I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
,
p
p
.
1
782
~1
7
9
2
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
38.
i
3
.
p
p
1
7
8
2
-
1
7
9
2
1782
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Identifica
tion a
nd
seg
menta
tion o
f
t
umo
r using
deep
l
ea
rning
a
nd ima
g
e seg
menta
tion a
lg
o
rithm
s
Sh
ilp
a
Ch
ipp
a
la
k
a
t
t
i
1
,
Renu
M
a
dh
a
v
i
Cho
da
v
a
ra
pu
2
,
An
dh
e
P
a
lla
v
i
3
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
,
S
i
r
M
V
i
sv
e
sv
a
r
a
y
a
I
n
st
i
t
u
t
e
o
f
Te
c
h
n
o
l
o
g
y
,
A
f
f
i
l
i
a
t
e
d
t
o
V
i
s
v
e
s
v
a
r
a
y
a
T
e
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
si
t
y
,
B
e
n
g
a
l
u
r
u
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
I
n
st
r
u
men
t
a
t
i
o
n
,
R
V
C
o
l
l
e
g
e
o
f
En
g
i
n
e
e
r
i
n
g
,
A
f
f
i
l
i
a
t
e
d
t
o
V
i
s
v
e
s
v
a
r
a
y
a
T
e
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
si
t
y
,
B
e
n
g
a
l
u
r
u
,
In
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
,
R
N
S
I
n
st
i
t
u
t
e
o
f
Te
c
h
n
o
l
o
g
y
,
A
f
f
i
l
i
a
t
e
d
t
o
V
i
s
v
e
s
v
a
r
a
y
a
T
e
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
si
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
Ju
n
20
,
2
0
2
4
R
ev
is
ed
No
v
13
,
2
0
2
4
Acc
ep
ted
No
v
30
,
2
0
2
4
Bra
in
tu
m
o
r
is
a
t
y
p
ica
l
m
a
ss
o
f
ti
ss
u
e
th
a
t
d
e
v
e
l
o
p
s
w
h
e
n
c
e
ll
s
p
ro
li
fe
ra
te
a
n
d
d
iv
i
d
e
e
x
c
e
ss
iv
e
ly
.
Bra
in
tu
m
o
r
p
e
rc
e
p
ti
o
n
re
q
u
ires
a
g
re
a
t
d
e
a
l
o
f
wo
r
k
a
n
d
e
x
p
e
rien
c
e
fro
m
t
h
e
m
e
d
ica
l
p
ro
fe
ss
io
n
a
l
in
o
rd
e
r
to
id
e
n
t
ify
t
h
e
tu
m
o
r'
s
p
re
c
ise
lo
c
a
ti
o
n
.
If
a
b
ra
i
n
tu
m
o
r
is
n
o
t
d
isc
o
v
e
re
d
i
n
a
ti
m
e
ly
m
a
n
n
e
r,
it
a
ffe
c
ts
a
p
e
rso
n
'
s
a
b
il
it
y
to
fu
n
c
ti
o
n
n
o
rm
a
ll
y
a
n
d
ra
ise
s
th
e
d
e
a
th
ra
te.
Th
is
stu
d
y
fo
c
u
se
s
o
n
tu
m
o
r
se
g
m
e
n
t
a
ti
o
n
a
n
d
tu
m
o
r
d
e
tec
ti
o
n
u
sin
g
m
a
g
n
e
ti
c
re
so
n
a
n
c
e
ima
g
in
g
(
M
RI
)
ima
g
e
s.
Th
is
wo
r
k
h
e
l
p
s
th
e
m
e
d
ica
l
p
r
o
fe
ss
io
n
a
l
to
p
re
c
ise
ly
i
d
e
n
t
ify
t
h
e
tu
m
o
r
l
o
c
a
ti
o
n
a
n
d
se
g
m
e
n
tati
o
n
p
r
o
c
e
ss
p
ro
v
id
e
s
c
o
st
e
ffe
c
ti
v
e
d
a
ta
st
o
ra
g
e
.
Th
e
YO
LOv
8
s
m
o
d
e
l
is
u
t
il
ize
d
fo
r
t
u
m
o
r
id
e
n
ti
fica
ti
o
n
,
wh
i
le
th
e
ima
g
e
se
g
m
e
n
tatio
n
tec
h
n
iq
u
e
is
e
m
p
lo
y
e
d
f
o
r
tu
m
o
r
se
g
m
e
n
tati
o
n
.
T
h
e
ima
g
e
s
c
o
m
e
fro
m
a
n
o
p
e
n
-
so
u
rc
e
d
a
tas
e
t
u
se
d
fo
r
re
se
a
rc
h
,
a
n
d
Ro
b
o
fl
o
w
1
0
0
tr
a
n
sfo
rm
s
th
e
m
in
t
o
.
y
a
m
l
fil
e
s
th
a
t
a
re
c
o
n
g
e
n
ial
with
YO
LOv
8
s.
T
o
train
th
e
m
o
d
e
l
th
e
d
a
tas
e
t
is
sp
li
t
in
t
o
train
in
g
,
v
a
li
d
a
ti
o
n
a
n
d
tes
ti
n
g
.
P
ro
p
o
se
d
m
o
d
e
l
c
o
n
sists
o
f
d
a
tas
e
t
wh
ich
c
o
m
p
rise
s
6
3
9
ima
g
e
s,
o
f
w
h
ic
h
4
5
3
a
re
u
ti
li
z
e
d
f
o
r
train
i
n
g
,
1
2
2
fo
r
v
a
li
d
a
ti
o
n
,
a
n
d
6
4
f
o
r
tes
ti
n
g
,
re
su
lt
in
g
in
a
ra
ti
o
o
f
7
1
:1
9
:1
0
.
Th
e
d
a
tas
e
t
is
su
b
jec
ted
t
o
p
re
p
r
o
c
e
ss
in
g
a
n
d
a
u
g
m
e
n
tatio
n
.
Th
e
su
g
g
e
ste
d
m
o
d
e
l
p
e
rfo
rm
a
n
c
e
is
a
ss
e
ss
e
d
d
e
p
e
n
d
i
n
g
o
n
t
h
e
p
a
ra
m
e
ters
li
k
e
p
re
c
isio
n
,
re
c
a
ll
,
m
AP5
0
a
n
d
m
AP5
0
-
9
5
.
K
ey
w
o
r
d
s
:
B
r
ain
tu
m
o
r
Data
au
g
m
en
tatio
n
Dee
p
lear
n
in
g
al
g
o
r
ith
m
Ma
g
n
etic
r
eso
n
an
ce
im
a
g
e
Pre
p
r
o
ce
s
s
in
g
Seg
m
en
tatio
n
YOL
Ov
8
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Sh
ilp
a
C
h
ip
p
alak
atti
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
C
o
m
m
u
n
icatio
n
,
Sir
M
Vis
v
e
s
v
ar
ay
a
I
n
s
titu
te
o
f
T
ec
h
n
o
lo
g
y
Af
f
iliated
to
Vis
v
esv
ar
ay
a
T
e
ch
n
o
lo
g
ical
Un
iv
er
s
ity
Kr
is
h
n
ad
ev
ar
ay
a
n
ag
ar
,
Hu
n
as
em
ar
an
ah
alli,
Ye
lah
an
k
a,
B
en
g
alu
r
u
,
Kar
n
atak
a,
I
n
d
ia
E
m
ail: sh
ilp
ac
h
ip
p
alak
atti_
ec
@
s
ir
m
v
it.e
d
u
1.
I
NT
RO
D
UCT
I
O
N
Ma
n
y
d
iv
er
s
e
co
g
n
itiv
e,
s
e
n
s
o
r
y
,
an
d
m
o
to
r
p
r
o
ce
s
s
es
th
at
ar
e
ess
en
tial
to
h
u
m
a
n
life
an
d
ex
p
er
ien
ce
a
r
e
d
o
n
e
b
y
h
u
m
an
b
r
ain
,
an
in
c
r
ed
ib
ly
co
m
p
licated
an
d
co
m
p
lex
o
r
g
an
[
1
]
,
[
2
]
.
T
h
e
h
u
m
an
an
ato
m
y
co
n
s
is
ts
o
f
m
illi
o
n
s
o
f
ce
lls
,
an
d
as p
er
o
u
r
u
n
d
er
s
tan
d
in
g
o
f
b
io
lo
g
y
,
ce
lls
h
av
e
th
e
n
atu
r
al
ab
ilit
y
to
p
r
o
life
r
ate,
g
r
o
w,
a
n
d
d
i
v
id
e
t
o
cr
ea
te
n
ew
ce
lls
.
Ho
wev
er
,
ce
r
tain
ex
tr
in
s
ic
f
ac
to
r
s
ca
n
d
is
r
u
p
t
th
is
b
alan
ce
,
lead
in
g
to
u
n
co
n
t
r
o
lled
ce
ll
g
r
o
wth
an
d
th
e
f
o
r
m
atio
n
o
f
tu
m
o
r
s
.
T
h
e
r
is
in
g
m
o
r
tality
r
at
es
in
th
e
in
f
an
ts
an
d
ad
u
lts
,
an
d
s
p
ec
if
ically
in
th
e
eld
er
ly
is
d
u
e
to
th
e
ex
is
ten
c
e
o
f
tu
m
o
r
.
As
claim
ed
b
y
th
e
GL
OB
O
C
ON
th
e
b
r
ain
t
u
m
o
r
ca
s
es
was
3
0
8
,
1
0
2
an
d
2
.
5
%
o
f
p
eo
p
le
d
ied
with
th
e
tu
m
o
r
[
3
]
,
[
4
]
.
T
u
m
o
r
s
c
an
b
e
ass
o
r
ted
i
n
to
two
k
in
d
s
:
B
en
ig
n
tu
m
o
r
s
d
o
n
o
t
s
p
r
ea
d
to
o
th
er
ce
lls
,
m
a
k
in
g
th
em
n
o
n
-
ca
n
ce
r
o
u
s
an
d
co
m
p
ar
ativ
ely
less
d
an
g
er
o
u
s
.
Ma
lig
n
an
t
tu
m
o
r
s
,
o
n
th
e
o
th
er
h
an
d
,
ar
e
m
ad
e
u
p
o
f
ca
n
ce
r
o
u
s
ce
lls
,
ar
e
ex
t
r
em
ely
d
an
g
e
r
o
u
s
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
I
d
en
tifi
ca
tio
n
a
n
d
s
eg
men
t
a
tio
n
o
f tu
mo
r
u
s
in
g
d
ee
p
lea
r
n
i
n
g
a
n
d
ima
g
e
… (
S
h
ilp
a
C
h
ip
p
a
la
ka
tti
)
1783
an
d
ar
e
m
o
r
e
lik
ely
to
s
p
r
ea
d
to
n
ea
r
b
y
ce
lls
an
d
tis
s
u
e.
E
s
s
en
tially
,
a
tu
m
o
r
is
an
ac
cu
m
u
latio
n
o
f
c
ells
th
at
f
o
r
m
a
tis
s
u
e,
b
u
t
u
n
lik
e
h
ea
lth
y
ce
lls
,
th
ey
lack
r
eg
u
latio
n
o
f
g
r
o
wth
a
n
d
ex
h
i
b
it
an
u
n
co
n
tr
o
lled
p
r
o
life
r
atio
n
r
ate
[
5
]
.
T
h
e
ea
r
ly
ex
p
o
s
u
r
e
o
f
th
e
tu
m
o
r
is
v
er
y
im
p
o
r
ta
n
t
to
d
ec
r
ea
s
e
th
e
m
o
r
tality
r
ate.
Ma
n
u
al
d
etec
tio
n
is
v
er
y
ted
io
u
s
,
tim
e
co
n
s
u
m
in
g
an
d
r
e
q
u
ir
es
ex
p
e
r
tis
e
in
i
t
s
o
a
u
to
m
ated
m
eth
o
d
s
is
p
r
ef
er
r
e
d
.
T
h
e
tr
ad
itio
n
al
m
eth
o
d
f
o
r
th
e
d
iag
n
o
s
es o
f
th
e
b
r
ain
tu
m
o
r
o
v
er
th
e
d
ec
ad
es is
ML
b
ased
tech
n
iq
u
es
lik
e
Su
p
er
v
is
ed
lear
n
i
n
g
(
ANN,
SVM
[
6
]
,
DT
,
KNN
a
n
d
L
VQ)
an
d
Un
s
u
p
e
r
v
is
ed
lear
n
i
n
g
(
FC
M,
K
-
Me
an
s
,
SOM,
PC
NN,
FP
C
NN,
Hi
er
ar
ch
ical
C
lu
s
ter
in
g
)
an
d
d
ee
p
le
ar
n
in
g
(
DL
)
b
ased
tech
n
iq
u
es
lik
e
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN)
[
7
]
,
d
ee
p
c
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
DC
NN)
,
VGG1
9
[
8
]
,
an
d
a
u
to
-
en
c
o
d
er
tech
n
iq
u
es
.
No
w
-
a
-
d
ay
s
,
DL
tech
n
iq
u
es
ar
e
m
o
r
e
ex
p
lo
r
ed
b
y
th
e
r
esear
ch
er
d
u
e
to
its
f
ast
an
d
ac
cu
r
ately
d
etec
tio
n
r
ates.
On
e
o
f
th
e
v
a
r
io
u
s
DL
tech
n
iq
u
es,
y
o
u
o
n
ly
lo
o
k
o
n
ce
(
YOL
O)
,
is
u
tili
ze
d
f
o
r
r
ea
l
-
tim
e
d
ata
r
ec
o
g
n
itio
n
with
ex
ce
llen
t
a
cc
u
r
ac
y
ev
en
in
c
o
m
p
licate
d
s
ettin
g
s
,
m
ak
in
g
it
ap
p
r
o
p
r
iate
f
o
r
m
ag
n
etic
r
eso
n
an
ce
im
ag
in
g
(
MRI
)
im
ag
e
an
aly
s
is
[
9
]
.
T
h
er
e
ar
e
d
if
f
er
en
t
v
er
s
io
n
s
o
f
YOL
O
alg
o
r
ith
m
YOL
Ov
2
,
YOL
Ov
3
,
YOL
Ov
4
,
YOL
Ov
5
,
YOL
Ov
6
,
YOL
Ov
7
an
d
r
e
ce
n
tly
YOL
Ov
8
.
T
h
ese
en
tire
alg
o
r
ith
m
s
ca
n
b
e
u
s
ed
f
o
r
d
etec
tio
n
o
f
tu
m
o
r
b
u
t
s
till
th
er
e
i
s
a
lacu
n
a
in
s
y
s
te
m
atic
an
d
f
ast
d
etec
ti
o
n
.
As
YOL
Ov
8
alg
o
r
ith
m
is
a
im
p
r
o
v
is
ed
v
er
s
io
n
o
f
r
e
m
ain
in
g
m
o
d
els
it
is
u
s
ed
to
a
ch
iev
e
b
etter
r
esu
lts
co
m
p
ar
e
d
to
o
th
er
s
[
1
0
]
.
T
h
e
co
r
e
o
f
th
e
YOL
Ov
8
m
o
d
el
is
th
e
C
SP
Dar
k
n
et5
3
f
ea
tu
r
e
ex
tr
ac
to
r
,
wh
ich
is
f
o
llo
wed
b
y
th
e
C
2
f
m
o
d
u
le
in
p
lace
o
f
t
h
e
co
n
v
en
tio
n
al
Y
OL
O
n
ec
k
ar
c
h
itectu
r
e.
T
h
is
h
elp
s
with
o
b
ject
r
ec
o
g
n
itio
n
th
at
is
q
u
ick
er
an
d
m
o
r
e
p
r
ec
is
e
[
1
1
]
,
[
1
2
]
.
H
o
s
p
itals
s
to
r
e
th
e
p
atien
t’
s
in
f
o
r
m
atio
n
in
th
e
clo
u
d
to
m
ain
tain
th
e
p
atien
t
m
ed
ical
h
is
to
r
y
[
1
3
]
.
C
o
n
tin
u
o
u
s
s
to
r
ag
e
o
f
all
th
e
m
ed
ical
d
ata
i
n
to
th
e
clo
u
d
m
ay
in
cr
ea
s
e
th
e
s
to
r
ag
e
c
o
s
t
[
1
4
]
.
I
m
ag
e
s
eg
m
e
n
tatio
n
m
i
g
h
t
b
e
u
s
ed
to
ef
f
ec
tiv
ely
r
ed
u
ce
th
e
clo
u
d
s
to
r
ag
e
s
p
ac
e
o
f
m
ed
ical
im
ag
es.
I
m
ag
e
s
eg
m
en
tatio
n
is
d
elin
ea
tio
n
o
f
tu
m
o
r
r
eg
io
n
o
n
m
ed
ical
im
a
g
es
u
s
in
g
an
y
alg
o
r
ith
m
[
1
5
]
.
T
h
u
s
,
th
e
s
cie
n
tific
g
o
al
o
f
th
is
in
v
esti
g
atio
n
is
to
ex
p
lo
r
e
th
e
YOL
Ov
8
m
o
d
el
f
o
r
b
r
ain
ca
n
ce
r
d
etec
tio
n
an
d
s
eg
m
en
t
th
e
in
f
ec
ted
ar
ea
u
s
in
g
im
a
g
e
s
eg
m
en
tatio
n
alg
o
r
ith
m
.
Fu
r
th
er
,
s
eg
m
en
ted
im
ag
e
will
b
e
s
av
e
d
in
th
e
clo
u
d
u
s
in
g
p
atien
t
id
an
d
l
o
ca
tio
n
o
f
im
ag
e
to
s
av
e
th
e
clo
u
d
s
p
ac
e
an
d
in
tu
r
n
co
s
t
o
f
s
to
r
ag
e
w
ill
b
e
r
ed
u
ce
d
.
T
h
e
p
u
r
s
u
an
ce
o
f
th
e
p
r
o
to
ty
p
e
is
ev
alu
ated
b
ased
o
n
d
ataset
u
s
ed
,
n
u
m
b
er
o
f
ep
o
ch
s
it
tak
e
s
,
p
r
ec
is
io
n
,
r
ec
all,
m
AP5
0
,
m
AP5
0
-
9
5
an
d
im
ag
e
s
ize
af
ter
s
eg
m
en
tatio
n
.
Mu
h
am
m
ad
I
r
f
a
n
Sh
ar
i
f
et
al
.
[
1
6
]
u
s
ed
lesi
o
n
e
n
h
an
ce
m
e
n
t,
f
ea
tu
r
e
ex
tr
ac
tio
n
a
n
d
s
el
ec
tio
n
f
o
r
class
if
icatio
n
,
lo
ca
lizatio
n
,
an
d
s
eg
m
en
tatio
n
.
I
t
co
m
p
r
is
es
t
h
e
f
o
u
r
s
tag
es
o
f
th
e
p
r
o
p
o
s
e
d
m
o
d
el.
T
o
r
e
d
u
ce
n
o
is
e,
a
h
o
m
o
m
o
r
p
h
ic
wav
el
et
f
ilter
is
em
p
l
o
y
ed
.
T
h
e
Y
OL
Ov
2
-
in
ce
p
tio
n
v
3
m
o
d
el
is
in
ten
d
e
d
to
s
er
v
e
i
n
tu
m
o
r
lo
ca
lizatio
n
.
I
n
ter
m
s
o
f
b
r
ain
lesi
o
n
lo
ca
tio
n
,
s
eg
m
en
tatio
n
,
an
d
class
if
icatio
n
,
th
e
m
o
d
el
p
r
o
d
u
ce
d
p
r
ed
ictio
n
v
alu
es
h
ig
h
er
th
an
0
.
9
0
.
Ok
s
u
z
a
n
d
Gu
llu
[
1
7
]
u
s
ed
YOL
Ov
2
s
in
g
le
s
tag
e
DL
m
o
d
el
f
o
r
th
e
b
r
ain
tu
m
o
r
tis
s
u
e
d
etec
tio
n
.
Ali
et
al
.
[
1
8
]
p
r
o
p
o
s
ed
two
-
s
tag
e
p
a
r
ad
ig
m
f
o
r
m
e
d
ical
im
ag
e
an
a
ly
s
is
.
C
las
s
if
icatio
n
u
s
in
g
th
e
Go
o
g
leNe
t
m
o
d
el
i
s
th
e
f
ir
s
t
s
tep
.
T
u
m
o
r
lo
ca
li
za
tio
n
u
s
in
g
YOL
Ov
3
is
th
e
s
ec
o
n
d
s
tep
.
T
h
e
DI
C
OM
d
ataset
was
u
tili
ze
d
in
th
is
in
s
tan
ce
.
A
to
tal
o
f
1
0
0
0
p
ict
u
r
es
wer
e
u
tili
ze
d
,
o
f
wh
ich
7
0
0
s
h
o
wed
tu
m
o
r
s
an
d
3
0
0
s
h
o
wed
n
o
r
m
als.
9
7
%
o
b
ject
class
if
icatio
n
p
er
f
ec
tio
n
was
attain
ed
u
s
in
g
th
e
Go
o
g
leNe
t
m
o
d
el.
8
1
.
9
%
ac
cu
r
ac
y
o
n
t
r
ain
in
g
d
ataset
an
d
9
4
.
3
%
ac
cu
r
ac
y
o
n
th
e
test
in
g
d
ataset
is
o
b
tain
ed
u
s
in
g
YOL
Ov
3
f
o
r
tu
m
o
r
lo
ca
lizati
o
n
.
T
h
e
YOL
Ov
4
s
m
all
m
o
d
e
l
is
u
s
ed
b
y
th
e
r
esear
ch
er
in
[
1
9
]
to
tr
ain
tu
m
o
r
id
en
tific
atio
n
.
T
h
e
tag
g
ed
im
ag
es
in
th
e
d
ataset
wer
e
r
etr
i
ev
ed
f
r
o
m
th
e
f
ig
s
h
ar
e
d
ata
r
ep
o
s
ito
r
y
.
8
0
:1
0
:1
0
r
atio
o
f
th
e
d
ataset
is
s
et
a
s
id
e
f
o
r
test
in
g
,
v
alid
atio
n
,
an
d
tr
ain
in
g
,
co
r
r
esp
o
n
d
in
g
ly
.
Pre
p
r
o
ce
s
s
in
g
m
eth
o
d
s
wer
e
ap
p
lied
to
r
ef
in
e
th
e
im
ag
e'
s
a
ttrib
u
tes.
Fo
r
r
aw
d
ata
,
th
e
m
o
d
el
y
ield
ed
an
av
er
a
g
e
m
ea
n
p
r
ec
is
io
n
(
m
AP)
o
f
0
.
8
0
7
4
,
wh
er
ea
s
f
o
r
p
r
o
ce
s
s
ed
d
ata,
it
was
0
.
8
3
2
4
.
Usi
n
g
YOL
Ov
5
,
Pau
l
et
a
l.
[
2
0
]
in
v
est
ig
ates
b
r
ain
tu
m
o
r
s
eg
m
e
n
tatio
n
.
T
h
e
B
R
AT
S
2
0
1
8
d
ataset,
wh
ich
in
co
r
p
o
r
ate
1
9
9
2
p
h
o
to
s
,
was
th
e
d
ataset
u
tili
ze
d
f
o
r
th
e
s
tu
d
y
.
T
h
e
p
r
o
to
ty
p
e
was
tr
ain
ed
o
n
7
2
0
p
h
o
to
s
,
an
d
it
was
v
alid
ated
o
n
1
8
0
im
ag
es.
W
ith
an
p
r
ec
is
io
n
o
f
8
5
.
9
5
%,
th
e
m
o
d
el
d
em
o
n
s
tr
ated
g
o
o
d
p
er
f
o
r
m
an
ce
.
Sh
elatk
ar
et
a
l.
[
2
1
]
,
YOL
Ov
5
was
u
s
ed
to
d
etec
t
an
d
ca
teg
o
r
ize
b
r
ain
tu
m
o
r
s
.
T
h
e
B
R
AT
S
2
0
2
1
d
ata
s
et
f
r
o
m
th
e
R
SN
A
-
MI
C
C
AI
b
r
ain
tu
m
o
r
r
ad
i
o
g
en
o
m
ic
ca
teg
o
r
izatio
n
is
u
s
ed
.
T
h
e
m
o
d
el
o
f
f
er
s
an
ac
c
u
r
ac
y
o
f
8
8
%
.
Ab
d
u
s
alo
m
o
v
et
a
l.
[
2
2
]
f
o
r
th
e
p
r
ec
is
e
id
en
tific
atio
n
o
f
p
itu
it
ar
y
g
lan
d
tu
m
o
r
s
,
g
lio
m
as,
an
d
m
en
in
g
io
m
as,
u
s
es
YOL
Ov
7
.
Fo
r
tr
ain
in
g
,
th
e
m
o
d
el
u
s
es
a
p
u
b
lically
ac
ce
s
s
ib
le
d
ataset
th
at
in
clu
d
es
2
5
0
0
p
h
o
t
o
s
o
f
n
o
n
-
tu
m
o
r
o
u
s
ca
s
es,
2
6
5
8
im
a
g
es
o
f
p
itu
itar
y
tu
m
o
r
s
,
2
,
5
8
2
i
m
ag
e
s
o
f
m
e
n
in
g
io
m
as,
an
d
2
5
4
8
i
m
ag
es
o
f
g
lio
m
as.
T
o
en
h
a
n
c
e
YOL
Ov
7
'
s
f
ea
tu
r
e
ex
tr
ac
tio
n
ca
p
ab
ilit
ies,
th
e
co
n
v
o
lu
tio
n
al
b
lo
c
k
atten
tio
n
m
o
d
u
le
(
C
B
AM
)
i
s
in
co
r
p
o
r
ated
,
f
o
cu
s
in
g
o
n
s
alien
t
r
eg
io
n
s
ass
o
ciate
d
with
b
r
ai
n
ca
n
ce
r
.
Ad
d
itio
n
ally
,
t
h
e
m
o
d
el
in
teg
r
ates
t
h
e
s
p
atial
p
y
r
a
m
id
p
o
o
lin
g
f
ast
+
(
SP
PF
+)
lay
er
in
to
its
f
u
n
d
am
en
tal
ar
ch
itectu
r
e
t
o
en
h
an
ce
s
en
s
itiv
ity
.
Fu
r
th
er
m
o
r
e,
th
e
b
id
ir
ec
tio
n
al
f
ea
tu
r
e
p
y
r
am
id
n
etwo
r
k
(
B
iFP
N)
is
em
p
lo
y
e
d
f
o
r
m
u
lti
-
s
ca
le
f
ea
tu
r
e
f
u
s
io
n
,
ef
f
icien
tly
ca
p
tu
r
in
g
tu
m
o
r
-
r
elev
an
t
in
f
o
r
m
atio
n
.
W
ith
th
ese
en
h
an
ce
m
en
ts
,
th
e
p
r
o
p
o
s
ed
m
o
d
e
l
ac
h
iev
es
n
o
tab
le
o
v
er
all
p
r
e
cisi
o
n
co
m
p
ar
ed
to
ex
is
tin
g
m
o
d
els”
.
Pas
s
a
et
a
l.
[
2
3
]
p
r
o
p
o
s
ed
Y
OL
Ov
8
alg
o
r
ith
m
alo
n
g
with
d
ata
au
g
m
e
n
tatio
n
ca
n
d
etec
t
th
e
b
r
ain
tu
m
o
r
(
m
e
n
in
g
io
m
a
,
g
lio
m
a
an
d
p
itu
itar
y
)
e
f
f
i
cien
tly
.
T
h
er
e
ar
e
th
r
ee
ca
teg
o
r
ies
o
f
b
r
ain
tu
m
o
r
s
in
th
e
d
ataset,
wh
ich
co
n
s
is
ts
o
f
3
0
6
1
T
1
-
weig
h
te
d
co
n
tr
ast
-
en
h
an
ce
d
im
ag
es.
T
esti
n
g
,
v
alid
atio
n
,
an
d
tr
ai
n
in
g
s
ets
o
f
d
ata
ar
e
s
ep
ar
ated
ap
ar
t.
4
9
6
tr
ain
in
g
,
1
4
1
v
alid
atio
n
,
an
d
7
1
test
in
g
p
ictu
r
es a
r
e
a
v
ailab
le
f
o
r
m
en
in
g
i
o
m
a.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
7
8
2
-
1
7
9
2
1784
T
h
er
e
ar
e
1
4
3
test
in
g
,
2
8
5
v
al
id
atin
g
,
a
n
d
9
9
8
tr
ain
in
g
im
a
g
es
f
o
r
g
lio
m
as.
1
8
6
v
alid
atin
g
im
ag
es,
8
3
test
in
g
im
ag
es,
an
d
6
5
1
tr
ai
n
in
g
im
ag
es
ar
e
ac
ce
s
s
ib
le
f
o
r
Pit
u
i
tar
y
.
T
h
e
d
ata
p
r
e
p
r
o
ce
s
s
in
g
is
d
o
n
e
lik
e
d
ata
n
o
r
m
aliza
tio
n
,
r
em
o
v
in
g
t
h
e
r
ed
u
n
d
an
t
d
ata
a
n
d
d
ata
c
o
n
v
er
s
io
n
is
d
o
n
e
in
R
o
b
o
f
lo
w.
Data
au
g
m
en
tatio
n
lik
e
f
lip
,
9
0
0
r
o
tate,
cr
o
p
,
r
o
ta
tio
n
,
s
h
ea
r
,
g
r
ay
s
ca
le,
b
r
ig
h
tn
ess
,
ex
p
o
s
u
r
e,
b
lu
r
an
d
n
o
is
e
is
ap
p
lied
.
Yo
lo
v
8
s
o
f
h
y
p
er
p
ar
a
m
eter
co
n
f
ig
u
r
ati
o
n
with
th
e
in
p
u
t size
o
f
6
4
0
×
6
4
0
,
1
0
0
e
p
o
ch
s
an
d
b
atch
s
iz
e
o
f
8
is
th
e
u
tili
ze
d
f
o
r
tr
ain
in
g
th
e
d
ata.
R
esear
ch
er
an
aly
ze
d
p
u
r
s
u
an
ce
o
f
th
e
m
o
d
el
f
o
r
b
o
th
c
o
n
d
itio
n
with
d
ata
au
g
m
en
tatio
n
an
d
with
o
u
t
d
ata
a
u
g
m
en
tatio
n
.
T
h
e
m
o
d
el
p
er
f
o
r
m
s
b
etter
with
au
g
m
en
tatio
n
y
ield
in
g
a
p
r
ec
is
io
n
as
0
.
9
4
2
,
r
ec
all
as 0
.
9
0
8
,
m
AP5
0
as 0
.
9
5
2
an
d
m
AP5
0
-
9
5
as 0
.
7
3
3
.
Me
r
ca
ld
o
et
a
l.
[
2
4
]
also
u
s
ed
Y
OL
Ov
8
s
alg
o
r
ith
m
f
o
r
t
h
e
d
is
clo
s
u
r
e
o
f
th
e
b
r
ain
ca
n
ce
r
.
T
h
e
r
ea
l
-
wo
r
ld
d
ata
u
tili
ze
d
in
t
h
is
s
tu
d
y
c
am
e
f
r
o
m
a
s
o
u
r
ce
th
at
is
o
p
en
ly
ac
ce
s
s
ib
le
f
o
r
ac
ad
em
i
c
u
s
ag
e.
T
h
er
e
ar
e
3
0
0
b
r
ain
MRIs
in
to
tal,
o
f
wh
ich
2
1
0
a
r
e
u
s
ed
f
o
r
tr
ain
in
g
,
6
0
f
o
r
v
alid
atio
n
,
a
n
d
3
0
f
o
r
test
in
g
.
T
h
e
d
ataset
in
clu
d
e
s
a
v
ar
iety
o
f
tu
m
o
r
f
o
r
m
s
,
i
n
clu
d
in
g
p
itu
itar
y
,
g
lio
m
a,
an
d
m
en
i
n
g
io
m
a.
T
h
e
im
ag
es
wer
e
r
esized
to
5
1
2
×
5
1
2
p
ix
els,
th
e
n
u
m
b
e
r
o
f
ep
o
ch
s
u
s
ed
is
5
0
a
n
d
b
atch
s
ize
is
1
6
.
T
h
e
d
ata
a
u
g
m
en
tatio
n
m
et
h
o
d
s
lik
e
9
0
0
cl
o
ck
wis
e
,
9
0
0
co
u
n
ter
clo
ck
wis
e
an
d
u
p
s
id
e
d
o
w
n
is
ap
p
lied
.
T
h
e
m
o
d
els
p
er
f
o
r
m
an
ce
is
ev
al
u
ated
b
ased
o
n
th
e
p
r
ec
is
io
n
(
0
.
9
4
3
)
,
r
ec
a
ll
(
0
.
9
3
2
)
,
m
AP5
0
(
0
.
9
4
1
)
,
s
p
ec
if
icity
(
0
.
9
3
8
)
an
d
m
AP5
0
-
95
(
0
.
4
2
1
)
.
Vin
ee
la
et
al
.
[
2
5
]
u
s
ed
YOL
Ov
8
alg
o
r
ith
m
to
d
etec
t
th
e
b
r
ain
tu
m
o
r
.
T
h
e
y
h
a
v
e
u
s
ed
f
r
ee
ly
av
ailab
le
d
ataset
wh
ich
co
n
s
is
t
o
f
1
9
2
3
im
ag
es
o
u
t
o
f
th
at
8
7
.
5
%
o
f
im
ag
es
ar
e
m
ea
n
t
f
o
r
tr
ai
n
in
g
,
8
.
3
%
ar
e
m
ea
n
t
f
o
r
v
alid
a
tio
n
an
d
4
.
2
%
f
o
r
test
in
g
.
T
h
e
p
u
r
s
u
an
ce
o
f
th
e
m
o
d
el
is
ev
al
u
ated
u
s
in
g
th
e
p
ar
am
eter
p
r
ec
is
io
n
,
r
ec
all,
m
A
P5
0
,
m
AP5
0
-
9
5
,
b
o
x
lo
s
s
,
class
lo
s
s
an
d
d
f
l
lo
s
s
.
Pre
d
icted
th
e
tu
m
o
r
with
a
a
cc
u
r
ac
y
o
f
9
6
.
4
%.
Acc
o
r
d
in
g
to
Hash
em
i
et
a
l.
[
2
6
]
t
he
d
ata
ef
f
icien
t
im
ag
e
tr
an
s
f
o
r
m
er
m
o
d
el
(
DeiT
)
a
n
d
v
is
io
n
tr
a
n
s
f
o
r
m
e
r
m
o
d
els
f
r
o
m
a
f
in
e
-
t
u
n
ed
R
esNet1
5
2
as
a
teac
h
er
in
th
e
class
if
icatio
n
p
h
ase
ca
n
im
p
r
o
v
e
YOL
Ov
8
n
p
er
f
o
r
m
an
ce
.
T
h
ey
m
ad
e
a
d
v
an
ta
g
e
o
f
th
e
n
atio
n
al
b
r
ain
m
ap
p
in
g
lab
(
NB
ML
)
,
w
h
ich
h
as
8
1
p
atien
ts
,
3
0
o
f
wh
o
m
h
a
v
e
tu
m
o
r
s
an
d
5
1
o
f
wh
o
m
ar
e
h
ea
lth
y
.
C
h
en
et
a
l.
[
2
7
]
p
h
o
to
ac
o
u
s
tic
im
ag
in
g
(
PAI
)
is
u
s
ed
in
s
te
ad
o
f
MRI
im
ag
es
an
d
to
cla
s
s
if
y
-
d
etec
t
b
en
ig
n
tu
m
o
r
an
d
m
alig
n
an
t
tu
m
o
r
YOL
Ov
8
-
Me
d
SAM
m
o
d
el
is
u
s
ed
.
R
en
et
a
l.
[
2
8
]
m
u
ltis
ca
le
d
ilated
atten
tio
n
an
d
m
u
lti
-
h
ea
d
s
elf
atten
tio
n
ar
e
in
teg
r
ated
in
s
id
e
th
e
YO
L
Ov
8
n
etwo
r
k
in
t
h
e
p
r
o
p
o
s
ed
en
h
a
n
ce
d
lesi
o
n
d
etec
tio
n
m
o
d
el
DHC
-
YOL
O.
Fo
r
im
p
r
o
v
e
d
f
ea
t
u
r
es,
it
also
in
co
r
p
o
r
ates
th
e
f
ea
tu
r
e
p
y
r
am
id
n
etwo
r
k
.
T
h
e
d
atasets
f
o
r
eso
p
h
ag
ea
l
ca
n
ce
r
,
co
lo
n
ic
p
o
ly
p
s
,
an
d
b
r
ain
t
u
m
o
r
s
ar
e
u
s
ed
to
ev
alu
ate
t
h
is
ap
p
r
o
ac
h
.
T
h
e
ac
cu
r
ac
y
o
f
th
e
b
r
ai
n
tu
m
o
r
d
a
taset wa
s
8
8
.
3
%.
Ho
wev
er
,
af
te
r
a
th
o
r
o
u
g
h
r
ev
iew
o
f
th
e
liter
atu
r
e,
we
c
an
co
n
clu
d
e
th
at
YOL
Ov
8
m
o
d
el
m
a
y
p
r
o
v
id
e
a
h
ig
h
er
lev
el
o
f
p
er
f
ec
tn
ess
in
tu
m
o
r
d
etec
tio
n
th
en
YOL
Ov
2
,
YOL
Ov
3
,
YOL
Ov
4
,
YOL
Ov
5
an
d
YOL
Ov
7
with
o
u
t
an
y
ad
d
itio
n
al
en
h
a
n
ce
r
an
d
f
ea
tu
r
e
ex
t
r
a
cto
r
m
o
d
el,
w
h
ich
m
ig
h
t
s
lo
w
d
o
wn
p
er
f
o
r
m
an
ce
.
I
n
all
th
e
ab
o
v
e
ar
ticles,
th
e
r
esear
ch
er
h
as
o
n
ly
co
n
ce
n
tr
at
ed
o
n
th
e
lo
ca
lizatio
n
,
d
etec
tio
n
an
d
class
if
icatio
n
o
f
b
r
ain
t
u
m
o
r
b
u
t sto
r
in
g
im
a
g
es in
clo
u
d
with
th
e
co
m
p
r
es
s
ed
f
o
r
m
at
is
n
o
t d
is
c
u
s
s
ed
.
I
n
p
r
o
p
o
s
ed
r
esear
ch
,
YOL
Ov
8
is
u
s
ed
f
o
r
p
r
ec
is
e
d
etec
tio
n
o
f
tu
m
o
r
an
d
im
ag
e
s
eg
m
en
tatio
n
is
d
o
n
e
to
r
e
d
u
ce
th
e
d
ata
s
to
r
ag
e
in
clo
u
d
.
T
h
is
r
esear
ch
p
ap
er
is
f
o
r
m
u
lated
as:
s
ec
tio
n
1
c
o
n
ta
in
s
in
tr
o
d
u
ctio
n
an
d
s
u
r
v
ey
o
f
th
e
ex
is
tin
g
wo
r
k
,
s
ec
tio
n
2
co
n
tain
s
m
eth
o
d
s
an
d
im
p
lem
en
tatio
n
,
s
ec
tio
n
3
co
n
tain
s
r
esu
lt
an
d
d
is
cu
s
s
io
n
s
an
d
s
ec
tio
n
4
co
n
tain
s
co
n
clu
s
io
n
an
d
f
u
tu
r
e
s
co
p
e.
2.
M
E
T
H
O
DS A
ND
I
M
P
L
E
M
E
NT
AT
I
O
N
T
h
e
tu
m
o
r
'
s
s
ize,
f
o
r
m
,
an
d
p
r
o
p
e
r
lo
ca
tio
n
ca
n
all
f
lu
ctu
ate,
m
ak
in
g
m
an
u
al
a
n
al
y
s
is
q
u
ite
ch
allen
g
in
g
f
r
o
m
th
e
MRI
im
ag
e.
T
o
aid
th
e
m
ed
ical
co
m
m
u
n
ity
in
ac
cu
r
ately
an
d
c
o
n
v
en
ien
tly
d
iag
n
o
s
e
tu
m
o
r
an
d
s
eg
m
en
t th
e
tu
m
o
r
r
eg
io
n
th
e
p
r
o
p
o
s
ed
m
eth
o
d
m
ig
h
t b
e
u
s
ed
.
T
h
e
s
u
g
g
ested
ap
p
r
o
ac
h
u
tili
ze
s
th
e
m
o
s
t
r
ec
en
t
v
er
s
io
n
YOL
Ov
8
s
m
o
d
el
to
i
d
en
tify
tu
m
o
r
,
wh
ich
ar
e
s
eg
m
en
ted
u
s
in
g
i
m
ag
e
s
eg
m
en
tatio
n
tech
n
iq
u
es
a
n
d
th
e
s
eg
m
e
n
ted
im
ag
e
is
s
to
r
ed
in
t
h
e
clo
u
d
u
s
in
g
th
e
p
atien
t
id
an
d
p
o
s
iti
o
n
o
f
th
e
t
u
m
o
r
f
o
r
th
e
f
u
tu
r
e
r
ef
er
e
n
ce
.
B
ef
o
r
e
ap
p
ly
in
g
YOL
Ov
8
s
m
o
d
el
im
ag
e
p
r
ep
r
o
ce
s
s
in
g
is
d
o
n
e
wh
ich
in
clu
d
es
co
n
v
er
s
io
n
o
f
MRI
im
ag
e
in
to
th
e
.
y
am
l f
ile
wh
ich
is
ap
p
r
o
p
r
iate
f
o
r
th
e
m
o
d
el,
r
esizin
g
th
e
im
ag
e,
r
em
o
v
i
n
g
r
ep
ea
ted
im
ag
es,
au
to
o
r
ien
tat
io
n
an
d
d
ata
a
u
g
m
e
n
tatio
n
.
T
h
e
ef
f
ec
tiv
e
e
v
alu
atio
n
o
f
t
h
e
p
r
o
to
ty
p
e
is
d
o
n
e
b
y
u
s
in
g
th
e
m
etr
ics
lik
e
p
r
ec
is
io
n
,
r
ec
all,
m
AP5
0
,
m
AP5
0
-
9
5
.
Fig
u
r
e
1
d
ep
icts
th
e
b
lo
ck
d
iag
r
am
f
o
r
th
e
b
r
ai
n
tu
m
o
r
id
e
n
tific
atio
n
u
s
in
g
YO
L
Ov
8
s
.
2
.
1
.
M
a
g
net
ic
re
s
o
na
nce
ima
g
e
A
n
o
n
-
in
v
asiv
e
im
ag
in
g
tech
n
o
lo
g
y
ca
p
ab
le
o
f
g
e
n
er
atin
g
d
etailed
th
r
ee
-
d
im
en
s
io
n
al
a
n
ato
m
ical
im
ag
es
is
co
m
m
o
n
ly
em
p
lo
y
e
d
f
o
r
d
is
ea
s
e
d
etec
tio
n
,
d
iag
n
o
s
is
,
tr
ea
tm
en
t
an
d
m
o
n
ito
r
i
n
g
.
T
h
is
tech
n
o
l
o
g
y
r
elies
o
n
ad
v
a
n
ce
d
m
ec
h
an
is
m
s
to
s
tim
u
late
an
d
d
etec
t
alt
er
atio
n
s
in
th
e
r
o
tatio
n
al
ax
is
o
f
p
r
o
to
n
s
with
in
th
e
wate
r
co
n
s
titu
tin
g
liv
in
g
tis
s
u
es.
T
h
is
is
b
est
s
u
ited
to
d
e
tect
tu
m
o
r
b
ec
au
s
e
th
ey
c
r
ea
t
e
im
ag
e
with
h
i
g
h
r
eso
lu
tio
n
th
at
clea
r
l
y
s
h
o
w
t
h
e
b
r
ain
s
tr
u
ct
u
r
e,
s
ize
an
d
lo
ca
tio
n
.
So
h
e
r
e
we
ar
e
u
s
in
g
MRI
im
ag
es
to
d
etec
t
tu
m
o
r
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
I
d
en
tifi
ca
tio
n
a
n
d
s
eg
men
t
a
tio
n
o
f tu
mo
r
u
s
in
g
d
ee
p
lea
r
n
i
n
g
a
n
d
ima
g
e
… (
S
h
ilp
a
C
h
ip
p
a
la
ka
tti
)
1785
Fi
g
u
r
e
1
.
B
r
ain
tu
m
o
r
d
etec
tio
n
u
s
in
g
YOL
Ov
8
s
m
o
d
el
2
.
2
.
Da
t
a
s
et
T
h
e
d
ataset
u
s
ed
in
p
r
o
p
o
s
e
d
m
eth
o
d
is
f
r
ee
ly
av
ailab
le
o
p
en
-
s
o
u
r
ce
d
ataset
wh
ich
is
th
eir
f
o
r
-
r
esear
ch
p
u
r
p
o
s
e,
th
e
lin
k
o
f
t
h
e
d
ataset
is
g
iv
en
h
er
e
[
2
9
]
.
T
h
e
d
ataset
co
n
s
is
ts
o
f
6
3
9
i
m
ag
es
wh
ich
co
n
s
is
t
o
f
tu
m
o
r
o
u
s
,
n
o
n
-
tu
m
o
r
o
u
s
,
a
n
d
also
v
ar
io
u
s
tu
m
o
r
ty
p
es
lik
e
m
en
i
n
g
io
m
a,
g
lio
m
a
an
d
p
i
tu
itar
y
.
Fo
r
tr
ain
in
g
th
e
m
o
d
el
th
e
d
ata
is
s
p
lit
in
t
o
tr
ain
in
g
,
v
alid
atio
n
an
d
test
in
g
.
I
n
o
u
r
d
ataset
we
h
a
v
e
ta
k
en
4
5
3
im
ag
es
f
o
r
tr
ain
in
g
,
1
2
2
f
o
r
v
ali
d
atio
n
an
d
6
4
f
o
r
test
in
g
in
a
r
atio
o
f
7
1
:1
9
:1
0
.
T
h
e
im
ag
e
s
ize
is
tak
en
as
6
4
0
×
6
4
0
.
On
th
e
d
ataset,
im
ag
e
p
r
ep
r
o
ce
s
s
in
g
lik
e
d
ata
co
n
v
er
s
io
n
to
th
e
s
u
it
th
e
f
o
r
m
at
o
f
YOL
Ov
8
s
,
r
em
o
v
in
g
r
ep
ea
ted
im
ag
es,
r
esizin
g
th
e
im
ag
e
a
n
d
g
r
ey
s
ca
le
co
n
v
er
s
io
n
is
ap
p
l
ied
u
s
in
g
R
o
b
o
f
lo
w
a
p
p
licatio
n
.
T
h
e
n
ex
t
s
tep
is
d
ata
au
g
m
e
n
tatio
n
.
2
.
3
.
Da
t
a
a
ug
m
ent
a
t
io
n
A
ML
o
r
DL
m
o
d
el'
s
ef
f
icac
y
is
h
ea
v
ily
r
ely
in
g
o
n
q
u
ality
,
q
u
an
tity
,
an
d
r
elev
a
n
ce
o
f
t
r
ain
in
g
d
ata.
T
h
e
m
ajo
r
ch
allen
g
e
in
im
p
le
m
en
tin
g
m
ac
h
in
e
lear
n
in
g
ap
p
licati
o
n
is
th
e
co
s
t
an
d
t
im
e
tak
en
to
co
llect
th
e
d
ata.
Au
g
m
en
tatio
n
m
eth
o
d
s
wer
e
d
ev
elo
p
ed
to
an
s
wer
th
i
s
p
r
o
b
lem
.
T
o
en
lar
g
e
th
e
s
ize
o
f
d
ata
ar
tific
ially
an
d
to
cr
ea
te
a
n
ew
d
ataset
with
in
ex
is
tin
g
o
n
e
th
e
au
g
m
en
tat
io
n
m
eth
o
d
is
u
s
ed
.
T
h
e
m
o
tiv
e
o
f
th
is
r
ap
id
a
n
d
ef
f
ec
tiv
e
ap
p
r
o
a
ch
is
to
en
h
an
ce
a
m
o
d
el'
s
ca
p
ac
ity
to
in
d
u
ce
,
n
o
v
el,
u
n
s
ee
n
s
am
p
les
b
y
d
iv
er
s
if
y
in
g
th
e
tr
ain
in
g
d
ata.
I
n
v
ar
i
o
u
s
ar
ea
s
o
f
r
esear
ch
,
in
clu
d
in
g
s
ig
n
al
p
r
o
ce
s
s
in
g
,
co
m
p
u
ter
v
is
io
n
,
s
p
ee
ch
p
r
o
ce
s
s
in
g
,
an
d
n
atu
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
,
au
g
m
en
tatio
n
is
b
ec
o
m
in
g
m
o
r
e
p
o
p
u
la
r
.
Ap
p
r
o
ac
h
es
s
u
ch
as
n
o
is
e
a
d
d
itio
n
,
d
ata
r
o
tatio
n
an
d
s
ca
lin
g
in
ten
tio
n
ally
a
u
g
m
en
t
th
e
d
ataset
s
ize.
Ad
d
itio
n
ally
,
m
o
d
if
icatio
n
s
s
u
ch
as
z
o
o
m
in
g
,
h
o
r
izo
n
tal
o
r
v
e
r
tical
f
lip
p
in
g
,
a
n
d
b
r
ig
h
tn
ess
ad
ju
s
tm
en
ts
co
n
tr
ib
u
t
e
to
en
lar
g
i
n
g
p
ictu
r
es.
B
y
em
p
lo
y
in
g
th
ese
m
eth
o
d
s
,
d
ata
au
g
m
en
tatio
n
ef
f
ec
tiv
ely
in
cr
ea
s
es
th
e
d
im
en
s
io
n
ality
o
f
th
e
t
r
ain
in
g
d
ata,
th
er
eb
y
in
c
r
ea
s
in
g
th
e
p
e
r
f
o
r
m
a
n
ce
an
d
f
lex
i
b
ilit
y
o
f
ML
an
d
DL
m
o
d
els.
Op
er
atio
n
s
lik
e
r
o
tatio
n
-
15
0
t
o
+1
5
0
,
g
r
ay
s
ca
le
u
p
to
1
5
%
a
n
d
n
o
is
e
ad
d
itio
n
is
u
s
ed
o
n
th
is
d
ataset.
2
.
4
.
YO
L
O
v
8
YOL
Ov
8
,
th
e
latest
an
d
m
o
s
t
s
o
p
h
is
ticated
YOL
O
m
o
d
el,
m
ay
b
e
u
s
ed
f
o
r
a
p
p
licatio
n
s
in
clu
d
in
g
o
b
ject
d
etec
tio
n
,
in
s
tan
ce
s
eg
m
en
tatio
n
,
a
n
d
im
a
g
e
class
if
icatio
n
.
YOL
Ov
8
was
cr
ea
ted
b
y
Ultr
aly
tics
,
th
e
s
am
e
f
ir
m
th
at
cr
ea
ted
th
e
p
o
p
u
lar
a
n
d
in
d
u
s
tr
y
-
d
e
f
in
in
g
YOL
Ov
2
,
YOL
Ov
3
,
Y
OL
Ov
4
,
YOL
Ov
5
,
YOL
Ov
6
,
an
d
YOL
Ov
7
m
o
d
els.
Kee
p
in
g
th
e
g
e
n
er
al
s
tr
u
c
tu
r
e
as
YOL
Ov
5
,
YOL
Ov
8
m
ak
es
m
o
d
if
icatio
n
s
to
C
SP
L
ay
er
,
wh
ich
is
also
ca
lled
as
C
2
f
m
o
d
u
le.
"Cro
s
s
-
s
tag
e
p
ar
tial
b
o
ttlen
ec
k
with
two
co
n
v
o
lu
tio
n
s
,
"
a
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
7
8
2
-
1
7
9
2
1786
th
is
m
o
d
u
le
is
s
h
o
r
t
ca
lled
,
ef
f
icien
tly
co
m
b
i
n
es
co
n
tex
tu
al
in
f
o
r
m
atio
n
with
h
ig
h
-
le
v
e
l
ch
ar
ac
ter
is
tics
to
in
cr
ea
s
e
d
etec
tio
n
ac
cu
r
ac
y
.
T
h
e
an
ch
o
r
-
f
r
ee
m
o
d
el
o
f
Y
OL
Ov
8
,
wh
ich
h
as
a
d
ec
o
u
p
l
ed
h
ea
d
an
d
allo
ws
e
ac
h
b
r
an
ch
to
f
o
cu
s
o
n
its
o
wn
jo
b
,
im
p
r
o
v
es
o
v
er
all
ac
cu
r
ac
y
b
y
a
d
d
r
ess
in
g
o
b
jectn
es
s
,
class
if
icatio
n
,
an
d
r
eg
r
ess
io
n
task
s
in
d
iv
id
u
ally
.
Y
O
L
O
v
8
c
o
m
p
u
t
e
s
o
b
je
c
t
n
es
s
s
c
o
r
e
,
o
r
p
r
o
b
a
b
i
l
i
t
y
o
f
a
n
o
b
j
e
c
t
b
e
i
n
g
p
r
e
s
e
n
t
w
i
t
h
i
n
a
b
o
u
n
d
i
n
g
b
o
x
,
u
s
i
n
g
t
h
e
s
i
g
m
o
i
d
f
u
n
c
t
i
o
n
.
F
o
r
c
l
a
s
s
p
r
o
b
a
b
i
l
i
ti
es
,
i
n
d
i
c
a
t
in
g
t
h
e
l
i
k
e
l
i
h
o
o
d
o
f
a
n
i
t
e
m
t
o
b
e
o
w
n
e
d
b
y
e
a
c
h
c
l
a
s
s
,
t
h
e
“
s
o
f
t
m
a
x
f
u
n
c
t
i
o
n
”
is
u
t
i
l
i
ze
d
.
Y
O
L
O
v
8
e
n
h
a
n
c
e
s
d
e
t
e
c
t
i
o
n
e
f
f
i
c
i
e
n
c
y
,
es
p
e
c
ia
l
l
y
f
o
r
s
m
a
l
l
o
b
j
e
c
ts
,
b
y
i
n
t
e
g
r
a
t
i
n
g
d
i
s
t
a
n
c
e
f
e
a
t
u
r
e
le
v
e
l
(
D
F
L
)
a
n
d
c
o
m
p
l
e
t
e
i
n
te
r
s
e
c
ti
o
n
o
v
e
r
u
n
i
o
n
(
C
I
o
U
)
l
o
s
s
f
u
n
c
t
i
o
n
s
f
o
r
b
o
u
n
d
i
n
g
b
o
x
l
o
s
s
.
A
d
d
i
t
i
o
n
al
ly
,
“
b
i
n
a
r
y
c
r
o
s
s
-
e
n
t
r
o
p
y
”
i
s
e
n
f
o
r
c
e
d
t
o
m
i
t
i
g
a
t
e
c
l
ass
i
f
i
c
at
i
o
n
l
o
s
s
.
Fu
r
th
er
m
o
r
e,
YOL
Ov
8
p
r
esen
ts
YOL
Ov
8
-
Seg
,
a
s
em
an
tic
s
eg
m
en
tatio
n
m
o
d
el.
T
h
is
m
o
d
el
d
ep
ar
ts
f
r
o
m
th
e
co
n
v
e
n
tio
n
al
YOL
O
n
ec
k
d
esig
n
b
y
h
a
v
in
g
a
C
2
f
m
o
d
u
le
an
d
a
C
SP
Dar
k
n
et5
3
b
ac
k
b
o
n
e.
Fo
r
th
e
s
ak
e
o
f
an
ticip
ate
s
em
an
tic
s
eg
m
en
tatio
n
m
ask
s
,
YO
L
Ov
8
-
Seg
co
m
b
in
es
two
s
eg
m
en
tatio
n
h
ea
d
s
.
YOL
Ov
8
'
s
d
etec
tio
n
h
ea
d
s
h
av
e
th
e
s
am
e
f
iv
e
m
o
d
u
les
a
n
d
p
r
e
d
ictio
n
lay
er
as
th
e
id
e
n
tific
atio
n
h
ea
d
s
in
YOL
Ov
8
.
On
s
ev
er
al
o
b
ject
id
en
tific
atio
n
an
d
s
em
an
tic
s
eg
m
en
tatio
n
b
e
n
ch
m
ar
k
s
,
YOL
Ov
8
-
Seg
r
eg
u
lar
ly
d
eli
v
er
s
s
tate
-
of
-
th
e
-
ar
t
p
er
f
o
r
m
an
ce
wh
ile
r
etai
n
in
g
ex
ce
p
tio
n
al
s
p
ee
d
an
d
ec
o
n
o
m
y
[
1
1
]
,
[
1
2
]
.
YOL
Ov
8
s
im
p
lifie
s
in
s
tallatio
n
th
r
o
u
g
h
a
PIP
p
ac
k
ag
e
an
d
p
r
o
v
id
es
a
co
m
m
an
d
-
lin
e
i
n
ter
f
ac
e
(
C
L
I
)
f
o
r
e
x
ec
u
tio
n
.
I
t
o
f
f
er
s
v
ar
io
u
s
in
teg
r
atio
n
s
f
o
r
d
ep
l
o
y
m
en
t,
t
r
ain
in
g
,
an
d
lab
elin
g
task
s
.
Fig
u
r
e
2
s
h
o
ws
th
e
g
en
er
alize
d
ar
ch
itectu
r
e
o
f
th
e
YOL
Ov
8
.
Fro
m
o
b
j
ec
t
d
etec
tio
n
to
m
o
r
e
co
m
p
licated
task
s
lik
e
in
s
tan
ce
s
eg
m
en
tati
o
n
,
p
o
s
e/k
ey
p
o
in
ts
id
en
tific
atio
n
,
o
r
ien
ted
o
b
ject
r
ec
o
g
n
itio
n
,
an
d
class
if
icatio
n
,
th
ese
m
o
d
els
ar
e
m
ad
e
to
m
ee
t
a
v
ar
iety
o
f
n
ee
d
s
.
T
h
er
e
ar
e
f
iv
e
v
er
s
io
n
s
o
f
YOL
Ov
8
:
I
n
ter
m
s
o
f
p
ar
am
eter
c
o
u
n
t
an
d
f
lo
atin
g
o
p
e
r
atio
n
s
,
YOL
Ov
8
n
(
n
an
o
)
is
th
e
s
m
allest
m
o
d
e
l.
YOL
Ov
8
s
(
s
m
all)
is
th
e
n
ex
t
s
m
allest
m
o
d
el,
f
o
llo
w
ed
b
y
YOL
Ov
8
n
,
YOL
Ov
8
m
(
m
e
d
iu
m
)
,
YOL
Ov
8
l
(
lar
g
e
)
,
a
n
d
YOL
Ov
8
x
(
ex
tr
a
lar
g
e)
[
2
3
]
,
[
2
4
]
.
T
h
e
s
u
g
g
ested
ap
p
r
o
ac
h
m
ak
es
u
s
e
o
f
YOL
Ov
8
s
as
it
s
tr
ik
es
a
co
m
p
r
o
m
is
e
b
et
wee
n
h
av
in
g
m
o
r
e
p
ar
am
ete
r
s
an
d
FLOPs
th
an
YOL
Ov
8
n
wh
ile
r
em
ain
in
g
co
m
p
ar
ativ
ely
q
u
ick
in
co
m
p
a
r
is
o
n
to
th
e
b
ig
g
e
r
m
o
d
els.
Fig
u
r
e
2
.
YOL
Ov
8
ar
ch
itectu
r
e
2
.
5
.
P
er
f
o
r
m
a
nce
ev
a
lua
t
io
n
T
h
e
p
u
r
s
u
a
n
ce
o
f
th
e
p
r
o
to
t
y
p
e
is
in
ten
t
o
n
u
s
in
g
p
a
r
am
ete
r
lik
e
d
ataset
u
s
ed
,
n
u
m
b
er
o
f
ep
o
ch
s
to
tr
ain
th
e
d
ata
,
p
r
ec
is
io
n
,
r
ec
al
l,
m
AP5
0
an
d
m
AP5
0
-
9
5
.
Af
t
er
tr
ain
in
g
th
e
m
o
d
el
v
alid
atio
n
o
f
th
e
p
r
o
t
o
ty
p
e
is
p
er
f
o
r
m
e
d
u
s
in
g
th
e
c
o
n
f
u
s
io
n
m
atr
ix
.
I
t
co
n
tain
s
tr
u
e
p
o
s
itiv
es
in
wh
ich
th
e
m
o
d
el
p
r
ec
i
s
ely
p
r
ed
icts
a
lab
el
th
at
is
co
n
s
is
ten
t
with
r
ea
lity
.
Fals
e
p
o
s
itiv
e
ev
en
ts
i
n
d
icate
th
at
p
r
o
to
ty
p
e
h
as
p
r
ed
icted
a
lab
el
th
at
wasn
'
t
in
th
e
g
r
o
u
n
d
tr
u
th
.
W
h
en
a
m
o
d
el
p
r
e
d
icts
a
n
e
g
ativ
e
lab
el,
it
is
tr
u
e
n
eg
ativ
e
s
in
ce
it
is
co
m
p
atib
le
with
th
e
g
r
o
u
n
d
t
r
u
th
n
o
t
h
a
v
in
g
t
h
at
lab
el.
W
h
en
a
m
o
d
el
p
r
ed
icts
a
lab
el
t
h
at
is
r
ea
l
ly
a
elem
e
n
t
o
f
t
h
e
g
r
o
u
n
d
tr
u
t
h
,
it is
ap
p
ar
e
n
tly
p
r
o
d
u
cin
g
f
alse n
eg
ativ
e
o
cc
u
r
r
en
ce
s
[
2
3
]
.
Pre
cisi
o
n
is
a
m
ea
s
u
r
e
th
at
ex
p
r
ess
es
th
e
p
er
ce
n
tag
e
o
f
co
r
r
ec
t
p
r
ed
ictio
n
s
with
in
t
h
e
to
ta
l
n
u
m
b
er
o
f
p
r
ed
ictio
n
s
an
d
is
u
tili
ze
d
to
ass
es
s
h
o
w
ac
cu
r
ately
th
e
s
y
s
tem
r
esp
o
n
d
s
to
u
s
er
r
eq
u
ests
.
C
o
r
r
ec
tn
ess
o
f
th
e
m
o
d
el
is
ass
ess
ed
b
y
th
is
m
etr
ics.
C
o
n
v
er
s
ely
,
r
ec
all
is
th
e
am
o
u
n
t
o
f
g
e
n
u
in
e
p
o
s
itiv
e
s
to
all
item
s
.
Fo
r
in
s
tan
ce
,
if
an
im
a
g
e
h
a
d
1
0
0
tr
ee
s
an
d
th
e
m
o
d
el
p
r
o
p
er
ly
r
ec
o
g
n
ized
7
5
o
f
th
em
,
th
en
r
ec
all
is
7
5
%.
T
h
e
s
u
b
s
eq
u
en
t (
1
)
an
d
(
2
)
s
h
o
ws th
e
f
o
r
m
u
la
f
o
r
p
r
ec
is
io
n
an
d
r
ec
all.
=
/
+
(
1)
/
=
/
+
(
2)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
I
d
en
tifi
ca
tio
n
a
n
d
s
eg
men
t
a
tio
n
o
f tu
mo
r
u
s
in
g
d
ee
p
lea
r
n
i
n
g
a
n
d
ima
g
e
… (
S
h
ilp
a
C
h
ip
p
a
la
ka
tti
)
1787
T
h
e
m
ea
n
av
er
a
g
e
p
r
ec
is
io
n
(
m
AP)
is
th
e
av
e
r
ag
e
o
f
th
e
av
er
ag
e
p
r
ec
is
io
n
(
AP)
f
o
r
e
ac
h
class
.
R
ec
all
v
alu
es
b
et
wee
n
0
a
n
d
1
ar
e
u
s
ed
to
ca
lc
u
late
av
er
a
g
e
ac
cu
r
ac
y
.
A
n
u
m
b
er
o
f
p
ar
a
m
eter
s
in
f
lu
en
ce
th
e
m
ea
n
av
er
a
g
e
p
r
ec
is
io
n
ca
lcu
l
atio
n
,
in
clu
d
i
n
g
th
e
in
ter
s
ec
tio
n
o
v
er
u
n
io
n
(
I
OU)
,
p
r
ec
is
io
n
,
r
ec
all,
p
r
ec
is
io
n
-
r
ec
all
cu
r
v
e,
a
n
d
a
v
er
ag
e
p
r
e
cisi
o
n
(
AP)
.
T
h
e
am
o
u
n
t
o
f
o
v
er
lap
b
etwe
en
two
b
o
u
n
d
in
g
b
o
x
es
is
m
ea
s
u
r
ed
u
s
in
g
th
e
I
OU
m
etr
ic.
A
n
u
m
b
er
b
etwe
en
0
a
n
d
1
r
ep
r
esen
t
s
th
e
I
OU
v
alu
e,
wh
er
e
0
in
d
i
ca
tes
n
o
o
v
er
lap
a
n
d
1
in
d
icate
s
a
p
er
f
ec
t
m
atch
o
r
co
m
p
lete
o
v
er
la
p
[
2
4
]
.
T
h
e
m
etr
ics
m
AP5
0
ass
es
s
ac
cu
r
ac
y
at
th
e
0
.
5
o
r
5
0
%
I
OU
th
r
esh
o
ld
.
All
th
at
m
atter
s
is
th
at
if
th
er
e
is
m
o
r
e
th
an
5
0
%
o
v
er
lap
b
etwe
en
th
e
an
ti
cip
ated
an
d
g
r
o
u
n
d
tr
u
th
b
o
u
n
d
in
g
b
o
x
es,
th
e
p
r
o
ce
s
s
is
d
ee
m
ed
s
u
cc
ess
f
u
l.
I
n
a
s
im
ilar
way
,
m
AP5
0
-
9
5
d
en
o
tes
I
OU
b
etwe
en
50
an
d
9
5
.
In
(
3
)
a
n
d
(
4
)
wh
ic
h
is
s
h
o
wn
b
elo
w
is
in
ten
d
e
d
f
o
r
th
e
ca
lcu
latio
n
o
f
t
h
e
I
OU
a
n
d
m
AP.
=
/
(
3
)
=
1
/
=
1
(
4
)
2
.
6
.
I
m
ple
m
ent
a
t
io
n
T
h
e
wo
r
k
is
im
p
lem
en
ted
o
n
Go
o
g
le
C
o
lab
u
s
in
g
Py
th
o
n
lan
g
u
a
g
e.
T
h
e
s
y
s
tem
s
p
ec
if
icatio
n
f
o
r
tr
ain
in
g
th
e
m
o
d
el
ar
e
Ultr
aly
tics
YOL
Ov
8
.
0
.
1
9
6
,
Py
th
o
n
3
.
1
0
.
1
2
,
to
r
ch
-
2
.
3
.
0
+c
u
1
2
1
an
d
C
PU
(
I
n
tel
Xeo
n
2
.
2
0
GHz
)
.
T
h
e
d
ataset
co
n
s
id
er
ed
f
o
r
ev
alu
atio
n
c
o
n
s
is
t
o
f
th
e
6
3
9
im
a
g
es
o
u
t
o
f
wh
ic
h
4
5
3
im
a
g
es
wer
e
f
o
r
tr
ain
i
n
g
,
1
2
2
im
ag
es
wer
e
m
e
an
t
f
o
r
v
alid
atio
n
a
n
d
6
4
im
a
g
es
wer
e
m
ea
n
t
f
o
r
test
in
g
i
n
a
r
atio
o
f
7
1
:1
9
:1
0
.
T
h
e
T
ab
le
1
s
p
ec
if
ies
th
e
e
n
g
i
n
e
o
r
th
e
tr
ain
er
s
p
ec
if
icatio
n
i
n
d
etail.
T
h
e
im
ag
es
u
tili
ze
d
f
o
r
tr
ain
i
n
g
p
u
r
p
o
s
e
h
as f
o
llo
win
g
s
p
ec
if
icatio
n
.
T
ab
le
1
.
T
r
ai
n
er
s
p
ec
if
icatio
n
C
o
n
f
i
g
u
r
a
t
i
o
n
V
a
l
u
e
A
l
g
o
r
i
t
h
m
u
s
e
d
Y
O
LO
v
8
s
I
mag
e
s
i
z
e
6
4
0
×
6
4
0
B
a
t
c
h
si
z
e
16
P
a
t
i
e
n
c
e
50
e
p
o
c
h
s
20
3.
RE
SU
L
T
AND
DI
SCUS
SI
O
N
T
u
m
o
r
d
etec
tio
n
ca
p
ab
ilit
y
o
f
th
e
YOL
Ov
8
is
ev
alu
ated
u
s
in
g
p
a
r
am
eter
s
u
ch
as
p
r
e
cisi
o
n
,
r
ec
all
an
d
m
AP5
0
an
d
m
AP5
0
-
95.
T
h
e
p
r
o
t
o
ty
p
e
d
etec
tio
n
f
in
d
i
n
g
s
ar
e
ex
am
in
e
d
to
ev
alu
ate
h
o
w
well
it
d
etec
ts
tu
m
o
r
.
T
h
e
p
r
o
to
ty
p
e
was
tr
ain
ed
f
o
r
2
0
ep
o
c
h
s
,
tak
in
g
in
s
tan
ce
o
f
5
,
1
0
,
1
5
an
d
2
0
an
d
e
v
alu
atin
g
v
alu
es
o
f
p
r
ec
is
io
n
,
r
ec
all,
m
AP5
0
an
d
m
AP5
0
-
9
5
f
o
r
ea
ch
in
s
t
an
ce
.
T
h
e
m
o
d
el
h
as
1
6
8
lay
e
r
s
,
1
1
1
2
5
9
7
1
p
ar
am
eter
s
,
0
g
r
a
d
ien
ts
an
d
2
8
.
4
GFLO
PS
.
W
e
s
tar
ted
tr
ain
in
g
f
o
r
5
,
1
0
,
1
5
,
2
0
ep
o
ch
s
r
esp
ec
tiv
ely
an
d
n
o
ted
th
e
v
alu
es
o
f
th
e
p
r
ec
is
io
n
,
r
ec
all,
m
AP5
0
an
d
m
AP5
0
-
9
5
.
T
h
e
e
p
o
ch
s
wer
e
s
to
p
p
ed
f
o
r
2
0
b
ec
a
u
s
e
af
ter
2
0
e
p
o
c
h
s
t
h
er
e
was
n
o
t
m
u
ch
ch
an
g
e
in
th
e
v
alu
es
o
f
p
r
ec
is
io
n
,
r
ec
all
an
d
m
AP.
T
h
e
f
o
llo
win
g
T
a
b
le
2
s
h
o
ws
th
e
o
u
tco
m
e
o
f
d
is
tin
ct
ep
o
c
h
s
.
T
ab
le
2
.
Ou
tp
u
t o
f
d
is
tin
ct
ep
o
ch
s
Ep
o
c
h
s
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
mA
P
5
0
mA
P
5
0
-
95
5
0
.
7
5
9
0
.
7
3
3
0
.
8
0
2
0
.
5
9
5
10
0
.
9
3
3
0
.
9
1
5
0
.
9
4
7
0
.
7
5
6
15
0
.
9
3
9
0
.
9
1
5
0
.
9
6
1
0
.
7
8
20
0
.
9
4
4
0
.
9
2
1
0
.
9
6
9
0
.
8
1
1
T
h
e
Fig
u
r
e
3
s
h
o
ws
th
e
tr
ain
in
g
lo
s
s
es
an
d
Fig
u
r
e
s
3
(
a
)
-
(
c
)
s
h
o
ws
th
e
b
o
x
_
lo
s
s
,
cl
s
_
lo
s
s
,
d
f
l_
lo
s
s
r
esp
ec
tiv
ely
.
T
h
e
u
s
e
o
f
b
o
u
n
d
in
g
b
o
x
es
to
d
etec
t
s
ev
er
al
it
em
s
in
a
im
ag
e
is
s
h
o
wn
in
F
ig
u
r
e
3
(
a
)
.
Fin
d
i
n
g
th
e
er
r
o
r
b
etwe
en
th
e
ex
p
ec
te
d
an
d
g
r
o
u
n
d
tr
u
th
b
o
u
n
d
i
n
g
b
o
x
es
is
th
e
m
ain
g
o
al
o
f
th
e
b
o
x
lo
s
s
f
u
n
ctio
n
.
T
h
e
m
o
d
el'
s
ab
ilit
y
to
f
o
r
ec
ast th
e
b
o
u
n
d
in
g
b
o
x
es with
ac
cu
r
ac
y
will in
cr
ea
s
e
as th
ese
lo
s
s
es d
im
in
is
h
ac
r
o
s
s
ep
o
ch
s
.
T
h
e
o
b
jectn
ess
lo
s
s
,
o
r
h
o
w
th
e
o
b
jectn
ess
an
d
class
s
co
r
es
b
eh
av
e
in
th
e
m
o
d
el,
is
s
h
o
wn
in
Fig
u
r
e
3
(
b
)
.
C
lass
s
co
r
e
in
d
ic
ates
th
e
co
n
d
itio
n
al
p
r
o
b
ab
ilit
y
o
f
a
ce
r
tain
class
,
wh
er
ea
s
o
b
jectn
ess
r
elate
s
to
th
e
m
o
d
el'
s
co
n
f
id
en
ce
in
th
e
p
r
esen
ce
o
f
a
n
o
b
jec
t
in
s
id
e
th
e
b
o
u
n
d
in
g
b
o
x
.
T
h
e
o
b
jectn
ess
an
d
class
s
co
r
e
ar
e
m
u
ltip
lied
to
g
et
th
e
o
v
e
r
all
co
n
f
id
en
ce
c
u
r
v
e.
As
th
e
g
r
ap
h
illu
s
tr
ates,
o
b
jectn
ess
s
h
o
u
ld
id
ea
lly
f
all
to
war
d
s
ze
r
o
as
th
e
n
u
m
b
e
r
o
f
ep
o
ch
s
in
c
r
ea
s
es.
I
n
r
elatio
n
to
ca
teg
o
r
izatio
n
,
Fig
u
r
e
3
(
c
)
.
Dete
r
m
in
in
g
if
a
n
o
b
ject
is
in
a
p
ictu
r
e
an
d
id
e
n
tify
in
g
its
class
ar
e
th
e
two
co
m
p
o
n
en
ts
o
f
ca
teg
o
r
izatio
n
.
T
h
e
p
r
o
to
ty
p
e'
s
ab
ilit
y
to
ca
teg
o
r
ize
in
s
id
e
th
e
an
ticip
ated
b
o
u
n
d
ar
y
b
o
x
es
is
d
ep
icted
in
th
e
g
r
ap
h
.
T
h
e
v
alid
atio
n
lo
s
s
es
ar
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
7
8
2
-
1
7
9
2
1788
s
h
o
wn
in
Fig
u
r
e
4
,
an
d
th
e
v
alid
atio
n
b
o
x
_
lo
s
s
,
cls_
lo
s
s
,
an
d
d
f
l_
lo
s
s
ar
e
s
h
o
wn
in
Fig
u
r
es
4
(
a
)
-
(
c
)
,
r
esp
ec
tiv
ely
.
T
h
e
p
er
f
o
r
m
an
ce
m
atr
ix
o
f
alg
o
r
ith
m
is
s
h
o
wn
in
Fig
u
r
e
5
,
wh
er
e
p
r
ec
is
io
n
,
r
ec
all,
m
AP5
0
,
an
d
m
AP5
0
-
9
5
o
f
th
e
tr
ain
in
g
d
at
aset
ar
e
s
h
o
wn
o
n
th
e
g
r
a
p
h
'
s
y
-
ax
is
i
n
Fig
u
r
es
5
(
a
)
-
(
d
)
,
w
h
ile
th
e
n
u
m
b
er
o
f
ep
o
ch
s
is
s
h
o
wn
o
n
th
e
x
-
a
x
is
.
At
th
e
co
n
clu
s
io
n
o
f
2
0
ep
o
ch
s
,
th
e
s
u
g
g
ested
m
o
d
el'
s
lo
s
s
es
ar
e
r
ep
o
r
ted
as
b
o
x
_
lo
s
s
o
f
0
.
6
7
6
5
,
cls_
lo
s
s
o
f
0
.
5
0
7
,
a
n
d
d
f
l_
lo
s
s
o
f
1
.
0
3
9
.
(
a)
(
b
)
(
c)
Fig
u
r
e
3
.
Gr
a
p
h
ical
r
ep
r
esen
tatio
n
o
f
tr
ai
n
in
g
lo
s
s
:
(
a)
b
o
x
l
o
s
s
,
(
b
)
cls lo
s
s
,
an
d
(
c)
d
f
l lo
s
s
(
a)
(
b
)
(
c)
Fig
u
r
e
4
.
Gr
a
p
h
ical
r
ep
r
esen
tatio
n
o
f
v
alid
atio
n
lo
s
s
:
(
a)
b
o
x
lo
s
s
,
(
b
)
cls lo
s
s
,
an
d
(
c)
d
f
l l
o
s
s
(
a)
(
b
)
(
c)
(
d
)
F
i
g
u
r
e
5
.
G
r
a
p
h
i
c
a
l
r
e
p
r
e
s
e
n
ta
ti
o
n
o
f
:
(
a
)
p
r
e
c
i
s
i
o
n
,
(
b
)
r
e
c
a
ll
,
(
c
)
m
A
P
5
0
,
a
n
d
(
d
)
m
A
P
5
0
-
9
5
w
.
r
.
t
n
u
m
b
e
r
o
f
e
p
o
c
h
s
Fig
u
r
e
6
s
h
o
ws
th
e
p
r
ec
is
io
n
-
r
ec
all
v
alu
es
o
b
tain
ed
f
r
o
m
ea
ch
ep
o
ch
s
ar
e
p
lo
tted
in
th
e
f
o
r
m
o
f
p
r
ec
is
io
n
-
r
ec
all
c
u
r
v
e.
T
h
is
g
r
ap
h
s
p
ec
if
ies
h
o
w
ef
f
icien
tly
th
e
m
o
d
el
is
g
o
in
g
to
p
r
ed
ict
th
e
tu
m
o
r
.
Acc
o
r
d
in
g
t
o
th
is
g
r
ap
h
o
u
r
m
o
d
el’
s
p
r
ed
ictio
n
r
ate
is
0
.
9
6
9
.
T
h
e
n
o
r
m
alize
d
co
n
f
u
s
io
n
m
a
tr
ix
f
o
r
th
e
s
u
g
g
ested
YOL
O
m
o
d
el
is
s
h
o
wn
in
Fig
u
r
e
7
.
T
h
e
b
est
-
p
er
f
o
r
m
in
g
an
d
wo
r
s
t
-
p
er
f
o
r
m
in
g
m
o
d
els
ar
e
id
en
tifie
d
b
y
u
s
in
g
th
e
co
n
f
u
s
io
n
m
atr
ix
,
wh
ich
p
r
o
v
i
d
es
a
m
o
r
e
th
o
r
o
u
g
h
v
iew
o
f
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
ac
r
o
s
s
s
ev
er
al
class
es.
I
n
ad
d
itio
n
,
th
e
co
n
f
u
s
io
n
m
atr
ix
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
I
d
en
tifi
ca
tio
n
a
n
d
s
eg
men
t
a
tio
n
o
f tu
mo
r
u
s
in
g
d
ee
p
lea
r
n
i
n
g
a
n
d
ima
g
e
… (
S
h
ilp
a
C
h
ip
p
a
la
ka
tti
)
1789
o
f
f
er
s
p
er
ce
p
tio
n
in
to
th
e
m
is
class
if
icatio
n
tr
en
d
s
an
d
ass
is
ts
in
p
in
p
o
in
tin
g
th
e
p
r
ec
is
e
i
n
s
tan
ce
s
th
at
h
av
e
b
ee
n
er
r
o
n
eo
u
s
ly
ca
teg
o
r
ized
.
On
e
m
ay
s
ee
th
e
d
is
tr
ib
u
tio
n
o
f
ac
tu
al
lab
els
f
o
r
ea
ch
clas
s
an
d
f
o
r
ec
asts
f
o
r
ea
ch
class
b
y
lo
o
k
i
n
g
at
th
e
co
n
f
u
s
io
n
m
atr
ix
.
I
t
m
ak
es
it
p
o
s
s
ib
le
to
ass
ess
th
e
co
r
r
ec
tn
ess
o
f
th
e
m
o
d
el
th
o
r
o
u
g
h
ly
,
e
m
p
h
asizin
g
r
e
g
io
n
s
wh
er
e
m
is
class
if
icatio
n
s
ar
e
m
o
r
e
co
m
m
o
n
an
d
p
i
n
p
o
in
tin
g
p
o
s
s
ib
le
m
is
tak
e
ca
u
s
es.
T
h
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
in
o
b
ject
id
e
n
tific
atio
n
task
s
m
ay
b
e
im
p
r
o
v
ed
a
n
d
r
ef
in
e
d
with
th
e
u
s
e
o
f
th
is
in
f
o
r
m
atio
n
.
Fig
u
r
e
6
.
Pre
cisi
o
n
-
r
ec
all
cu
r
v
e
Fig
u
r
e
7
.
No
r
m
alize
d
co
n
f
u
s
io
n
m
atr
ix
Af
ter
d
etec
tin
g
,
th
e
tu
m
o
r
s
eg
m
en
tatio
n
is
d
o
n
e
u
s
in
g
im
ag
e
s
eg
m
en
tatio
n
alg
o
r
ith
m
.
T
h
e
s
eg
m
en
ted
im
ag
e
is
p
lace
d
o
n
th
e
clo
u
d
f
o
r
f
u
tu
r
e
r
ef
e
r
en
ce
.
T
h
e
b
en
ef
it
o
f
s
av
in
g
th
e
im
ag
e
in
th
e
clo
u
d
m
ig
h
t
in
clu
d
e
m
ain
tain
i
n
g
th
e
p
atien
t
m
ed
ical
h
is
to
r
y
,
r
esear
c
h
p
u
r
p
o
s
e
etc.
T
h
e
s
eg
m
en
t
atio
n
p
r
o
ce
s
s
m
ay
r
ed
u
ce
th
e
s
to
r
ag
e
s
p
ac
e
i
n
t
h
e
clo
u
d
in
tu
r
n
co
s
t
o
f
s
to
r
ag
e
is
r
ed
u
ce
d
.
T
h
e
Fig
u
r
e
8
s
h
o
ws
p
r
o
ce
s
s
o
f
s
eg
m
en
tatio
n
.
T
h
e
tu
m
o
r
p
r
e
d
icted
u
s
in
g
YOL
Ov
8
s
alg
o
r
ith
m
is
d
ep
icted
in
Fig
u
r
e
s
8
(
a
)
an
d
8
(
b
)
s
h
o
ws
s
eg
m
en
t
ed
im
ag
e.
(
a)
(
b
)
Fig
u
r
e
8
.
Seg
m
e
n
tatio
n
p
r
o
ce
s
s
(
a)
p
r
ed
ictio
n
o
f
t
u
m
o
r
in
o
r
i
g
in
al
im
ag
e
u
s
in
g
YOL
Ov
8
s
m
o
d
el
an
d
(
b
)
t
u
m
o
r
s
eg
m
e
n
tatio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
38
,
No
.
3
,
J
u
n
e
20
25
:
1
7
8
2
-
1
7
9
2
1790
T
h
e
m
ain
o
b
jectiv
e
o
f
t
h
is
s
tu
d
y
is
to
lo
ca
te
th
e
tu
m
o
r
p
r
ec
is
ely
an
d
f
u
r
th
er
s
eg
m
en
t
t
h
e
in
f
ec
ted
ar
ea
to
s
av
e
i
n
clo
u
d
to
m
ain
t
ain
th
e
p
atien
t
m
ed
ical
h
is
to
r
y
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
y
ield
s
(
0
.
9
4
4
)
Pre
cisi
o
n
,
(
0
.
9
2
1
)
R
ec
all,
(
0
.
9
6
9
)
m
AP5
0
an
d
(
0
.
8
1
1
)
m
AP5
0
-
9
5
f
o
r
2
0
ep
o
ch
s
o
n
th
e
o
p
en
-
s
o
u
r
ce
d
ataset
wh
ich
is
u
s
ed
f
o
r
r
esea
r
c
h
.
T
h
e
d
ataset
co
n
s
is
ts
o
f
6
3
9
im
a
g
es
wh
ich
is
s
p
lit
in
th
e
r
atio
o
f
7
1
:1
9
:1
0
.
I
n
[
1
6
]
YOL
Ov
2
alo
n
g
with
i
n
ce
p
tio
n
v
3
is
u
s
e
d
to
e
n
h
an
ce
th
e
p
r
ed
ictio
n
an
d
th
ey
ac
h
iev
ed
ac
cu
r
ac
y
u
p
to
0
.
9
0
.
Acc
o
r
d
in
g
to
Ali
et
a
l.
[
1
8
]
YOL
Ov
3
is
u
s
e
d
f
o
r
tu
m
o
r
lo
ca
lizatio
n
wh
ic
h
g
av
e
ac
c
u
r
ac
y
o
f
8
3
%
af
ter
f
in
e
tu
n
in
g
th
e
p
r
e
-
tr
ain
ed
YOL
Ov
3
m
o
d
el.
Acc
o
r
d
in
g
to
[
1
9
]
YOL
Ov
4
is
u
s
ed
f
o
r
tu
m
o
r
id
en
tific
atio
n
wh
ic
h
g
av
e
an
ac
cu
r
ac
y
o
f
0
.
8
3
2
4
o
n
th
e
p
r
o
ce
s
s
ed
d
ata.
I
n
Sh
elatk
ar
et
a
l.
[
2
1
]
YOL
Ov
5
is
u
s
ed
f
o
r
t
u
m
o
r
d
etec
tio
n
a
n
d
cl
ass
if
icatio
n
o
n
B
R
AT
S2
0
2
1
d
ataset
an
d
m
o
d
el
g
av
e
an
ac
cu
r
ac
y
o
f
8
8
%.
Acc
o
r
d
in
g
to
Ab
d
u
s
alo
m
o
v
et
a
l.
[
2
2
]
YOL
Ov
7
is
u
s
ed
f
o
r
tu
m
o
r
d
etec
tio
n
an
d
class
if
icatio
n
alo
n
g
with
t
h
e
f
ea
tu
r
e
e
x
tr
ac
to
r
,
s
en
s
itiv
ity
en
h
an
ce
r
an
d
m
u
ltis
tag
e
f
ea
t
u
r
e
f
u
s
io
n
to
g
et
th
e
r
elev
a
n
t
i
n
f
o
r
m
atio
n
.
Pas
s
a
et
a
l.
[
2
3
]
p
r
o
p
o
s
ed
YOL
Ov
8
alo
n
g
with
d
ata
au
g
m
e
n
tatio
n
ca
n
d
etec
t
b
r
ain
tu
m
o
r
e
f
f
icien
tly
an
d
it
g
av
e
ac
c
u
r
ac
y
ar
o
u
n
d
0
.
9
5
.
T
h
e
r
esear
ch
er
s
[
2
4
]
also
u
s
ed
YOL
Ov
8
s
m
o
d
el
f
o
r
th
e
d
etec
tin
g
tu
m
o
r
o
n
th
e
o
p
en
-
s
o
u
r
ce
d
at
aset a
n
d
it c
o
n
tain
s
o
n
ly
3
0
0
im
a
g
es,
d
ata
au
g
m
e
n
tatio
n
was
also
ap
p
lied
an
d
t
h
ey
g
o
t
ac
cu
r
ac
y
o
f
0
.
9
4
1
.
Ac
co
r
d
in
g
to
Hash
em
i
et
a
l.
[
2
6
]
to
e
n
h
an
ce
t
h
e
p
er
f
o
r
m
an
ce
o
f
YOL
Ov
8
n
,
DeiT
m
o
d
el
is
u
s
ed
.
I
n
C
h
en
et
a
l.
[
2
7
]
p
h
o
to
ac
o
u
tic
im
ag
in
g
is
u
s
ed
in
s
tead
o
f
M
R
I
im
ag
es
to
g
et
b
etter
r
esu
lts
f
r
o
m
YOL
Ov
8
.
C
o
m
p
a
r
ed
t
o
all
th
e
ab
o
v
e
wo
r
k
,
p
r
o
p
o
s
ed
m
o
d
el
is
g
iv
in
g
b
et
ter
r
esu
lts
with
o
p
en
-
s
o
u
r
ce
d
ataset.
Data
p
r
ep
r
o
ce
s
s
in
g
is
u
s
ed
in
m
o
d
el
f
o
r
n
o
is
e
r
ed
u
ctio
n
an
d
d
ata
au
g
m
en
tatio
n
is
u
s
ed
to
en
h
an
ce
th
e
d
ata.
T
h
e
r
esear
ch
er
s
in
[
2
3
]
,
[
2
4
]
,
[
2
6
]
,
[
2
7
]
u
s
es
YOL
Ov
8
alg
o
r
ith
m
b
u
t
f
o
r
d
if
f
er
e
n
t
d
ataset
an
d
p
ar
am
eter
s
co
m
p
ar
ed
to
o
u
r
s
tu
d
y
.
T
h
e
d
ataset
u
s
ed
in
p
r
o
p
o
s
ed
m
o
d
el
was
test
ed
o
n
th
e
YOL
Ov
5
m
o
d
el
k
ee
p
i
n
g
all
th
e
p
ar
am
eter
s
s
am
e
as
u
s
ed
in
th
e
YOL
Ov
8
an
d
r
esu
lts
wer
e
n
o
ted
.
T
ab
le
3
s
h
o
ws th
e
co
m
p
ar
is
o
n
o
f
Y
OL
Ov
5
an
d
YOL
Ov
8
alg
o
r
ith
m
u
s
ed
f
o
r
th
e
s
am
e
d
ataset,
f
r
o
m
th
e
ta
b
le
it
is
clea
r
ly
n
o
ted
th
at
YOL
Ov
8
alg
o
r
ith
m
is
g
iv
in
g
b
etter
ac
cu
r
a
cy
co
m
p
a
r
ed
to
t
h
e
YOL
Ov
5
.
As
well
as
in
all
th
e
ab
o
v
e
p
ap
er
r
esear
ch
e
r
h
as
o
n
ly
co
n
ce
n
tr
ated
o
n
th
e
class
if
icatio
n
an
d
d
etec
tio
n
o
f
t
u
m
o
r
b
u
t
in
s
u
g
g
ested
m
o
d
el
im
ag
e
s
eg
m
en
ta
tio
n
is
d
o
n
e
to
r
ed
u
ce
th
e
im
a
g
e
s
to
r
ag
e
s
p
ac
e
in
clo
u
d
.
R
ed
u
ctio
n
in
im
ag
e
s
to
r
ag
e
s
p
ac
e
is
d
ep
icte
d
in
T
ab
l
e
4
f
o
r
th
e
s
am
p
le
im
a
g
e
u
s
ed
f
o
r
s
eg
m
e
n
tatio
n
,
f
r
o
m
th
e
tab
le
it
is
clea
r
th
at
b
y
th
e
im
ag
e
s
eg
m
en
tatio
n
t
ec
h
n
iq
u
e
we
ca
n
s
u
b
s
tan
tially
r
ed
u
ce
th
e
clo
u
d
s
to
r
ag
e
s
p
ac
e.
T
ab
le
3
.
C
o
m
p
a
r
is
o
n
o
f
YOL
Ov
5
an
d
YOL
Ov
8
o
n
th
e
s
am
e
d
ataset
A
l
g
o
r
i
t
h
m
Ep
o
c
h
s
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
mA
P
5
0
mA
P
5
0
-
95
Y
O
LO
v
5
20
0
.
8
6
1
0
.
7
8
3
0
.
8
5
5
0
.
5
8
7
P
r
o
p
o
se
d
m
o
d
e
l
(
Y
O
LO
v
8
s)
20
0
.
9
4
4
0
.
9
2
1
0
.
9
6
9
0
.
8
1
1
T
ab
le
4
.
I
m
ag
e
s
to
r
ag
e
s
p
ac
e
b
ef
o
r
e
a
n
d
af
te
r
s
eg
m
en
tatio
n
M
e
m
o
r
y
st
o
r
a
g
e
r
e
q
u
i
r
e
d
b
e
f
o
r
e
se
g
m
e
n
t
a
t
i
o
n
M
e
m
o
r
y
st
o
r
a
g
e
r
e
q
u
i
r
e
d
a
f
t
e
r
se
g
m
e
n
t
a
t
i
o
n
2
8
k
B
1
6
k
B
4.
CO
NCLU
SI
O
N
T
h
e
au
to
m
ated
m
eth
o
d
f
o
r
a
cc
u
r
ate
d
etec
tio
n
o
f
tu
m
o
r
with
less
co
m
p
lex
ity
an
d
co
s
t
-
ef
f
ec
tiv
e
s
to
r
ag
e
o
f
p
atien
t’
s
m
ed
ical
h
i
s
to
r
y
was
n
ee
d
ed
.
Ou
r
p
r
o
p
o
s
ed
m
o
d
el
a
n
s
wer
ed
th
is
is
s
u
e
b
y
u
s
in
g
YOL
Ov
8
s
m
o
d
el
f
o
r
tu
m
o
r
d
etec
tio
n
an
d
im
ag
e
s
eg
m
en
tatio
n
is
d
o
n
e
t
o
s
av
e
th
e
s
to
r
ag
e
s
p
ac
e
in
clo
u
d
.
T
h
e
ef
f
ec
tiv
en
ess
o
f
t
h
e
YOL
Ov
8
s
m
o
d
el
in
d
etec
tin
g
b
r
ain
tu
m
o
r
s
an
d
s
eg
m
en
tin
g
th
e
tu
m
o
r
u
s
in
g
an
im
ag
e
s
eg
m
en
tatio
n
m
eth
o
d
is
p
r
esen
ted
in
t
h
is
s
tu
d
y
.
On
e
o
f
th
e
DL
m
o
d
els,
YOL
Ov
8
s
,
is
ac
cu
s
to
m
ed
to
d
etec
t
th
e
tu
m
o
r
.
T
h
e
d
ataset
th
at
is
b
ein
g
u
s
ed
h
er
e
is
an
o
p
en
-
s
o
u
r
ce
d
ataset
th
at
is
u
tili
ze
d
t
o
s
u
p
p
o
r
t
s
cien
tific
r
esear
ch
.
6
3
9
tu
m
o
r
im
ag
es
m
ak
e
u
p
th
e
d
ataset,
wh
ich
is
s
p
lit
in
a
r
atio
o
f
7
1
:1
9
:1
0
ac
r
o
s
s
tr
ain
in
g
,
v
alid
atio
n
,
a
n
d
test
in
g
.
Pre
p
r
o
ce
s
s
in
g
an
d
Data
au
g
m
e
n
tatio
n
is
ca
r
r
ie
d
o
u
t
p
r
i
o
r
to
tr
ai
n
in
g
.
2
0
e
p
o
ch
s
a
r
e
u
tili
ze
d
to
tr
ain
t
h
e
m
o
d
el.
I
n
co
m
p
ar
is
o
n
to
p
r
e
v
io
u
s
r
esear
ch
er
s
wh
o
h
av
e
u
s
ed
t
h
e
s
am
e
ap
p
r
o
ac
h
f
o
r
tu
m
o
r
d
etec
tio
n
b
u
t
h
av
e
d
i
f
f
er
en
t
d
ataset
an
d
p
ar
am
eter
s
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
r
esu
lts
af
ter
2
0
ep
o
c
h
s
s
h
o
w
th
at
it
is
ab
le
to
d
etec
t
th
e
tu
m
o
r
m
o
r
e
p
r
ec
is
ely
with
a
p
r
ec
is
io
n
v
alu
e
o
f
0
.
9
4
4
,
r
ec
all
o
f
0
.
9
2
1
,
m
AP5
0
o
f
0
.
9
6
9
,
an
d
m
AP5
0
-
9
5
o
f
0
.
8
1
1
.
T
h
u
s
,
we
m
ay
s
ay
th
at
th
e
YOL
Ov
8
s
m
o
d
el
h
as
f
ast
an
d
g
o
o
d
tu
m
o
r
d
etec
tio
n
r
ate.
Fu
r
th
e
r
m
o
r
e
,
t
h
e
tu
m
o
r
is
s
eg
m
e
n
ted
u
s
in
g
th
e
im
a
g
e
s
eg
m
e
n
tatio
n
a
p
p
r
o
ac
h
a
n
d
o
n
l
y
th
e
r
eg
io
n
o
f
in
ter
est
is
s
av
ed
in
clo
u
d
wh
ich
i
n
tu
r
n
r
ed
u
ce
th
e
clo
u
d
s
to
r
a
g
e
s
p
ac
e
wh
ic
h
lead
s
to
lo
w
-
co
s
t
s
to
r
ag
e.
Fu
tu
r
e
s
co
p
e
o
f
t
h
is
wo
r
k
is
to
en
h
a
n
ce
th
e
p
er
f
e
ctn
ess
o
f
b
r
ai
n
tu
m
o
r
d
etec
tio
n
u
s
in
g
YOL
Ov
8
s
m
o
d
els
b
y
tr
ain
in
g
m
o
d
el
wit
h
a
v
ar
io
u
s
k
in
d
o
f
d
ataset
.
R
eb
u
ild
in
g
th
e
s
eg
m
en
ted
im
ag
es
an
d
r
ed
u
cin
g
th
e
im
ag
e
r
ec
o
n
s
tr
u
ctio
n
lo
s
s
es.
YOL
Ov
8
s
m
o
d
el
c
an
also
b
e
u
s
ed
f
o
r
p
r
ed
ictio
n
o
f
d
is
ea
s
es
in
X
-
r
ay
,
C
T
s
ca
n
,
u
ltra
s
o
u
n
d
,
PET
s
ca
n
.
Ou
r
r
ese
ar
ch
m
i
g
h
t
m
ak
e
it
ea
s
ier
a
n
d
m
o
r
e
p
r
ec
is
e
f
o
r
m
ed
ical
p
r
o
f
ess
io
n
als
to
lo
ca
te
tu
m
o
r
an
d
s
eg
m
en
tatio
n
p
r
o
ce
s
s
m
ay
h
elp
in
co
s
t e
f
f
ec
tiv
e
d
ata
s
to
r
ag
e
in
clo
u
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
I
d
en
tifi
ca
tio
n
a
n
d
s
eg
men
t
a
tio
n
o
f tu
mo
r
u
s
in
g
d
ee
p
lea
r
n
i
n
g
a
n
d
ima
g
e
… (
S
h
ilp
a
C
h
ip
p
a
la
ka
tti
)
1791
F
U
N
DI
N
G
I
N
F
O
RM
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
A
U
T
H
O
R
C
O
NT
R
I
B
UT
I
O
N
S
ST
AT
E
M
E
NT
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Sh
ilp
a
C
h
ip
p
alak
atti
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
R
en
u
Ma
d
h
av
i
C
h
o
d
av
ar
ap
u
✓
✓
✓
✓
✓
✓
✓
An
d
h
e
Pallav
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
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
o
p
e
n
l
y
a
v
a
i
l
a
b
l
e
i
n
R
o
b
o
f
l
o
w
U
n
i
v
e
r
s
e
a
t
h
t
t
p
s
:
/
/
a
p
p
.
r
o
b
o
f
l
o
w
.
c
o
m
/
s
h
i
l
p
a
1
/
b
r
t
u
m
o
r
/
b
r
o
w
s
e
?
q
u
e
r
y
T
e
x
t
=
&
p
a
g
e
S
i
z
e
=
5
0
&
s
t
a
r
t
i
n
g
I
n
d
e
x
=
0
&
b
r
o
w
s
e
Q
u
e
r
y
=
t
r
u
e
,
r
e
f
e
r
e
n
c
e
[
2
9
]
.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
K
a
i
f
i
,
“
A
r
e
v
i
e
w
o
f
r
e
c
e
n
t
a
d
v
a
n
c
e
s
i
n
b
r
a
i
n
t
u
m
o
r
d
i
a
g
n
o
si
s
b
a
se
d
o
n
A
I
-
b
a
sed
c
l
a
ssi
f
i
c
a
t
i
o
n
,
”
D
i
a
g
n
o
st
i
c
s
,
v
o
l
.
1
3
,
n
o
.
1
8
,
p
.
3
0
0
7
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
d
i
a
g
n
o
st
i
c
s
1
3
1
8
3
0
0
7
.
[
2
]
C
.
W
a
t
s
o
n
,
M
.
K
i
r
k
c
a
l
d
i
e
,
a
n
d
G
.
P
a
x
i
n
o
s,
“
T
h
e
b
r
a
i
n
:
a
n
i
n
t
r
o
d
u
c
t
i
o
n
t
o
f
u
n
c
t
i
o
n
a
l
n
e
u
r
o
a
n
a
t
o
my
,
”
Ac
a
d
e
m
i
c
Pr
e
ss
,
2
0
1
0
.
[
3
]
S
.
A
l
i
,
J.
Li
,
Y
.
P
e
i
,
R
.
K
h
u
r
r
a
m
,
K
.
u
r
R
e
h
m
a
n
,
a
n
d
T
.
M
a
h
mo
o
d
,
“
A
c
o
m
p
r
e
h
e
n
si
v
e
s
u
r
v
e
y
o
n
b
r
a
i
n
t
u
m
o
r
d
i
a
g
n
o
s
i
s u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
a
n
d
e
m
e
r
g
i
n
g
h
y
b
r
i
d
t
e
c
h
n
i
q
u
e
s
w
i
t
h
m
u
l
t
i
-
m
o
d
a
l
M
R
i
m
a
g
e
,
”
Ar
c
h
i
v
e
s
o
f
C
o
m
p
u
t
a
t
i
o
n
a
l
M
e
t
h
o
d
s
i
n
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
2
9
,
n
o
.
7
,
p
p
.
4
8
7
1
–
4
8
9
6
,
M
a
y
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
8
3
1
-
022
-
0
9
7
5
8
-
z.
[
4
]
F
.
B
r
a
y
,
J.
F
e
r
l
a
y
,
I
.
S
o
e
r
j
o
m
a
t
a
r
a
m
,
R
.
L
.
S
i
e
g
e
l
,
L.
A
.
T
o
r
r
e
,
a
n
d
A
.
Jema
l
,
“
G
l
o
b
a
l
c
a
n
c
e
r
st
a
t
i
st
i
c
s
2
0
1
8
:
G
LO
B
O
C
A
N
e
st
i
mat
e
s
o
f
i
n
c
i
d
e
n
c
e
a
n
d
m
o
r
t
a
l
i
t
y
w
o
r
l
d
w
i
d
e
f
o
r
3
6
c
a
n
c
e
r
s
i
n
1
8
5
c
o
u
n
t
r
i
e
s,
”
C
A:
A
C
a
n
c
e
r
J
o
u
r
n
a
l
f
o
r
C
l
i
n
i
c
i
a
n
s
,
v
o
l
.
6
8
,
n
o
.
6
,
p
p
.
3
9
4
–
4
2
4
,
S
e
p
.
2
0
1
8
,
d
o
i
:
1
0
.
3
3
2
2
/
c
a
a
c
.
2
1
4
9
2
.
[
5
]
H
.
B
y
a
l
e
,
G
.
M
.
L
i
n
g
a
r
a
j
u
,
a
n
d
S
.
S
i
v
a
s
u
b
r
a
ma
n
i
a
n
,
“
A
u
t
o
ma
t
i
c
se
g
me
n
t
a
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
b
r
a
i
n
t
u
m
o
r
u
s
i
n
g
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
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
En
g
i
n
e
e
ri
n
g
R
e
se
a
rc
h
,
v
o
l
.
1
3
,
n
o
.
1
4
,
p
p
.
1
1
6
8
6
–
1
1
6
9
2
,
2
0
1
8
.
[
6
]
Y
.
A
.
Ef
f
e
n
d
i
,
A
.
S
o
f
i
a
h
,
N
.
A
.
H
i
d
a
y
a
t
,
A
.
S
.
E
b
r
i
e
,
a
n
d
Z.
H
a
m
z
a
h
,
“
P
r
e
d
i
c
t
i
n
g
v
u
l
n
e
r
a
b
i
l
i
t
y
f
o
r
b
r
a
i
n
t
u
mo
r
:
d
a
t
a
-
d
r
i
v
e
n
a
p
p
r
o
a
c
h
u
t
i
l
i
z
i
n
g
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
I
n
d
o
n
e
si
a
n
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
3
5
,
n
o
.
3
,
p
p
.
1
5
7
9
–
1
5
8
9
,
S
e
p
.
2
0
2
4
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
e
c
s
.
v
3
5
.
i
3
.
p
p
1
5
7
9
-
1
5
8
9
.
[
7
]
A
.
S
.
A
h
m
e
d
a
n
d
H
.
A
.
S
a
l
a
h
.
"
Th
e
I
o
T
a
n
d
r
e
g
i
s
t
r
a
t
i
o
n
o
f
M
R
I
b
r
a
i
n
d
i
a
g
n
o
s
i
s
b
a
se
d
o
n
g
e
n
e
t
i
c
a
l
g
o
r
i
t
h
m
a
n
d
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
d
o
n
e
s
i
a
n
J
o
u
rn
a
l
o
f
E
l
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
5
,
n
o
.
1
,
p
p
.
2
7
3
-
2
8
0
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
e
c
s.
v
2
5
.
i
1
.
p
p
2
7
3
-
2
8
0
.
[
8
]
S
e
d
e
e
q
,
N
i
h
a
y
a
.
"
C
N
N
-
B
a
se
d
S
e
g
me
n
t
a
t
i
o
n
a
n
d
D
e
t
e
c
t
i
o
n
o
f
B
r
a
i
n
T
u
m
o
r
s
M
R
I
I
mag
e
s:
A
R
e
v
i
e
w
.
"
T
h
e
I
n
d
o
n
e
si
a
n
J
o
u
rn
a
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
1
3
,
n
o
.
3
(
2
0
2
4
)
.
[
9
]
J.
R
e
d
mo
n
,
S
.
D
i
v
v
a
l
a
,
R
.
G
i
r
sh
i
c
k
,
a
n
d
A
.
F
a
r
h
a
d
i
,
“
Y
o
u
o
n
l
y
l
o
o
k
o
n
c
e
:
U
n
i
f
i
e
d
,
r
e
a
l
-
t
i
m
e
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
,
”
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
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r
Vi
s
i
o
n
a
n
d
Pa
t
t
e
rn
Re
c
o
g
n
i
t
i
o
n
,
J
u
n
.
2
0
1
6
,
v
o
l
.
2
0
1
6
-
D
e
c
e
m
b
e
r
,
p
p
.
7
7
9
–
7
8
8
,
d
o
i
:
1
0
.
1
1
0
9
/
C
V
P
R
.
2
0
1
6
.
9
1
.
[
1
0
]
D
.
R
e
i
s
,
J
.
K
u
p
e
c
,
J
.
H
o
n
g
,
a
n
d
A
.
D
a
o
u
d
i
,
“
R
e
a
l
-
t
i
m
e
f
l
y
i
n
g
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
w
i
t
h
Y
O
L
O
v
8
,
”
a
r
X
i
v
p
r
e
p
r
i
n
t
a
r
X
i
v
:
2
3
0
5
.
0
9
9
7
2
,
2
0
2
3
.
[
1
1
]
“
A
w
e
so
m
e
-
y
o
l
o
v
8
-
m
o
d
e
l
s
,”
g
i
t
h
u
b
r
e
p
o
si
t
o
r
y
,
2
0
2
3
[
O
n
l
i
n
e
]
.
A
v
a
i
l
a
b
l
e
:
h
t
t
p
s
:
/
/
g
i
t
h
u
b
.
c
o
m/
k
e
r
e
mb
e
r
k
e
/
a
w
e
so
me
-
y
o
l
o
v
8
-
mo
d
e
l
s
[
1
2
]
J.
T
e
r
v
e
n
,
D
.
M
.
C
ó
r
d
o
v
a
-
Esp
a
r
z
a
,
a
n
d
J.
A
.
R
o
mer
o
-
G
o
n
z
á
l
e
z
,
“
A
c
o
mp
r
e
h
e
n
si
v
e
r
e
v
i
e
w
o
f
Y
O
LO
a
r
c
h
i
t
e
c
t
u
r
e
s
i
n
c
o
m
p
u
t
e
r
v
i
s
i
o
n
:
f
r
o
m
Y
O
LO
v
1
t
o
Y
O
LO
v
8
a
n
d
Y
O
LO
-
N
A
S
,
”
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
K
n
o
w
l
e
d
g
e
Ex
t
r
a
c
t
i
o
n
,
v
o
l
.
5
,
n
o
.
4
,
p
p
.
1
6
8
0
–
1
7
1
6
,
N
o
v
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
ma
k
e
5
0
4
0
0
8
3
.
[
1
3
]
K
.
P
.
K
u
m
a
r
,
B
.
R
.
P
r
a
t
h
a
p
,
M
.
M
.
T
h
i
r
u
t
h
u
v
a
n
a
t
h
a
n
,
H
.
M
u
r
t
h
y
,
a
n
d
V
.
J
.
P
i
l
l
a
i
,
“
S
e
c
u
r
e
a
p
p
r
o
a
c
h
t
o
sh
a
r
i
n
g
d
i
g
i
t
i
z
e
d
m
e
d
i
c
a
l
d
a
t
a
i
n
a
c
l
o
u
d
e
n
v
i
r
o
n
me
n
t
,
”
D
a
t
a
S
c
i
e
n
c
e
a
n
d
M
a
n
a
g
e
m
e
n
t
,
v
o
l
.
7
,
n
o
.
2
,
p
p
.
1
0
8
–
1
1
8
,
Ju
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
d
sm
.
2
0
2
3
.
1
2
.
0
0
1
.
[
1
4
]
A
.
Ta
h
i
r
e
t
a
l
.
,
“
A
s
y
st
e
ma
t
i
c
r
e
v
i
e
w
o
n
c
l
o
u
d
st
o
r
a
g
e
mec
h
a
n
i
sms
c
o
n
c
e
r
n
i
n
g
e
-
h
e
a
l
t
h
c
a
r
e
sy
s
t
e
ms
,
”
S
e
n
s
o
rs
(
S
w
i
t
zerl
a
n
d
)
,
v
o
l
.
2
0
,
n
o
.
1
8
,
p
p
.
1
–
3
2
,
S
e
p
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
0
1
8
5
3
9
2
.
[
1
5
]
W
.
L
i
e
t
a
l
.
,
“
M
a
g
n
e
t
i
c
r
e
s
o
n
a
n
c
e
i
m
a
g
e
(
M
R
I
)
sy
n
t
h
e
si
s
f
r
o
m
b
r
a
i
n
c
o
m
p
u
t
e
d
t
o
m
o
g
r
a
p
h
y
(
C
T)
i
ma
g
e
s
b
a
s
e
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
met
h
o
d
s
f
o
r
ma
g
n
e
t
i
c
r
e
so
n
a
n
c
e
(
M
R
)
-
g
u
i
d
e
d
r
a
d
i
o
t
h
e
r
a
p
y
,
”
Q
u
a
n
t
i
t
a
t
i
v
e
I
m
a
g
i
n
g
i
n
M
e
d
i
c
i
n
e
a
n
d
S
u
r
g
e
r
y
,
v
o
l
.
1
0
,
n
o
.
6
,
p
p
.
1
2
2
3
–
1
2
3
6
,
Ju
n
.
2
0
2
0
,
d
o
i
:
1
0
.
2
1
0
3
7
/
Q
I
M
S
-
19
-
8
8
5
.
Evaluation Warning : The document was created with Spire.PDF for Python.