes,
e.
g
.
co
lo
r
in
f
o
r
m
atio
n
in
th
ese
d
e
tectio
n
p
ip
elin
es.
T
h
e
cu
r
r
e
n
t
m
o
d
els
ar
e
m
ain
ly
b
ased
o
n
f
ea
tu
r
es
o
f
s
p
ac
e
an
d
s
h
ap
e
with
o
u
t
m
u
ch
f
o
cu
s
o
n
co
l
o
r
f
ea
tu
r
es
wh
ich
ar
e
e
s
s
en
tial
in
v
ar
io
u
s
p
r
ac
tical
s
itu
atio
n
s
.
T
o
d
em
o
n
s
tr
ate,
o
n
e
o
f
th
e
m
ai
n
d
is
tin
ctio
n
s
in
th
e
u
s
e
o
f
co
lo
r
is
in
tr
af
f
ic
s
ig
n
al
d
etec
tio
n
,
s
o
r
tin
g
o
f
f
r
u
its
in
a
g
r
icu
ltu
r
e,
o
r
th
e
s
ca
n
n
i
n
g
o
f
d
ef
ec
ts
o
n
i
n
d
u
s
tr
ial
p
r
o
d
u
cts.
T
h
e
f
ail
u
r
e
to
p
ay
atten
tio
n
to
co
lo
r
m
a
y
r
esu
lt
in
m
is
class
if
icatio
n
s
in
s
itu
atio
n
s
wh
er
e
c
o
lo
r
co
d
ed
o
b
jects
o
r
ap
p
ea
r
an
ce
s
im
ilar
o
b
jects
ar
e
s
ep
ar
ate
d
p
r
im
ar
ily
b
y
co
lo
r
.
Mo
r
e
o
v
er
,
th
e
s
ec
o
n
d
s
ig
n
if
ica
n
t
d
r
aw
b
ac
k
o
f
th
e
s
tate
-
of
-
th
e
-
ar
t
is
th
e
in
ab
ilit
y
to
in
t
er
p
r
et
d
ee
p
lear
n
in
g
m
o
d
els.
E
v
en
th
o
u
g
h
C
NN
-
b
ased
d
etec
to
r
s
ar
e
h
ig
h
ly
p
er
f
o
r
m
in
g
,
th
e
y
ar
e
b
lack
b
o
x
es,
wh
ich
m
ea
n
s
th
at
en
d
-
u
s
er
s
in
s
af
ety
-
cr
itical
s
y
s
tem
s
(
au
to
n
o
m
o
u
s
v
eh
icles
o
r
m
ed
ical
d
ev
ices)
ca
n
n
o
t
r
ely
o
n
m
o
d
el
o
u
tp
u
t
u
n
less
th
er
e
is
an
in
ter
p
r
etab
ilit
y
m
ec
h
an
is
m
.
T
o
o
v
er
co
m
e
th
em
,
we
in
tr
o
d
u
ce
a
n
ew
h
y
b
r
i
d
f
r
a
m
ewo
r
k
th
a
t
will
in
v
o
lv
e
C
NN
-
b
ased
o
b
j
ec
t
d
etec
tio
n
(
Fas
t
R
-
C
NN
with
R
e
s
Net5
0
-
f
ea
tu
r
e
p
y
r
am
id
n
etwo
r
k
(
FPN
)
b
ac
k
b
o
n
e)
,
co
lo
r
an
al
y
s
is
with
th
e
k
-
m
ea
n
s
clu
s
ter
in
g
alg
o
r
ith
m
a
n
d
e
x
p
lain
ab
le
ar
tific
ial
in
tellig
en
ce
(
AI
)
m
eth
o
d
s
th
r
o
u
g
h
g
r
a
d
ien
t
-
weig
h
ted
class
ac
tiv
atio
n
m
ap
p
i
n
g
(
Gr
a
d
-
C
AM
)
an
d
h
is
to
g
r
am
o
f
o
r
ien
te
d
g
r
ad
ien
ts
(
HOG
)
[
7
]
.
T
h
e
r
em
ai
n
d
er
o
f
th
is
ar
ticl
e
is
s
tr
u
ctu
r
ed
as
:
Sectio
n
2
p
r
o
v
id
es
a
b
r
ief
o
v
er
v
iew
o
f
r
elate
d
wo
r
k
.
Sectio
n
3
o
u
tlin
es
t
h
e
m
eth
o
d
.
I
n
s
ec
tio
n
4
r
esu
lts
a
r
e
d
is
cu
s
s
ed
,
an
d
f
i
n
ally
s
ec
tio
n
6
co
n
clu
d
es
th
e
ar
ticle.
2.
RE
L
AT
E
D
WO
RK
R
ec
en
t
ad
v
an
ce
s
in
o
b
ject
d
et
ec
tio
n
tech
n
o
lo
g
ies
h
av
e
b
ee
n
s
ig
n
if
ican
tly
im
p
r
o
v
e
d
in
v
ar
io
u
s
f
ield
s
an
d
ar
ch
itectu
r
e
.
Stu
d
ies
[
8
]
h
av
e
s
h
o
wn
th
at
in
teg
r
atin
g
R
e
s
Net
-
5
0
with
d
etec
tio
n
with
tr
an
s
f
o
r
m
er
s
(
DE
T
R
)
(
a
tr
an
s
f
o
r
m
e
r
-
b
ased
m
o
d
el)
ac
h
iev
ed
ap
p
r
o
x
im
atel
y
9
0
%
b
etter
r
esu
lts
th
an
tr
ad
itio
n
al
C
NNs,
wh
ile
tr
an
s
f
o
r
m
er
s
co
m
b
in
ed
with
C
NNs
ac
h
iev
ed
an
av
er
ag
e
p
r
ec
is
io
n
(
AP)
o
f
2
0
.
6
f
o
r
s
m
al
l
o
b
jects
[
9
]
.
Fas
ter
R
-
C
NN
h
as
b
ee
n
ef
f
ec
tiv
e
in
n
ich
e
ap
p
licatio
n
s
s
u
ch
as
m
an
g
a
ch
ar
ac
ter
an
d
tex
t
d
etec
tio
n
,
ac
h
iev
in
g
m
etr
ics
as
h
ig
h
as
0
.
8
1
6
a
n
d
0
.
8
9
8
,
r
esp
ec
tiv
ely
[
1
0
]
.
Sate
llit
e
im
ag
er
y
an
aly
s
is
[
1
1
]
u
s
i
n
g
a
cu
s
to
m
C
NN
ap
p
r
o
ac
h
ac
h
iev
ed
an
ac
c
u
r
ac
y
o
f
9
4
.
6
5
%
to
d
etec
t
v
eh
icles,
b
u
ild
in
g
s
,
an
d
tr
ee
s
.
T
r
af
f
ic
s
ig
n
d
etec
tio
n
with
a
two
-
s
tag
e
F
aster
R
-
C
NN
m
o
d
el
ac
h
ie
v
ed
a
d
etec
tio
n
p
r
ec
is
io
n
o
f
8
8
.
9
9
%
[
1
2
]
,
wh
ile
s
p
ec
ialized
ap
p
licatio
n
s
s
u
ch
as
u
n
m
an
n
ed
ae
r
ial
v
e
h
icle
(
UAV)
d
etec
tio
n
s
h
o
wed
b
etter
r
esu
lts
with
L
SL
Net
[
1
3
]
.
Dete
ctio
n
o
f
h
o
u
s
eh
o
ld
o
b
jec
ts
u
s
in
g
R
esNet5
0
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM
)
is
p
r
esen
ted
in
[
1
4
]
.
C
NN
i
s
u
s
ed
to
d
etec
t
o
b
ject
s
,
R
e
s
Net5
0
is
u
s
ed
to
cla
s
s
if
y
th
e
im
ag
es
in
to
o
b
jects,
an
d
th
en
th
e
SVM
is
u
s
ed
to
tr
ain
o
b
jects
an
d
s
to
r
ed
in
th
e
o
b
ject
d
atab
ase.
A
two
-
p
h
ase
ap
p
r
o
ac
h
c
o
m
b
in
i
n
g
c
o
r
r
elativ
e
f
ilter
tr
ac
k
in
g
a
n
d
C
NN
s
h
o
wed
s
ig
n
if
ican
t
im
p
r
o
v
e
m
en
ts
in
s
t
u
d
y
[
1
5
]
f
o
r
v
a
r
io
u
s
s
ce
n
ar
i
o
s
.
A
F
aster
R
-
C
N
N
m
o
d
el
with
a
r
eg
i
o
n
al
p
r
o
p
o
s
al
n
etwo
r
k
[
1
6
]
s
h
o
we
d
s
tr
o
n
g
r
esu
lts
o
n
b
en
c
h
m
ar
k
s
,
s
u
ch
as
MS
C
O
C
O
an
d
Pas
ca
l V
O
C
[
1
7
]
.
T
h
e
o
b
ject
id
en
tific
atio
n
alg
o
r
ith
m
b
ased
o
n
th
e
c
h
ar
ac
ter
i
s
tic
co
lo
r
with
th
e
YUV
co
lo
r
s
p
ac
e
is
p
r
esen
ted
in
s
tu
d
y
[
1
8
]
.
A
s
tu
d
y
-
ex
p
an
s
io
n
alg
o
r
ith
m
to
o
b
tain
th
e
s
p
atial
d
is
tr
ib
u
ti
o
n
o
f
th
e
c
o
m
p
atib
le
o
b
ject
co
lo
r
c
h
ar
ac
ter
is
tic.
T
h
e
im
ag
es
co
n
tain
in
g
C
C
T
V
f
ac
es
ar
e
s
eg
m
en
ted
in
s
tu
d
y
[
1
9
]
,
ex
p
er
im
en
ts
wer
e
ca
r
r
ied
o
u
t
o
n
a
s
in
g
le
f
ac
e
im
ag
e
in
m
ask
d
etec
tio
n
,
r
esu
ltin
g
in
a
n
ac
cu
r
ac
y
o
f
9
7
.
3
3
%.
Similar
ly
,
th
er
m
al
im
ag
e
d
etec
tio
n
r
ep
o
r
ted
a
m
ea
n
AP
(
m
AP)
o
f
2
6
.
5
%
[
2
0
]
.
I
n
o
th
er
a
p
p
licatio
n
s
,
an
m
ask
ed
im
a
g
e
m
o
d
elin
g
(
MI
M
)
-
p
r
etr
ai
n
ed
v
an
illa
v
is
io
n
tr
an
s
f
o
r
m
er
(
ViT
)
[
2
1
]
en
c
o
d
er
ac
h
ie
v
ed
an
AP o
f
5
1
.
5
,
a
n
d
v
is
io
n
tr
an
s
f
o
r
m
er
s
d
em
o
n
s
tr
ated
p
r
o
m
is
in
g
o
p
e
n
-
v
o
ca
b
u
la
r
y
d
etec
tio
n
,
ac
h
iev
i
n
g
a
n
AP
o
f
3
1
.
2
%
o
n
th
e
L
VI
S
v
1
.
0
d
ataset
[
2
2
]
.
Sp
ec
ialized
u
n
d
er
wate
r
d
etec
tio
n
m
o
d
els
f
o
r
lo
w
-
p
o
wer
s
y
s
tem
s
,
s
u
ch
a
s
th
e
R
asp
b
er
r
y
Pi,
d
em
o
n
s
tr
ated
b
y
u
s
in
g
m
u
lt
i
-
s
ca
le
f
ea
tu
r
e
lear
n
in
g
to
i
m
p
r
o
v
e
co
m
p
u
tatio
n
al
ef
f
ici
en
cy
[
2
3
]
.
I
m
ag
es
ca
p
tu
r
ed
f
r
o
m
r
o
a
d
s
o
r
h
illy
r
eg
io
n
s
o
f
ten
s
u
f
f
er
f
r
o
m
p
o
o
r
v
is
ib
ilit
y
d
u
e
to
h
az
e
[
2
4
]
.
R
em
o
v
in
g
th
is
h
az
e
ca
n
s
ig
n
if
ican
tly
en
h
a
n
ce
co
l
o
r
r
ec
o
g
n
itio
n
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
ad
d
r
ess
es
th
is
b
y
in
teg
r
atin
g
two
k
e
y
tech
n
iq
u
es:
f
ir
s
t,
it
em
p
lo
y
s
th
e
d
ar
k
c
h
an
n
el
p
r
io
r
m
eth
o
d
to
elim
in
ate
h
az
e;
n
e
x
t,
it
u
tili
ze
s
C
NN
f
o
r
f
ea
tu
r
e
lear
n
in
g
.
On
ce
t
h
e
f
ea
tu
r
es
a
r
e
ex
tr
ac
ted
,
an
ef
f
ec
tiv
e
clas
s
if
icatio
n
m
eth
o
d
,
s
u
ch
as
th
e
SVM,
is
u
s
ed
to
p
er
f
o
r
m
th
e
f
in
al
class
if
icati
o
n
.
Py
th
o
n
-
O
p
en
C
V
[
2
5
]
h
as
p
r
o
v
en
ef
f
ec
tiv
e
in
R
GB
co
lo
r
r
ec
o
g
n
itio
n
f
o
r
co
lo
r
d
etec
tio
n
in
co
m
p
u
ter
v
is
io
n
.
Gu
p
ta
et
a
l.
[
2
6
]
h
ig
h
lig
h
t
h
o
w
th
e
m
o
d
el
u
tili
ze
s
d
ee
p
lear
n
in
g
an
d
n
atu
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
to
g
e
n
er
ate
co
n
cis
e
d
escr
ip
tio
n
s
o
f
in
p
u
t
im
ag
es.
C
o
m
p
r
eh
en
s
iv
e
r
ev
iew
o
f
r
ec
en
t
ad
v
an
ce
s
,
m
eth
o
d
s
,
c
h
allen
g
es,
an
d
f
u
tu
r
e
d
ir
ec
tio
n
s
in
o
b
ject
d
etec
tio
n
an
d
c
o
lo
r
id
en
tific
atio
n
,
f
o
cu
s
in
g
o
n
d
ee
p
lear
n
in
g
an
d
p
r
ac
tical
ap
p
lic
atio
n
s
[
2
7
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
ted
d
ee
p
lea
r
n
in
g
a
p
p
r
o
a
ch
fo
r
r
ea
l
-
time
o
b
ject
d
etec
tio
n
a
n
d
…
(
S
r
in
iva
s
Dib
b
u
r
B
yra
p
p
a
)
865
3.
M
E
T
H
O
D
T
h
e
f
r
a
m
ewo
r
k
as
illu
s
tr
ated
in
Fig
u
r
e
1
in
tr
o
d
u
ce
s
a
r
ea
l
t
im
e
o
b
ject
d
etec
tio
n
a
n
d
c
o
lo
r
an
aly
s
is
o
b
ject
o
f
r
aw
im
ag
es
s
y
s
tem
.
I
t
is
b
ased
o
n
a
Fas
ter
R
-
C
N
N
f
r
am
ewo
r
k
with
a
R
esNet
-
5
0
b
ac
k
b
o
n
e
a
n
d
a
FP
N
th
at
is
u
s
ed
to
d
etec
t
o
b
jects
with
h
ig
h
lev
els
o
f
r
o
b
u
s
tn
ess
an
d
at
m
u
ltip
le
s
ca
les.
T
h
e
in
itial
s
tag
e
in
v
o
lv
es
p
r
e
-
p
r
o
ce
s
s
in
g
an
d
a
u
g
m
en
tatio
n
o
f
th
e
im
ag
es
a
n
d
th
en
t
h
r
o
u
g
h
d
etec
tio
n
n
etwo
r
k
.
A
n
ex
p
e
r
t
co
lo
r
r
ec
o
g
n
itio
n
u
n
it
o
p
e
r
ates
o
n
th
e
id
en
tifie
d
ar
ea
s
th
r
o
u
g
h
an
a
n
aly
s
is
o
f
clu
s
ter
in
g
to
f
in
d
th
e
p
r
ev
ailin
g
co
l
o
r
s
an
d
th
eir
p
e
r
ce
n
tag
es.
T
h
e
s
y
s
tem
ca
n
v
is
u
alize
th
e
r
esu
lts
with
b
o
u
n
d
ar
ies,
co
lo
r
d
etails,
an
d
h
ea
tm
ap
s
o
f
Gr
ad
-
C
AM
,
wh
ich
h
av
e
c
o
n
ti
n
u
o
u
s
d
y
n
am
ic
u
p
d
ates,
to
o
f
f
er
b
etter
u
s
er
e
x
p
er
ien
ce
.
Fig
u
r
e
1
.
Pro
p
o
s
ed
f
r
am
ewo
r
k
3
.
1
.
F
a
s
t
er
R
-
CNN
Fas
ter
R
-
C
NN
(
A
lg
o
r
ith
m
1
)
s
er
v
es
as
th
e
p
r
im
ar
y
d
etec
tio
n
m
o
d
u
le
d
u
e
to
its
s
tr
o
n
g
ac
cu
r
ac
y
in
r
eg
io
n
-
b
ased
o
b
ject
lo
ca
lizati
o
n
.
T
h
e
s
y
s
tem
u
s
es
a
R
esNet
-
5
0
b
ac
k
b
o
n
e
with
an
FP
N
to
ex
tr
ac
t
r
ich
m
u
lti
-
s
ca
le
f
ea
tu
r
es
s
u
itab
le
f
o
r
d
etec
tin
g
o
b
jects
o
f
d
i
f
f
er
en
t
s
izes.
A
r
eg
io
n
p
r
o
p
o
s
al
n
etwo
r
k
(
R
PN
)
th
en
g
en
er
ates
ca
n
d
id
ate
o
b
ject
r
eg
io
n
s
b
y
e
v
alu
atin
g
an
ch
o
r
b
o
x
es
o
f
v
a
r
io
u
s
s
ca
les
an
d
asp
ec
t
r
atio
s
an
d
p
r
ed
ictin
g
o
b
ject
n
ess
s
co
r
es a
n
d
r
ef
in
e
d
co
o
r
d
in
ates.
T
h
e
to
p
p
r
o
p
o
s
als ar
e
p
r
o
ce
s
s
ed
u
s
in
g
r
eg
io
n
o
f
i
n
ter
est
(
R
OI
)
alig
n
,
wh
ic
h
p
r
o
d
u
ce
s
s
p
atially
co
n
s
is
ten
t,
f
ix
e
d
-
s
ize
f
ea
tu
r
e
m
a
p
s
,
en
ab
li
n
g
p
r
ec
is
e
class
if
icatio
n
an
d
b
o
u
n
d
in
g
b
o
x
r
ef
in
em
e
n
t.
3
.
2
.
Co
l
o
r
re
co
g
nitio
n inte
g
ra
t
io
n
T
h
e
in
teg
r
atio
n
o
f
co
l
o
r
r
ec
o
g
n
itio
n
in
to
th
e
Fas
ter
R
-
C
NN
f
r
am
ewo
r
k
in
v
o
lv
es
an
aly
zin
g
th
e
R
OI
s
d
etec
ted
b
y
th
e
o
b
ject
d
etec
ti
o
n
m
o
d
el
an
d
p
e
r
f
o
r
m
in
g
d
eta
iled
co
lo
r
an
aly
s
is
f
o
r
ea
c
h
id
en
tifie
d
o
b
ject.
T
h
e
p
r
o
ce
s
s
(
A
lg
o
r
ith
m
2
)
s
tar
ts
b
y
ex
tr
ac
tin
g
a
cir
cu
lar
r
e
g
io
n
f
r
o
m
th
e
ce
n
ter
o
f
ea
ch
d
etec
te
d
o
b
ject'
s
b
o
u
n
d
in
g
b
o
x
,
e
n
s
u
r
in
g
th
e
an
aly
s
is
f
o
c
u
s
es
o
n
th
e
m
o
s
t
r
e
p
r
esen
tativ
e
p
ar
t
o
f
th
e
o
b
ject
wh
ile
m
in
i
m
izin
g
b
ac
k
g
r
o
u
n
d
in
ter
f
er
en
ce
.
T
h
is
e
x
tr
ac
ted
r
eg
io
n
is
t
h
en
s
u
b
jecte
d
to
k
-
m
ea
n
s
clu
s
ter
in
g
,
an
ef
f
icien
t
an
d
wid
ely
u
s
ed
alg
o
r
ith
m
f
o
r
id
e
n
tify
in
g
d
o
m
in
an
t
co
lo
r
s
with
in
im
a
g
e
p
atc
h
es
b
y
g
r
o
u
p
i
n
g
s
im
ilar
p
ix
el
co
lo
r
s
in
to
clu
s
ter
s
b
ased
o
n
th
eir
p
r
o
x
im
it
y
to
clu
s
ter
ce
n
ter
s
.
T
h
e
n
u
m
b
er
o
f
clu
s
ter
s
(
k
)
ca
n
b
e
p
r
ed
ef
in
ed
,
an
d
th
e
ce
n
ter
o
f
ea
ch
clu
s
ter
r
ep
r
esen
ts
a
d
o
m
in
an
t
co
l
o
r
f
o
u
n
d
in
t
h
e
o
b
ject
r
eg
io
n
.
T
h
is
m
eth
o
d
e
n
ab
les
r
o
b
u
s
t
an
d
au
to
m
ated
c
o
lo
r
id
e
n
tific
atio
n
f
o
r
ea
c
h
d
etec
ted
o
b
jec
t,
s
u
p
p
o
r
tin
g
task
s
th
at
r
e
q
u
ir
e
p
r
ec
is
e
co
l
o
r
in
f
o
r
m
atio
n
with
in
o
b
ject
d
ete
ctio
n
p
ip
elin
es.
3
.
3
.
Vis
ua
liza
t
io
n
T
h
e
f
r
a
m
ewo
r
k
in
teg
r
ates
ex
p
lain
ab
le
AI
m
eth
o
d
s
HOG
a
n
d
Gr
a
d
-
C
AM
to
im
p
r
o
v
e
tr
a
n
s
p
ar
en
cy
an
d
in
ter
p
r
eta
b
ilit
y
in
o
b
ject
d
etec
tio
n
an
d
co
l
o
r
an
aly
s
is
.
HOG
p
r
o
v
id
es
a
tr
ad
itio
n
al,
h
an
d
-
c
r
af
ted
v
is
u
al
ex
p
lan
atio
n
b
y
m
a
p
p
in
g
ed
g
e
o
r
ien
tatio
n
s
an
d
g
r
ad
ie
n
ts
,
em
p
h
asizin
g
s
h
ap
es a
n
d
c
o
n
to
u
r
s
th
at
s
u
p
p
o
r
t
o
b
jec
t
r
ec
o
g
n
itio
n
.
I
t
allo
ws
co
m
p
ar
is
o
n
b
etwe
en
co
n
v
en
tio
n
al
f
ea
tu
r
e
ex
t
r
ac
tio
n
a
n
d
n
e
u
r
al
r
ep
r
esen
tatio
n
s
.
Gr
ad
-
C
AM
co
m
p
lem
en
ts
th
i
s
b
y
v
is
u
alizin
g
m
o
d
el
atten
tio
n
u
s
in
g
g
r
ad
ien
t
-
b
ased
h
e
atm
ap
s
f
r
o
m
d
ee
p
n
etwo
r
k
s
lik
e
R
esNet
-
5
0
with
FP
N.
T
h
ese
m
ap
s
h
ig
h
lig
h
t
t
h
e
im
ag
e
r
e
g
io
n
s
m
o
s
t
in
f
lu
e
n
tial
in
th
e
m
o
d
el’
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
6
3
-
872
866
class
if
icatio
n
an
d
lo
ca
lizatio
n
d
ec
is
io
n
s
,
r
ev
ea
lin
g
h
o
w
t
h
e
n
etwo
r
k
f
o
cu
s
es
o
n
s
p
ec
i
f
ic
s
p
atial
f
ea
tu
r
es
d
u
r
in
g
d
etec
tio
n
.
(
,
)
=
√
(
(
−
)
)
=
1
(
1
)
T
h
e
p
i
x
e
l
s
wi
t
h
i
n
t
h
e
i
d
e
n
t
i
f
ie
d
r
e
g
i
o
n
a
r
e
u
s
e
d
a
s
i
n
p
u
ts
t
o
t
h
e
k
-
m
e
a
n
s
a
l
g
o
r
it
h
m
(
A
lg
o
r
i
t
h
m
3
)
wh
ich
clu
s
ter
s
th
em
i
n
to
a
s
p
e
cif
ied
n
u
m
b
er
o
f
clu
s
ter
s
o
f
c
o
lo
r
s
with
(
1
)
(
E
u
clid
ea
n
d
is
tan
ce
)
as
th
e
d
is
tan
ce
m
ea
s
u
r
e.
T
h
e
m
eth
o
d
allo
ws b
r
ea
k
in
g
d
o
wn
n
o
t
o
n
ly
o
n
e
d
o
m
in
an
t
co
l
o
r
b
u
t
also
it
g
iv
es a
n
allo
ca
tio
n
o
f
th
e
m
ajo
r
co
lo
r
s
th
at
f
o
r
m
i
n
th
e
o
b
ject.
T
h
e
m
o
s
t r
ep
r
esen
tativ
e
co
lo
r
s
o
f
a
n
o
b
ject
ar
e
th
en
th
e
ce
n
tr
o
id
s
o
f
ea
c
h
clu
s
ter
.
T
h
e
ce
n
ter
s
o
f
th
ese
clu
s
ter
s
ar
e
th
e
m
o
s
t
d
escr
ip
tiv
e
co
lo
r
s
o
f
th
e
o
b
ject.
I
t
h
as
a
co
lo
r
-
n
am
in
g
co
m
p
o
n
en
t
wh
ich
m
ap
s
th
e
R
GB
v
alu
es
o
f
clu
s
ter
ce
n
tr
o
id
to
th
e
clo
s
est
n
am
ed
co
l
o
r
s
in
a
p
r
e
-
ex
is
tin
g
c
o
lo
r
s
p
ac
e,
e.
g
.
C
SS
3
co
lo
r
n
am
es
[
2
8
]
,
[
2
9
]
.
T
h
is
ass
i
s
ts
i
n
co
n
v
er
tin
g
co
lo
r
s
o
f
o
b
je
cts
in
to
a
h
u
m
an
u
n
d
er
s
tan
d
a
b
le
f
o
r
m
at,
wh
ich
ca
n
b
e
u
s
ed
in
ca
s
es
wh
er
e
th
e
ap
p
licatio
n
r
e
q
u
ir
es
th
e
u
s
e
o
f
n
atu
r
al
lan
g
u
a
g
e
co
lo
r
s
.
3
.
4
.
I
m
ple
m
ent
a
t
io
n
Fig
u
r
e
2
s
h
o
ws
th
e
im
p
lem
en
tatio
n
o
f
th
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
.
T
h
e
Py
T
o
r
ch
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
is
u
s
ed
in
th
e
f
r
am
ewo
r
k
.
T
h
e
m
eth
o
d
s
u
s
ed
co
n
s
is
ted
o
f
r
an
d
o
m
h
o
r
izo
n
tal
f
lip
p
in
g
at
5
0
p
er
ce
n
t
f
r
e
q
u
en
c
y
,
r
an
d
o
m
ap
p
licatio
n
o
f
th
e
r
o
tatio
n
s
o
f
±
1
0
d
eg
r
ee
s
an
d
b
r
ig
h
tn
ess
an
d
co
n
tr
ast
m
an
ip
u
latio
n
.
T
h
e
p
h
o
to
s
wer
e
r
esized
to
8
0
0
p
ix
els
o
n
t
h
e
s
h
o
r
ter
d
im
en
s
io
n
k
ee
p
in
g
th
e
o
r
ig
in
al
asp
ec
t
r
atio
to
m
ak
e
s
u
r
e
th
at
th
e
in
p
u
t
is
as
h
o
m
o
g
en
eo
u
s
as
p
o
s
s
ib
le
an
d
th
at
th
e
s
ig
n
if
ica
n
t
v
is
u
al
in
f
o
r
m
atio
n
is
n
o
t
lo
s
t.
T
h
e
s
to
ch
asti
c
g
r
ad
ien
t
d
escen
t
(
SGD)
alg
o
r
ith
m
was
u
s
ed
in
th
e
tr
ain
in
g
s
tag
e
wh
er
e
a
m
o
m
en
tu
m
f
ac
to
r
o
f
0
.
9
an
d
a
weig
h
t
d
e
ca
y
r
ate
o
f
0
.
0
0
0
5
wer
e
ap
p
li
ed
.
I
n
th
e
a
n
aly
s
is
o
f
co
l
o
r
,
k
-
m
ea
n
s
clu
s
ter
in
g
alg
o
r
ith
m
is
u
s
ed
to
d
eter
m
in
e
th
e
p
r
ev
ailin
g
c
o
lo
r
p
atter
n
s
in
th
e
d
etec
ted
o
b
je
ct
r
eg
io
n
s
.
I
t
was
ex
p
er
im
en
tally
test
ed
th
at
th
e
f
iv
e
clu
s
ter
ap
p
r
o
ac
h
(
K=
5
)
o
f
f
er
ed
a
g
o
o
d
b
alan
ce
b
et
wee
n
co
m
p
u
tatio
n
ef
f
icien
cy
an
d
ca
p
ac
ity
to
d
escr
ib
e
ad
eq
u
ate
co
lo
r
v
ar
iab
ilit
y
.
T
o
m
ak
e
th
e
s
am
p
lin
g
co
n
s
tan
t
an
d
r
ep
r
esen
tativ
e,
a
cir
cu
lar
r
eg
i
o
n
is
s
am
p
led
o
u
t
o
f
an
y
g
iv
e
n
o
b
s
er
v
ed
o
b
ject,
an
d
th
e
r
a
d
iu
s
is
d
ef
in
ed
as
a
q
u
ar
ter
o
f
th
e
less
er
s
id
e
o
f
th
e
b
o
u
n
d
in
g
b
o
x
o
f
th
e
o
b
ject.
T
h
e
s
am
p
lin
g
s
tr
ateg
y
is
u
n
if
o
r
m
,
wh
ich
m
ak
es
th
e
m
eth
o
d
c
o
n
s
is
ten
t
in
an
al
y
zin
g
co
l
o
r
in
f
o
r
m
atio
n
o
f
o
b
je
cts
o
f
d
if
f
er
en
t
s
h
ap
es
an
d
s
iz
es.
All
ex
p
er
im
e
n
ts
wer
e
co
n
d
u
cte
d
u
s
in
g
Py
th
o
n
3
.
1
0
with
th
e
Py
T
o
r
ch
2
.
x
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
o
n
a
s
y
s
tem
eq
u
ip
p
e
d
with
an
1
2
th
g
en
in
tel(
r
)
co
r
e
(
tm
)
i5
-
1
2
5
0
0
,
3
0
0
0
Mh
z
6
co
r
es,
1
2
th
r
ea
d
s
,
3
2
GB
R
AM
,
an
d
Ub
u
n
tu
2
0
.
0
4
.
Fi
gur
e 2. Fl
o
w
c
har
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
ted
d
ee
p
lea
r
n
in
g
a
p
p
r
o
a
ch
fo
r
r
ea
l
-
time
o
b
ject
d
etec
tio
n
a
n
d
…
(
S
r
in
iva
s
Dib
b
u
r
B
yra
p
p
a
)
867
Alg
o
r
ith
m
1
.
Ob
ject
d
etec
tio
n
u
s
in
g
Fas
ter
R
-
C
NN
1: Feature Extraction
2: Extract features using ResNet50 backbone
3: Create multi
-
scale feature maps using FPN
4: Region
proposal network
(RPN)
5: Generate anchor boxes
6: Predict object proposals
7: Filter and refine proposals
8: Region of Interest (RoI) Processing
9: Align multi
-
scale feature maps with region proposals to get RoI features
10: Final Prediction
11: for each RoI do
12:
Classify the RoI
13:
Refine bounding box for the RoI
14: end for
15: Return Results
Alg
o
r
ith
m
2
.
C
o
lo
r
an
aly
s
is
p
r
o
ce
d
u
r
e
Require: image, bounding box
Ensure: dominant color, color name, cluster centers, counts
1
:
Extract circular region from the bounding box.
2
:
Get dominant color:
3
:
Reshape images to 2D and remove black pixels.
4
:
if pixels are not empty then
5
:
Perform k
-
means clustering.
6
:
Calculate mean color.
7
:
Find closest centroid to the mean color.
8
:
Store dominant color, cluster centers, and counts.
9
:
end if
10
:
Return Results:
11
:
if dominant color is found then
12
:
Get color name from RGB value.
13
:
return dominant color, color name, cluster centers, counts.
14
:
else
15
:
return None, None, None, None.
16
:
end if
Alg
o
r
ith
m
3
.
k
-
m
ea
n
s
clu
s
ter
in
g
p
r
o
ce
d
u
r
e
Require: dataset D, number of clusters k, maximum iterations max iterations
Ensure: centroids, assignments
1: Initialize k centroids randomly
2: for iteration = 1 to max iterations do
3:
Assign each point in D to the nearest centroid
4:
Update centroids as the mean of the assigned points
5:
if centroids do not change then
6:
break
7:
end if
8: end for
9: return Final centroids, Cluster assignments
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
The
p
er
f
o
r
m
an
ce
o
f
o
u
r
p
r
o
p
o
s
ed
f
r
a
m
ewo
r
k
is
tab
u
late
d
in
T
a
b
le
1
.
E
x
p
er
im
en
t
1
0
(
lear
n
i
n
g
r
ate
=
0
.
0
0
5
an
d
ep
o
ch
s
=
5
0
)
s
h
o
ws
th
e
h
i
g
h
est
m
ea
n
p
r
ec
is
io
n
o
f
0
.
8
1
%,
m
AP 0
.
8
1
%,
an
d
s
o
lid
F1
s
co
r
e
o
f
0
.
6
2
%,
in
d
icatin
g
th
at
m
o
r
e
ep
o
ch
s
(
5
0
)
at
a
m
o
d
er
ate
le
ar
n
in
g
r
ate
y
ield
th
e
b
est
o
v
er
all
p
er
f
o
r
m
a
n
ce
.
E
x
p
er
im
en
ts
1
an
d
6
also
s
h
o
w
h
ig
h
p
r
ec
is
io
n
an
d
m
AP
(
0
.
7
7
%
an
d
0
.
7
4
%),
s
u
g
g
esti
n
g
t
h
at
3
0
ep
o
c
h
s
with
0
.
0
0
5
o
r
0
.
0
0
1
lear
n
in
g
r
ate
al
s
o
p
er
f
o
r
m
well.
Similar
ly
,
e
x
p
er
im
en
t
7
(
lear
n
in
g
r
ate
=
0
.
0
0
0
1
,
e
p
o
ch
s
=
2
0
)
h
as
th
e
lo
west
p
e
r
f
o
r
m
an
ce
i
n
all
m
etr
ics,
esp
ec
ially
with
an
e
x
tr
em
ely
lo
w
F1
s
co
r
e
(
0
.
0
9
%)
an
d
m
AP
(
0
.
1
5
%),
in
d
icatin
g
th
at
to
o
lo
w
a
lear
n
in
g
r
ate
with
f
ewe
r
ep
o
ch
s
lead
s
to
u
n
d
er
f
itt
in
g
.
E
x
p
er
i
m
en
ts
2
(
3
0
ep
o
c
h
s
)
an
d
8
(
5
0
ep
o
ch
s
)
s
h
o
w
s
im
ilar
m
etr
ic
s
,
wh
ic
h
s
h
o
ws
th
at
h
ig
h
er
ep
o
ch
s
p
r
ev
en
t
f
u
r
t
h
er
g
ain
af
ter
a
p
o
i
n
t,
d
u
e
to
ea
r
ly
co
n
v
er
g
en
ce
.
T
h
e
Pas
ca
l
VOC
2
0
1
2
d
ata
s
et
is
u
s
ed
f
o
r
ex
p
er
im
e
n
tal
p
u
r
p
o
s
es.
T
h
is
d
ata
s
et
co
m
p
r
is
es
1
1
,
5
3
0
im
a
g
es with
2
7
,
4
5
0
an
n
o
tated
R
OI
s
an
d
6
,
9
2
9
s
eg
m
e
n
tatio
n
s
in
2
0
u
n
iq
u
e
o
b
ject
cl
ass
es.
T
h
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
e
f
f
ec
tiv
ely
d
etec
te
d
an
d
lo
ca
lized
o
b
jects
in
v
ar
io
u
s
s
ettin
g
s
.
I
n
Fig
u
r
e
3
,
it
ac
cu
r
ately
id
en
tifie
d
a
h
o
r
s
e
with
h
ig
h
co
n
f
id
e
n
ce
,
c
o
r
r
ec
tly
d
r
awin
g
th
e
b
o
u
n
d
in
g
b
o
x
es.
T
h
e
m
o
d
el
also
r
ec
o
g
n
ized
th
e
h
o
r
s
e’
s
d
o
m
in
an
t
co
at
co
l
o
r
as
d
ar
k
s
late
g
r
ay
(
R
GB
:
5
4
,
5
4
,
5
2
)
,
d
em
o
n
s
tr
atin
g
its
ab
ilit
y
to
p
r
o
v
id
e
d
etailed
in
f
o
r
m
atio
n
ab
o
u
t
d
etec
ted
o
b
jects.
Similar
ly
,
th
e
f
r
am
ewo
r
k
d
em
o
n
s
tr
ated
s
tr
o
n
g
co
w
im
ag
e
d
etec
tio
n
in
Fig
u
r
e
4
,
with
a
h
ig
h
co
n
f
id
e
n
ce
s
co
r
e
o
f
0
.
9
7
.
T
h
e
r
esu
lts
s
h
o
w
th
at
f
ac
to
r
s
s
u
c
h
as
co
w
p
o
s
tu
r
e,
lig
h
tin
g
,
an
d
th
e
ex
ten
t
t
o
wh
ich
th
e
a
n
im
al
o
cc
u
p
ies
th
e
f
r
am
e
ca
n
s
ig
n
if
ican
tly
in
f
lu
en
ce
th
e
m
o
d
el’
s
co
n
f
id
en
ce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
6
3
-
872
868
T
h
e
m
o
d
el
co
u
l
d
b
e
im
p
r
o
v
e
d
to
p
er
f
o
r
m
m
o
r
e
co
n
s
is
ten
tly
in
d
if
f
er
en
t
v
a
r
iatio
n
s
o
f
th
e
s
am
e
o
b
ject
class
.
I
ts
co
lo
r
an
aly
s
is
s
y
s
tem
p
r
o
v
id
es
ac
cu
r
ate
co
lo
r
in
f
o
r
m
atio
n
,
as
s
h
o
wn
b
y
co
r
r
ec
tly
id
e
n
tify
in
g
th
e
‘
Sien
n
a
’
co
lo
r
s
.
T
ab
le
1
.
Per
f
o
r
m
an
ce
m
etr
ics ac
r
o
s
s
ex
p
er
im
en
ts
Ex
p
.
Le
a
r
n
i
n
g
r
a
t
e
Ep
o
c
h
s
M
e
a
n
p
r
e
c
i
s
i
o
n
M
e
a
n
r
e
c
a
l
l
M
e
a
n
F
1
sc
o
r
e
M
e
a
n
I
o
U
mA
P
1
0
.
0
0
5
30
0
.
7
7
7
8
0
.
4
5
7
9
0
.
5
5
0
1
0
.
6
8
7
3
0
.
7
7
7
8
2
0
.
0
0
5
30
0
.
5
3
4
7
0
.
4
7
6
9
0
.
4
6
5
7
0
.
6
6
6
1
0
.
5
3
4
7
3
0
.
0
0
1
20
0
.
6
1
2
2
0
.
5
9
5
1
0
.
5
7
8
9
0
.
6
7
0
0
0
.
6
1
2
2
4
0
.
0
1
0
20
0
.
2
5
1
0
0
.
1
2
3
1
0
.
1
5
4
2
0
.
6
9
0
1
0
.
2
5
1
0
5
0
.
0
0
5
20
0
.
5
5
8
8
0
.
3
1
4
7
0
.
3
7
3
7
0
.
6
5
1
4
0
.
5
5
8
8
6
0
.
0
0
1
30
0
.
7
4
6
3
0
.
5
8
0
2
0
.
6
2
4
2
0
.
7
0
4
0
0
.
7
4
6
3
7
0
.
0
0
0
1
20
0
.
1
5
0
0
0
.
0
7
5
5
0
.
0
9
8
8
0
.
6
6
9
3
0
.
1
5
0
0
8
0
.
0
0
0
1
50
0
.
5
3
4
7
0
.
4
7
6
9
0
.
4
6
5
7
0
.
6
6
6
1
0
.
5
3
4
7
9
0
.
0
0
5
10
0
.
6
6
6
7
0
.
3
4
4
6
0
.
4
3
0
6
0
.
6
7
4
4
0
.
6
6
6
7
10
0
.
0
0
5
50
0
.
8
1
1
9
0
.
5
3
7
6
0
.
6
2
3
2
0
.
7
0
9
6
0
.
8
1
1
9
Fig
u
r
e
3
.
Ob
ject
d
etec
tio
n
an
d
co
lo
r
id
e
n
tific
atio
n
o
f
h
o
r
s
e
Fig
u
r
e
4
.
Ob
ject
d
etec
tio
n
an
d
co
lo
r
id
e
n
tific
atio
n
o
f
co
w
HOG
is
a
wid
ely
u
s
ed
f
ea
tu
r
e
d
escr
ip
to
r
in
co
m
p
u
ter
v
is
io
n
an
d
im
a
g
e
p
r
o
ce
s
s
in
g
f
o
r
o
b
ject
d
etec
tio
n
.
HOG
v
is
u
aliza
tio
n
s
p
r
o
v
id
e
in
f
o
r
m
atio
n
ab
o
u
t
h
o
w
an
im
ag
e
is
ch
ar
ac
ter
ized
b
y
its
ed
g
e
o
r
ien
tatio
n
s
an
d
in
ten
s
ities
.
Fig
u
r
es
5
an
d
6
s
h
o
w
HOG
v
is
u
aliza
tio
n
o
f
a
n
air
c
r
af
t
an
d
a
b
u
s
,
r
esp
ec
tiv
ely
.
I
n
Fig
u
r
e
5
,
s
tr
o
n
g
g
r
ad
ien
t
p
atter
n
s
alo
n
g
th
e
win
g
s
,
f
u
s
elag
e,
an
d
tail
o
f
th
e
p
lan
e
h
ig
h
li
g
h
t
its
ae
r
o
d
y
n
am
i
c
f
ea
tu
r
es
an
d
s
h
o
w
th
at
th
e
f
r
a
m
ewo
r
k
f
o
cu
s
es
o
n
s
h
ar
p
e
d
g
es
an
d
co
n
to
u
r
s
.
Similar
ly
,
th
e
HOG
v
is
u
aliza
tio
n
in
Fig
u
r
e
6
h
ig
h
lig
h
ts
s
tr
o
n
g
h
o
r
izo
n
tal
a
n
d
v
er
tical
g
r
ad
ien
t
s
ar
o
u
n
d
k
e
y
b
u
s
f
ea
tu
r
es
lik
e
th
e
win
d
s
h
ield
a
n
d
h
ea
d
lig
h
ts
.
T
h
is
in
d
icate
s
th
at
th
e
f
r
am
ewo
r
k
id
e
n
tifie
s
b
u
s
es
b
y
f
o
cu
s
in
g
o
n
g
eo
m
et
r
ic
s
h
ap
es
an
d
e
d
g
e
p
atter
n
s
,
ef
f
ec
tiv
ely
u
s
in
g
s
tr
u
ctu
r
al
an
d
a
n
g
u
lar
ch
ar
ac
ter
is
t
ics f
o
r
v
eh
icle
r
ec
o
g
n
itio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
ted
d
ee
p
lea
r
n
in
g
a
p
p
r
o
a
ch
fo
r
r
ea
l
-
time
o
b
ject
d
etec
tio
n
a
n
d
…
(
S
r
in
iva
s
Dib
b
u
r
B
yra
p
p
a
)
869
Gr
ad
-
C
AM
v
is
u
aliza
tio
n
s
ar
e
a
p
o
wer
f
u
l
to
o
l
f
o
r
in
ter
p
r
e
tin
g
o
u
r
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
,
r
ev
ea
lin
g
wh
ic
h
ar
ea
s
o
f
a
n
i
m
ag
e
ar
e
th
e
m
o
s
t
in
f
lu
en
tial
in
th
e
m
o
d
el
p
r
ed
ictio
n
s
.
Gr
a
d
-
C
AM
h
ea
tm
ap
i
n
Fig
u
r
es
7
an
d
8
s
h
o
ws
th
at
th
e
m
o
d
el
m
ain
ly
h
ig
h
lig
h
ts
th
e
co
w’
s
b
o
d
y
an
d
h
ea
d
.
T
h
e
f
r
am
ewo
r
k
f
o
cu
s
es
o
n
p
ar
ts
o
f
th
e
an
im
al
r
ic
h
in
ca
t
eg
o
r
y
-
d
ef
in
in
g
f
ea
tu
r
es
an
d
d
em
o
n
s
tr
ates
its
ab
ilit
y
to
ac
cu
r
ately
id
en
tify
a
n
d
lo
ca
te
co
ws
in
th
e
im
ag
e.
T
h
is
v
is
u
aliza
tio
n
s
h
o
ws
th
at
th
e
m
o
d
el
r
elies
n
o
t
o
n
ly
o
n
th
e
o
v
er
all
s
h
ap
e
o
f
th
e
co
w,
b
u
t a
ls
o
o
n
v
ar
iatio
n
s
in
s
u
r
f
ac
e
p
atter
n
s
,
co
l
o
r
s
,
an
d
tex
tu
r
es wh
en
m
a
k
in
g
its
p
r
e
d
ic
tio
n
s
.
Fig
u
r
e
5
.
HOG
v
is
u
aliza
tio
n
o
f
p
lan
e
Fig
u
r
e
6
.
HOG
v
is
u
aliza
tio
n
o
f
b
u
s
Fig
u
r
e
7
.
Gr
a
d
-
C
AM
v
is
u
aliza
tio
n
o
f
c
o
w
Fig
u
r
e
8
.
Gr
a
d
-
C
AM
v
is
u
aliza
tio
n
o
f
c
o
w
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
6
3
-
872
870
5.
C
O
N
C
L
U
SIO
N
Fin
ally
,
th
is
s
tu
d
y
also
b
r
i
n
g
s
a
co
n
tr
ib
u
tio
n
to
th
e
in
teg
r
at
ed
ex
am
i
n
atio
n
o
f
o
b
ject
d
ete
ctio
n
an
d
co
lo
r
i
n
ter
p
r
etatio
n
b
ec
a
u
s
e
o
f
u
s
in
g
lar
g
e
-
s
ca
le
d
atasets
,
co
m
p
le
x
a
u
g
m
en
tatio
n
m
et
h
o
d
s
,
an
d
th
e
m
o
s
t
r
ec
en
t
d
ee
p
lea
r
n
in
g
m
o
d
els,
in
clu
d
in
g
Fas
ter
R
-
C
NN
wi
th
a
R
esNet
-
5
0
b
ac
k
b
o
n
e
an
d
f
ea
tu
r
e
p
y
r
am
i
d
n
etwo
r
k
s
.
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
h
as
a
h
ig
h
q
u
an
titativ
e
p
er
f
o
r
m
a
n
ce
with
a
m
AP
o
f
0
.
8
1
1
9
o
n
th
e
Pas
ca
l
VOC
2
0
1
2
d
ataset
an
d
s
h
o
ws r
esil
ien
ce
in
a
v
ar
iety
o
f
o
b
jec
t
ca
teg
o
r
ies
an
d
d
em
a
n
d
in
g
v
i
s
u
al
s
ettin
g
s
d
u
e
to
th
e
in
teg
r
atio
n
o
f
k
-
m
ea
n
s
clu
s
ter
in
g
to
ex
tr
ac
t
d
o
m
i
n
an
t
c
o
lo
r
s
an
d
ex
p
lain
ab
le
AI
s
o
lu
ti
o
n
s
lik
e
HOG
an
d
Gr
ad
-
C
AM
.
T
h
e
ca
p
ab
ilit
y
o
f
th
e
m
o
d
el
to
g
iv
e
ac
c
u
r
ate
o
b
ject
lo
ca
lizatio
n
alo
n
g
with
s
em
an
tically
d
escr
ip
tiv
e
co
lo
r
f
ea
tu
r
es
m
ak
es
it
a
u
s
ef
u
l
to
o
l
in
r
ea
l
-
wo
r
ld
a
p
p
licatio
n
s
in
s
u
ch
f
ield
s
as
in
d
u
s
tr
ial
in
s
p
ec
tio
n
,
a
u
to
n
o
m
o
u
s
s
y
s
te
m
s
an
d
m
ed
ical
im
ag
in
g
.
Gen
er
ally
,
th
e
wo
r
k
p
r
esen
ts
a
v
al
u
ab
le
a
n
d
d
ev
elo
p
ed
f
r
am
ewo
r
k
th
at
s
titch
es
th
e
d
is
p
ar
ity
b
etwe
en
th
eo
r
etica
l
d
ev
elo
p
m
en
ts
an
d
p
r
ac
tical
asp
ec
ts
o
f
co
m
p
u
ter
v
is
io
n
n
ee
d
s
.
T
h
is
f
r
am
ewo
r
k
will
b
e
ex
ten
d
ed
in
f
u
tu
r
e
d
i
r
e
ctio
n
s
to
f
ac
ilit
ate
r
ea
l
-
tim
e
p
r
o
ce
s
s
in
g
,
ad
ap
tiv
e
b
eh
av
io
r
,
d
o
m
ain
g
en
e
r
aliza
tio
n
a
n
d
e
x
p
lain
ab
ilit
y
th
u
s
in
c
r
ea
s
in
g
its
ca
p
ab
ilit
y
to
b
e
f
u
r
th
er
u
s
ed
in
p
r
ac
tice
an
d
r
ea
l
-
life
s
itu
atio
n
s
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
d
id
n
o
t r
ec
eiv
e
a
n
y
g
r
a
n
ts
f
r
o
m
a
n
y
f
u
n
d
in
g
a
g
en
cies.
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
Srin
iv
as Dib
b
u
r
B
y
r
ap
p
a
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Ku
s
h
al
Gaje
n
d
r
a
✓
✓
✓
✓
✓
✓
✓
R
o
h
ith
Ho
len
ar
asip
u
r
a
Pu
ttar
aju
✓
✓
✓
✓
✓
✓
✓
✓
T
u
m
ak
alah
alli N
ag
ar
aj
Ma
lin
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
T
h
e
au
th
o
r
s
d
ec
lar
e
n
o
co
n
f
lict o
f
in
ter
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
a
t
a
t
h
a
t
s
u
p
p
o
r
t
t
h
e
f
i
n
d
i
n
g
s
o
f
t
h
i
s
s
t
u
d
y
a
r
e
a
v
a
il
a
b
l
e
o
n
r
e
q
u
e
s
t
f
r
o
m
t
h
e
c
o
r
r
e
s
p
o
n
d
in
g
a
u
t
h
o
r
.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
L
.
G
a
l
v
e
z
,
A
.
A
.
B
a
n
d
a
l
a
,
E
.
P
.
D
a
d
i
o
s
,
R
.
R
.
P
.
V
i
c
e
r
r
a
,
a
n
d
J
.
M
.
Z
.
M
a
n
i
n
g
o
,
“
O
b
j
e
c
t
d
e
t
e
c
t
i
o
n
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
,
”
i
n
T
E
N
C
O
N
2
0
1
8
-
2
0
1
8
I
E
E
E
R
e
g
i
o
n
1
0
C
o
n
f
e
r
e
n
c
e
,
O
c
t
.
2
0
1
8
,
p
p
.
2
0
2
3
–
2
0
2
7
,
d
o
i
:
1
0
.
1
1
0
9
/
T
E
N
C
O
N
.
2
0
1
8
.
8
6
5
0
5
1
7
.
[
2
]
G
.
C
h
a
n
d
a
n
,
A
.
J
a
i
n
,
H
.
J
a
i
n
,
a
n
d
M
o
h
a
n
a
,
“
R
e
a
l
t
i
m
e
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
t
r
a
c
k
i
n
g
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
O
p
e
n
C
V
,
”
i
n
2
0
1
8
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
I
n
v
e
n
t
i
v
e
Re
s
e
a
r
c
h
i
n
C
o
m
p
u
t
i
n
g
A
p
p
l
i
c
a
t
i
o
n
s
(
I
C
I
R
C
A)
,
Ju
l
.
2
0
1
8
,
p
p
.
1
3
0
5
–
1
3
0
8
,
doi
:
1
0
.
1
1
0
9
/
I
C
I
R
C
A
.
2
0
1
8
.
8
5
9
7
2
6
6
.
[
3
]
A
.
K
ö
r
e
z
a
n
d
N
.
B
a
r
ı
ş
ç
ı
,
“
O
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
l
o
w
c
a
p
a
c
i
t
y
G
P
U
s
y
s
t
e
ms
u
s
i
n
g
i
m
p
r
o
v
e
d
F
a
st
e
r
R
-
C
N
N
,
”
Ap
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
0
,
n
o
.
1
,
p
.
8
3
,
D
e
c
.
2
0
1
9
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
0
0
1
0
0
8
3
.
[
4
]
H
.
D
e
sh
p
a
n
d
e
,
A
.
S
i
n
g
h
,
a
n
d
H
.
H
e
r
u
n
d
e
,
“
C
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
n
Y
O
LO
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
O
p
e
n
C
V
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
rese
a
r
c
h
i
n
i
n
d
u
st
r
i
a
l
e
n
g
i
n
e
e
ri
n
g
,
v
o
l
.
9
,
n
o
.
1
,
p
p
.
4
6
–
6
4
,
2
0
2
0
.
[
5
]
H
.
G
o
n
g
e
t
a
l
.
,
“
S
w
i
n
-
t
r
a
n
sf
o
r
mer
-
e
n
a
b
l
e
d
Y
O
LO
v
5
w
i
t
h
a
t
t
e
n
t
i
o
n
me
c
h
a
n
i
sm
f
o
r
sma
l
l
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
o
n
sat
e
l
l
i
t
e
i
ma
g
e
s,
”
Re
m
o
t
e
S
e
n
s
i
n
g
,
v
o
l
.
1
4
,
n
o
.
1
2
,
p
.
2
8
6
1
,
J
u
n
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
r
s1
4
1
2
2
8
6
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
I
n
teg
r
a
ted
d
ee
p
lea
r
n
in
g
a
p
p
r
o
a
ch
fo
r
r
ea
l
-
time
o
b
ject
d
etec
tio
n
a
n
d
…
(
S
r
in
iva
s
Dib
b
u
r
B
yra
p
p
a
)
871
[
6
]
M
.
A
.
S
h
e
h
a
b
,
A
.
A
l
-
G
i
z
i
,
a
n
d
S
.
M
.
S
w
a
d
i
,
“
Ef
f
i
c
i
e
n
t
r
e
a
l
-
t
i
me
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
b
a
s
e
d
o
n
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
,
”
i
n
2
0
2
1
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ap
p
l
i
e
d
a
n
d
T
h
e
o
ret
i
c
a
l
El
e
c
t
r
i
c
i
t
y
(
I
C
ATE)
,
M
a
y
2
0
2
1
,
p
p
.
1
–
5
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
TE4
9
6
8
5
.
2
0
2
1
.
9
4
6
5
0
1
5
.
[
7
]
H
.
M
o
u
j
a
h
i
d
e
t
a
l
.
,
“
C
o
m
b
i
n
i
n
g
C
N
N
a
n
d
G
r
a
d
-
C
A
M
f
o
r
C
O
V
I
D
-
1
9
d
i
sea
se
p
r
e
d
i
c
t
i
o
n
a
n
d
v
i
s
u
a
l
e
x
p
l
a
n
a
t
i
o
n
,
”
I
n
t
e
l
l
i
g
e
n
t
Au
t
o
m
a
t
i
o
n
&
S
o
f
t
C
o
m
p
u
t
i
n
g
,
v
o
l
.
3
2
,
n
o
.
2
,
p
p
.
7
2
3
–
7
4
5
,
2
0
2
2
,
d
o
i
:
1
0
.
3
2
6
0
4
/
i
a
sc
.
2
0
2
2
.
0
2
2
1
7
9
.
[
8
]
E.
S
u
h
e
r
ma
n
,
B
.
R
a
h
ma
n
,
D
.
H
i
n
d
a
r
t
o
,
a
n
d
H
.
S
a
n
t
o
so
,
“
I
mp
l
e
m
e
n
t
a
t
i
o
n
o
f
R
e
sN
e
t
-
5
0
o
n
e
n
d
-
to
-
e
n
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
(
D
E
TR
)
o
n
o
b
j
e
c
t
s,
”
S
i
n
k
rO
n
,
v
o
l
.
7
,
n
o
.
2
,
p
p
.
1
0
8
5
–
1
0
9
6
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
3
9
5
/
s
i
n
k
r
o
n
.
v
8
i
2
.
1
2
3
7
8
.
[
9
]
C.
-
L.
L
i
n
,
Y
.
-
L.
C
h
e
n
,
a
n
d
Y
.
-
C
.
L
i
n
,
“
S
m
a
l
l
o
b
j
e
c
t
s
d
e
t
e
c
t
i
o
n
u
s
i
n
g
t
r
a
n
sf
o
r
m
e
r
-
C
N
N
,
”
S
S
R
N
El
e
c
t
ro
n
i
c
J
o
u
rn
a
l
,
2
0
2
2
,
d
o
i
:
1
0
.
2
1
3
9
/
ssr
n
.
4
2
3
2
8
5
0
.
[
1
0
]
H
.
Y
a
n
a
g
i
sawa
,
T.
Y
a
m
a
sh
i
t
a
,
a
n
d
H
.
W
a
t
a
n
a
b
e
,
“
A
st
u
d
y
o
n
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
m
e
t
h
o
d
f
r
o
m
m
a
n
g
a
i
ma
g
e
s
u
si
n
g
C
N
N
,
”
i
n
2
0
1
8
I
n
t
e
r
n
a
t
i
o
n
a
l
W
o
rks
h
o
p
o
n
A
d
v
a
n
c
e
d
I
m
a
g
e
T
e
c
h
n
o
l
o
g
y
(
I
WA
I
T
)
,
Ja
n
.
2
0
1
8
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
W
A
I
T.
2
0
1
8
.
8
3
6
9
6
3
3
.
[
1
1
]
S
.
A
.
N
a
n
d
S
.
M
.
N
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
a
n
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
o
n
sa
t
e
l
l
i
t
e
i
mag
e
s
u
s
i
n
g
c
u
st
o
m
C
N
N
a
r
c
h
i
t
e
c
t
u
r
e
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
I
n
n
o
v
a
t
i
v
e
Re
s
e
a
r
c
h
i
n
C
o
m
p
u
t
e
r
a
n
d
C
o
m
m
u
n
i
c
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
2
,
n
o
.
0
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
5
6
8
0
/
i
j
i
r
se
t
.
2
0
2
3
.
1
2
0
7
1
9
0
.
[
1
2
]
E.
G
ü
n
e
y
a
n
d
C
.
B
a
y
i
l
mi
s
,
“
A
n
i
m
p
l
e
me
n
t
a
t
i
o
n
o
f
t
r
a
f
f
i
c
si
g
n
s
a
n
d
r
o
a
d
o
b
j
e
c
t
s
d
e
t
e
c
t
i
o
n
u
si
n
g
f
a
s
t
e
r
R
-
C
N
N
,
”
S
a
k
a
ry
a
U
n
i
v
e
rsi
t
y
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
s
,
v
o
l
.
5
,
n
o
.
2
,
p
p
.
2
1
6
–
2
2
4
,
2
0
2
2
.
[
1
3
]
P
.
D
so
u
z
a
,
“
D
e
t
e
c
t
i
o
n
o
f
sma
l
l
o
b
j
e
c
t
s
u
s
i
n
g
C
N
N
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
R
e
se
a
rc
h
J
o
u
r
n
a
l
o
f
Mo
d
e
rn
i
za
t
i
o
n
i
n
E
n
g
i
n
e
e
ri
n
g
T
e
c
h
n
o
l
o
g
y
a
n
d
S
c
i
e
n
c
e
,
v
o
l
.
5
,
n
o
.
5
,
p
.
6
8
1
8
,
M
a
y
2
0
2
3
,
d
o
i
:
1
0
.
5
6
7
2
6
/
I
R
JM
ETS
4
0
3
4
3
.
[
1
4
]
S
.
V
a
t
c
h
a
l
a
,
S
.
S
a
si
d
e
v
i
,
D
h
a
n
a
l
a
k
s
h
mi
R
,
a
n
d
S
a
i
R
a
m
e
sh
L
,
“
S
mart
h
o
u
s
e
h
o
l
d
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
u
s
i
n
g
C
N
N
,
”
A
d
v
a
n
c
e
s
i
n
Pa
r
a
l
l
e
l
C
o
m
p
u
t
i
n
g
A
l
g
o
ri
t
h
m
s,
T
o
o
l
s a
n
d
P
a
ra
d
i
g
m
s
,
v
o
l
.
4
1
,
p
.
4
6
4
,
2
0
2
2
,
d
o
i
:
1
0
.
3
2
3
3
/
A
P
C
2
2
0
0
6
5
.
[
1
5
]
F
.
M
.
T
.
R
.
K
i
n
a
si
h
,
C
.
M
a
c
h
b
u
b
,
L.
Y
u
l
i
a
n
t
i
,
a
n
d
A
.
S
.
R
o
h
ma
n
,
“
Tw
o
-
st
a
g
e
mu
l
t
i
p
l
e
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
u
si
n
g
C
N
N
a
n
d
c
o
r
r
e
l
a
t
i
v
e
f
i
l
t
e
r
f
o
r
a
c
c
u
r
a
c
y
i
mp
r
o
v
e
men
t
,
”
H
e
l
i
y
o
n
,
v
o
l
.
9
,
n
o
.
1
,
p
.
e
1
2
7
1
6
,
J
a
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
2
.
e
1
2
7
1
6
.
[
1
6
]
S
.
R
e
n
,
K
.
H
e
,
R
.
G
i
r
s
h
i
c
k
,
a
n
d
J
.
S
u
n
,
“
F
a
s
t
e
r
R
-
C
N
N
:
t
o
w
a
r
d
s
r
e
a
l
-
t
i
me
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
r
e
g
i
o
n
p
r
o
p
o
sa
l
n
e
t
w
o
r
k
s,”
I
EE
E
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
Ma
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
3
9
,
n
o
.
6
,
p
p
.
1
1
3
7
–
1
1
4
9
,
J
u
n
.
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
1
6
.
2
5
7
7
0
3
1
.
[
1
7
]
K
.
T
o
n
g
a
n
d
Y
.
W
u
,
“
R
e
t
h
i
n
k
i
n
g
P
A
S
C
A
L
-
V
O
C
a
n
d
M
S
-
C
O
C
O
d
a
t
a
set
f
o
r
s
mal
l
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
Vi
s
u
a
l
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
I
m
a
g
e
Re
p
res
e
n
t
a
t
i
o
n
,
v
o
l
.
9
3
,
p
.
1
0
3
8
3
0
,
M
a
y
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
v
c
i
r
.
2
0
2
3
.
1
0
3
8
3
0
.
[
1
8
]
J.
S
u
i
,
L.
Y
a
n
g
,
Y
.
W
a
n
g
,
a
n
d
Z
.
H
u
a
,
“
A
n
o
b
j
e
c
t
c
o
l
o
r
r
e
c
o
g
n
i
t
i
o
n
a
l
g
o
r
i
t
h
m
b
a
s
e
d
o
n
st
u
d
y
-
e
x
p
a
n
si
o
n
m
e
t
h
o
d
,
”
i
n
2
0
0
9
9
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
E
l
e
c
t
r
o
n
i
c
Me
a
su
reme
n
t
& I
n
st
r
u
m
e
n
t
s
,
A
u
g
.
2
0
0
9
,
p
p
.
4
–
205
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
EM
I
.
2
0
0
9
.
5
2
7
4
1
0
6
.
[
1
9
]
R
.
P
.
S
i
d
i
k
a
n
d
E.
C
o
n
t
e
ssa
D
j
a
ma
l
,
“
F
a
c
e
mas
k
d
e
t
e
c
t
i
o
n
u
si
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
i
n
2
0
2
1
4
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
f
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
c
s
E
n
g
i
n
e
e
r
i
n
g
(
I
C
2
I
E)
,
S
e
p
.
2
0
2
1
,
p
p
.
8
5
–
89
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
2
I
E5
3
2
1
9
.
2
0
2
1
.
9
6
4
9
0
6
5
.
[
2
0
]
Y
u
h
a
n
d
r
i
,
M
u
s
l
i
Y
a
n
t
o
,
a
n
d
Ek
a
N
a
u
f
a
l
d
i
N
o
v
r
i
,
“
A
p
p
l
i
c
a
t
i
o
n
o
f
o
b
j
e
c
t
mask
d
e
t
e
c
t
i
o
n
u
si
n
g
t
h
e
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
(
C
N
N
)
,
”
J
u
rn
a
l
RE
S
T
I
(
Re
k
a
y
a
s
a
S
i
s
t
e
m
d
a
n
T
e
k
n
o
l
o
g
i
I
n
f
o
rm
a
s
i
)
,
v
o
l
.
7
,
n
o
.
4
,
p
p
.
9
2
2
–
9
2
9
,
A
u
g
.
2
0
2
3
,
d
o
i
:
1
0
.
2
9
2
0
7
/
r
e
st
i
.
v
7
i
4
.
5
0
5
9
.
[
2
1
]
Y
.
F
a
n
g
,
S
.
Y
a
n
g
,
S
.
W
a
n
g
,
Y
.
G
e
,
Y
.
S
h
a
n
,
a
n
d
X
.
W
a
n
g
,
“
U
n
l
e
a
sh
i
n
g
v
a
n
i
l
l
a
v
i
s
i
o
n
t
r
a
n
sf
o
r
mer
w
i
t
h
m
a
s
k
e
d
i
ma
g
e
mo
d
e
l
i
n
g
f
o
r
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
,
”
i
n
2
0
2
3
I
E
EE/
C
VF
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
V
i
si
o
n
(
I
C
C
V)
,
O
c
t
.
2
0
2
3
,
p
p
.
6
2
2
1
–
6
2
3
0
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
V
5
1
0
7
0
.
2
0
2
3
.
0
0
5
7
4
.
[
2
2
]
M
.
M
i
n
d
e
r
e
r
e
t
a
l
.
,
“
S
i
m
p
l
e
o
p
e
n
-
v
o
c
a
b
u
l
a
r
y
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
,
”
i
n
E
u
r
o
p
e
a
n
c
o
n
f
e
re
n
c
e
o
n
c
o
m
p
u
t
e
r v
i
si
o
n
,
2
0
2
2
,
p
p
.
7
2
8
–
7
5
5
.
[
2
3
]
C.
-
H
.
Y
e
h
e
t
a
l
.
,
“
L
i
g
h
t
w
e
i
g
h
t
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
j
o
i
n
t
l
e
a
r
n
i
n
g
o
f
u
n
d
e
r
w
a
t
e
r
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
c
o
l
o
r
c
o
n
v
e
r
s
i
o
n
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
N
e
u
ra
l
N
e
t
w
o
rk
s
a
n
d
L
e
a
r
n
i
n
g
S
y
s
t
e
m
s
,
v
o
l
.
3
3
,
n
o
.
1
1
,
p
p
.
6
1
2
9
–
6
1
4
3
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
TN
N
LS.
2
0
2
1
.
3
0
7
2
4
1
4
.
[
2
4
]
K
.
S
.
A
a
r
a
t
h
i
a
n
d
A
.
A
b
r
a
h
a
m
,
“
V
e
h
i
c
l
e
c
o
l
o
r
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
f
o
r
h
a
z
y
i
ma
g
e
s
,
”
i
n
2
0
1
7
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
v
e
n
t
i
v
e
C
o
m
m
u
n
i
c
a
t
i
o
n
a
n
d
C
o
m
p
u
t
a
t
i
o
n
a
l
T
e
c
h
n
o
l
o
g
i
e
s
(
I
C
I
C
C
T
)
,
M
a
r
.
2
0
1
7
,
p
p
.
3
3
5
–
3
3
9
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
C
C
T.
2
0
1
7
.
7
9
7
5
2
1
5
.
[
2
5
]
P
.
R
a
g
u
r
a
ma
n
,
A
.
M
e
g
h
a
n
a
,
Y
.
N
a
v
y
a
,
S
.
K
a
r
i
sh
ma,
a
n
d
S
.
I
sw
a
r
y
a
,
“
C
o
l
o
r
d
e
t
e
c
t
i
o
n
o
f
R
G
B
i
ma
g
e
s
u
si
n
g
P
y
t
h
o
n
a
n
d
O
p
e
n
C
v
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
S
c
i
e
n
t
i
f
i
c
Re
s
e
a
r
c
h
i
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
E
n
g
i
n
e
e
r
i
n
g
a
n
d
I
n
f
o
rm
a
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
p
p
.
1
0
9
–
1
1
2
,
2
0
2
1
,
d
o
i
:
1
0
.
3
2
6
2
8
/
C
S
EI
T2
1
7
1
1
9
.
[
2
6
]
S
.
C
.
G
u
p
t
a
,
N
.
R
.
S
i
n
g
h
,
T
.
S
h
a
r
m
a
,
A
.
T
y
a
g
i
,
a
n
d
R
.
M
a
j
u
m
d
a
r
,
“
G
e
n
e
r
a
t
i
n
g
i
ma
g
e
c
a
p
t
i
o
n
s
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
s
si
n
g
,
”
i
n
2
0
2
1
9
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
Re
l
i
a
b
i
l
i
t
y
,
I
n
f
o
c
o
m
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
O
p
t
i
m
i
z
a
t
i
o
n
(
T
re
n
d
s
a
n
d
Fu
t
u
r
e
D
i
r
e
c
t
i
o
n
s)
(
I
C
RI
T
O
)
,
S
e
p
.
2
0
2
1
,
p
p
.
1
–
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
R
I
TO
5
1
3
9
3
.
2
0
2
1
.
9
5
9
6
4
8
6
.
[
2
7
]
K
.
G
a
j
e
n
d
r
a
a
n
d
S
.
D
.
B
y
r
a
p
p
a
,
“
R
e
c
e
n
t
a
d
v
a
n
c
e
me
n
t
s
a
n
d
c
h
a
l
l
e
n
g
e
s
i
n
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
a
n
d
c
o
l
o
r
i
d
e
n
t
i
f
i
c
a
t
i
o
n
:
a
s
u
r
v
e
y
,
”
H
i
g
h
T
e
c
h
n
o
l
o
g
y
L
e
t
t
e
rs
,
v
o
l
.
3
0
,
n
o
.
5
,
2
0
2
4
.
A
c
c
e
s
sed
:
M
a
y
1
7
,
2
0
2
5
.
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
d
r
i
v
e
.
g
o
o
g
l
e
.
c
o
m
/
f
i
l
e
/
d
/
1
N
8
O
i
_
x
k
y
M
mn
n
k
_
e
W
f
w
5
e
0
-
y
Y
s
B
s
v
a
0
R
7
/
v
i
e
w
[
2
8
]
J.
v
a
n
d
e
W
e
i
j
e
r
,
C
.
S
c
h
mi
d
,
J
.
V
e
r
b
e
e
k
,
a
n
d
D
.
L
a
r
l
u
s,
“
Le
a
r
n
i
n
g
c
o
l
o
r
n
a
mes
f
o
r
r
e
a
l
-
w
o
r
l
d
a
p
p
l
i
c
a
t
i
o
n
s,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
m
a
g
e
Pro
c
e
ss
i
n
g
,
v
o
l
.
1
8
,
n
o
.
7
,
p
p
.
1
5
1
2
–
1
5
2
3
,
J
u
l
.
2
0
0
9
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
P
.
2
0
0
9
.
2
0
1
9
8
0
9
.
[
2
9
]
J.
v
a
n
d
e
W
e
i
j
e
r
,
C
.
S
c
h
m
i
d
,
a
n
d
J.
V
e
r
b
e
e
k
,
“
Le
a
r
n
i
n
g
c
o
l
o
r
n
a
mes
f
r
o
m
r
e
a
l
-
w
o
r
l
d
i
m
a
g
e
s,
”
i
n
2
0
0
7
I
EE
E
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
,
J
u
n
.
2
0
0
7
,
p
p
.
1
–
8
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
0
7
.
3
8
3
2
1
8
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
S
r
in
iv
a
s
Dib
b
u
r
B
y
r
a
p
p
a
re
c
e
iv
e
d
h
is
P
h
.
D
.
d
e
g
re
e
in
c
o
m
p
u
ter
a
n
d
in
fo
rm
a
ti
o
n
sc
ien
c
e
s
fro
m
Visv
e
sv
a
ra
y
a
Tec
h
n
o
lo
g
ica
l
Un
i
v
e
rsity
,
In
d
ia,
in
2
0
1
9
.
He
i
s
c
u
rre
n
tl
y
a
p
ro
fe
ss
o
r
in
th
e
De
p
a
rtme
n
t
o
f
I
n
fo
rm
a
ti
o
n
S
c
ien
c
e
a
n
d
E
n
g
i
n
e
e
rin
g
a
t
Nitt
e
M
e
e
n
a
k
sh
i
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
,
Ba
n
g
a
lo
re
,
I
n
d
ia.
His
re
se
a
rc
h
in
tere
sts
a
re
in
p
a
ra
ll
e
l
c
o
m
p
u
ti
n
g
,
c
lo
u
d
c
o
m
p
u
ti
n
g
,
g
rid
/clu
ste
r
c
o
m
p
u
ti
n
g
,
a
n
d
c
o
m
p
u
ter
v
isio
n
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
srin
i
v
a
s.d
b
@n
m
it
.
a
c
.
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
6
3
-
872
872
K
u
sha
l
G
a
je
n
d
r
a
p
u
rs
u
in
g
M
.
Tec
h
.
in
d
a
ta
sc
ien
c
e
a
t
Nitt
e
M
e
e
n
a
k
sh
i
In
stit
u
te
o
f
Tec
h
n
o
l
o
g
y
,
Ba
n
g
a
lo
re
,
In
d
ia
a
n
d
c
o
m
p
lete
d
h
is
Ba
c
h
e
lo
r
o
f
E
n
g
i
n
e
e
rin
g
in
e
lec
tro
n
ic
a
n
d
c
o
m
m
u
n
ica
ti
o
n
fro
m
Visv
e
s
v
a
ra
y
a
Tec
h
n
o
lo
g
ica
l
Un
i
v
e
rsity
in
2
0
2
2
.
His
re
se
a
rc
h
in
tere
st
in
c
lu
d
e
s
re
a
l
-
ti
m
e
o
b
jec
t
d
e
tec
ti
o
n
a
n
d
e
x
p
lai
n
a
b
le
AI
m
o
d
e
ls.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
k
u
sh
a
l
k
o
ws
h
@
g
m
a
il
.
c
o
m
.
Ro
h
ith
H
o
lena
r
a
si
p
u
r
a
Pu
tta
r
a
ju
c
o
m
p
lete
d
m
y
M
.
Tec
h
.
(CS
E)
fro
m
M
.
S
.
Ra
m
a
iah
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
,
c
u
rre
n
tl
y
wo
r
k
in
g
a
s
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
i
n
th
e
De
p
a
rtme
n
t
o
f
I
n
fo
rm
a
ti
o
n
S
c
ien
c
e
a
n
d
En
g
in
e
e
rin
g
.
He
is
p
u
r
su
in
g
m
y
P
h
.
D
.
i
n
n
a
tu
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
fr
o
m
VTU.
His
a
re
a
s
o
f
in
tere
st
a
re
a
rti
f
icia
l
in
telli
g
e
n
c
e
,
m
a
c
h
in
e
lea
rn
in
g
a
n
d
c
o
m
p
u
ter
n
e
two
r
k
in
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
r
o
h
it
.
h
p
@n
m
it
.
a
c
.
i
n
.
Tu
m
a
k
a
l
a
h
a
ll
i
Na
g
a
r
a
j
Ma
l
in
i
is
c
u
r
re
n
tl
y
wo
r
k
i
n
g
a
s
a
p
r
o
fe
ss
o
r
a
t
th
e
De
p
a
rtme
n
t
o
f
M
a
n
a
g
e
m
e
n
t
S
tu
d
ies
,
Nitt
e
M
e
e
n
a
k
sh
i
In
st
it
u
te
o
f
Tec
h
n
o
l
o
g
y
,
Ba
n
g
a
l
o
re
,
h
o
l
d
in
g
1
9
y
e
a
rs
o
f
tea
c
h
i
n
g
a
n
d
re
se
a
rc
h
e
x
p
e
rie
n
c
e
in
o
rg
a
n
iza
ti
o
n
a
l
b
e
h
a
v
io
r
,
h
u
m
a
n
re
so
u
rc
e
m
a
n
a
g
e
m
e
n
t,
su
p
p
ly
c
h
a
in
a
n
d
lo
g
isti
c
s
m
a
n
a
g
e
m
e
n
t,
ima
g
e
p
r
o
c
e
ss
in
g
a
n
d
c
o
m
p
u
ter v
isio
n
.
S
h
e
c
a
n
b
e
c
o
n
t
a
c
ted
a
t
e
m
a
il
:
m
a
li
n
i.
tn
@
n
m
it
.
a
c
.
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.