I
A
E
S
I
n
t
e
r
n
at
io
n
al
Jou
r
n
al
of
A
r
t
if
ic
ia
l
I
n
t
e
ll
ig
e
n
c
e
(
I
J
-
AI
)
V
ol
.
1
4
, N
o.
1
,
F
e
br
ua
r
y
202
5
, pp.
592
~
601
I
S
S
N
:
2252
-
8938
,
D
O
I
:
10.11591/
ij
a
i.
v
1
4
.i
1
.pp
592
-
601
592
Jou
r
n
al
h
om
e
page
:
ht
tp
:
//
ij
ai
.
ia
e
s
c
or
e
.c
om
D
e
e
p
l
e
ar
n
i
n
g ar
c
h
i
t
e
c
t
u
r
e
s f
o
r
l
oc
at
i
on
an
d
i
d
e
n
t
i
f
i
c
at
i
on
i
n
st
or
age
sys
t
e
m
s
A
n
n
y A
s
t
r
id
E
s
p
it
ia
C
u
b
il
lo
s
1
,
R
ob
in
s
on
Ji
m
é
n
e
z
-
M
or
e
n
o
2
,
E
s
p
e
r
an
z
a R
od
r
íg
u
e
z
C
ar
m
on
a
1
1
In
dus
t
r
i
a
l
E
ngi
ne
e
r
i
ng P
r
ogr
a
m
,
F
a
c
ul
t
y
of
E
ngi
ne
e
r
i
ng
,
U
ni
ve
r
s
i
t
y
M
i
l
i
t
a
r
N
u
e
va
G
r
a
na
da
,
B
ogot
á
,
C
ol
om
bi
a
2
M
e
c
ha
t
r
oni
c
E
ngi
ne
e
r
i
ng P
r
ogr
a
m
,
F
a
c
ul
t
y
of
E
ngi
ne
e
r
i
ng
,
U
ni
ve
r
s
i
t
y
M
i
l
i
t
a
r
N
ue
va
G
r
a
na
da
,
B
ogot
á
,
C
ol
om
bi
a
A
r
t
ic
le
I
n
f
o
A
B
S
T
R
A
C
T
A
r
ti
c
le
h
is
to
r
y
:
R
e
c
e
iv
e
d
A
pr
2
,
2024
R
e
vi
s
e
d
J
ul
27
,
2024
A
c
c
e
pt
e
d
A
ug
30
,
2024
This document exposes the application
of two deep lea
rning models ba
sed on
ResNet
-
18
architectures,
intended
for
the
location
and
identificat
ion
of
products
in
storage
areas.
One
model
obeys
a
tree
structure
and
the
other
a
structure
under
an
ouroboron
cyc
le.
The
performance
of
both
mo
dels
is
evaluated
using
the
metrics
of
training
time,
processin
g
time
and
le
vel
of
learning
precision,
which
allows
recommen
dations
to
be
made
reg
arding
which
one
should
be
used
for
order
preparation
purposes,
based
on
mul
tile
vel
feature
extraction.
The
total
training
time
of
the
first
model
is
34.65
minutes
and
the
second
40.43
minutes.
The
analysis
of
results
allowed
the
de
tection
parameters to be adjusted, fina
lly with the refined models, through
con
fusion
matrices,
precisio
n
results
greater
than
90%
and
processing
tim
es
are
obtained,
which
for
model
1
is
6.8565
seconds
and
for
model
2
is
4.884
seconds.
For
practical
purposes,
training
times
are
not
relevant,
as
are
the
precision
and
processing
times
for
selecting
the
most
co
nvenient
model
according t
o the end
user'
s object
ives.
K
e
y
w
o
r
d
s
:
A
c
c
ur
a
nc
y
C
onvolut
io
na
l
ne
twor
ks
D
e
e
p l
e
a
r
ni
ng
P
r
oduc
ts
i
de
nt
if
ic
a
ti
on
P
r
oduc
ts
l
oc
a
ti
on
S
to
r
a
ge
s
ys
te
m
s
This is an
open
acce
ss artic
le unde
r the
CC BY
-
SA
license.
C
or
r
e
s
pon
di
n
g A
u
th
or
:
A
nny As
tr
id
E
s
pi
ti
a
C
ubi
ll
os
I
ndus
tr
ia
l
E
ngi
ne
e
r
in
g P
r
og
r
a
m
,
F
a
c
ul
ty
of
E
ngi
ne
e
r
in
g
, U
ni
ve
r
s
id
a
d M
il
it
a
r
N
ue
va
G
r
a
na
da
C
a
r
r
e
r
a
11 #101
-
80, B
ogot
á
, C
ol
om
bi
a
E
m
a
il
:
a
nny.e
s
pi
ti
a
@
uni
m
il
it
a
r
.e
du.c
o
1.
I
N
T
R
O
D
U
C
T
I
O
N
C
ur
r
e
nt
ly
,
c
onc
e
pt
s
a
nd
im
pl
e
m
e
nt
a
ti
on
of
in
dus
tr
y
4.0
a
r
e
a
r
e
a
li
ty
in
pr
oduc
ti
on
a
t
va
r
io
us
le
ve
ls
[
1]
,
w
he
r
e
in
te
gr
a
ti
on
of
da
ta
a
na
ly
s
is
,
s
e
n
s
or
s
a
nd
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
go
h
a
nd
in
ha
nd
w
it
h
s
c
h
e
m
e
s
th
a
t
r
e
vol
ut
io
ni
z
e
i
ndus
tr
ia
l
pr
oc
e
s
s
e
s
,
s
uc
h a
s
i
nc
lu
s
io
n of
r
obot
s
[
2]
. B
e
in
g a
bl
e
t
o m
e
a
s
ur
e
a
nd
a
c
t
on output
s
of
a
pr
oduc
ti
on
li
ne
a
r
e
ta
s
ks
in
he
r
e
nt
to
d
e
ve
lo
pm
e
nt
of
in
dus
tr
y 4.0
a
nd
in
te
r
ne
t
of
th
in
gs
(
I
oT
)
-
ba
s
e
d s
ys
te
m
s
[
3]
.
A
ll
o
f
th
is
a
ll
ow
s
a
ut
om
a
ti
on
in
p
r
oduc
ti
on
li
ne
s
th
a
t
tr
a
ns
la
te
s
in
to
lo
w
e
r
c
os
ts
a
nd
ti
m
e
s
,
unde
r
th
e
s
m
a
r
t
f
a
c
to
r
y
c
onc
e
pt
[
4]
.
S
to
r
a
ge
of
pr
oduc
t
s
a
nd
th
e
ir
lo
c
a
ti
on
w
it
hi
n
c
ol
le
c
ti
on
s
it
e
a
r
e
ta
s
k
s
th
a
t
c
a
n
b
e
a
ut
om
a
te
d
gi
ve
n
hi
gh
vol
um
e
of
de
m
a
nd
a
nd
s
ta
nda
r
di
z
a
ti
on
of
th
e
p
r
o
c
e
s
s
.
I
n
th
e
c
a
s
e
of
a
n
a
ut
om
a
te
d
s
ys
te
m
th
a
t
a
ll
ow
s
th
is
ta
s
k
to
be
a
c
hi
e
ve
d,
it
m
us
t
id
e
nt
if
y
th
e
s
to
r
a
g
e
lo
c
a
ti
on
a
nd
di
s
c
r
im
in
a
te
e
a
c
h
pr
oduc
t
a
m
ong
th
e
pos
s
ib
le
one
s
t
ha
t
a
r
e
s
to
r
e
d, t
hi
s
a
s
pr
e
li
m
in
a
r
y a
c
ti
vi
ti
e
s
f
or
or
de
r
pr
e
pa
r
a
ti
on
.
I
n
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
,
th
e
us
e
of
de
e
p
le
a
r
ni
ng
a
lg
or
it
hm
s
a
ppl
ie
d
to
m
a
nuf
a
c
tu
r
in
g
pr
oc
e
s
s
e
s
in
in
dus
tr
y
s
ta
nds
out
[
5]
.
W
he
r
e
th
e
in
te
gr
a
ti
on
of
de
e
p
le
a
r
n
in
g
w
it
h
r
obot
ic
a
ut
om
a
ti
on
s
ys
te
m
s
a
ll
ow
s
a
ut
om
a
ti
ng
f
a
c
to
r
y
pr
oc
e
s
s
e
s
s
uc
h
a
s
pr
oduc
t
in
ve
nt
or
ie
s
[
6]
.
T
he
s
e
a
lg
or
it
hm
s
to
da
y
de
m
a
r
c
a
te
th
e
im
pl
e
m
e
nt
a
ti
on
of
I
ndus
tr
y
4.0
s
c
he
m
e
s
,
ba
s
e
d
on
te
c
hni
qu
e
s
f
or
m
ul
ti
di
m
e
ns
io
na
l
pa
tt
e
r
n
r
e
c
ogni
ti
on
[
7]
,
w
he
r
e
c
onvolut
io
na
l
ne
ur
a
l
ne
twor
ks
(
C
N
N
)
a
r
e
pr
e
s
e
nt
e
d
a
s
th
e
pr
e
dom
in
a
nt
te
c
hni
que
.
A
c
onvolut
io
na
l
ne
twor
k i
s
a
l
a
ye
r
-
ba
s
e
d l
e
a
r
ni
ng a
r
c
hi
te
c
tu
r
e
, w
he
r
e
t
he
s
im
pl
e
s
t
s
tr
uc
tu
r
e
i
s
t
he
uni
on of
c
onvolut
io
n, l
in
e
a
r
r
e
c
ti
f
ic
a
ti
on,
a
nd
pool
in
g
ope
r
a
ti
on
s
f
or
r
e
s
iz
in
g.
T
h
e
s
e
a
r
e
t
w
o
-
di
m
e
ns
io
na
l
or
de
r
ope
r
a
ti
ons
c
a
r
r
ie
d
out
th
r
ough ma
tr
ix
-
ty
pe
ope
r
a
ti
ons
[
8]
, hi
gh
ly
e
f
f
ic
ie
nt
f
or
de
te
c
ti
o
n of
obj
e
c
ts
i
n i
m
a
ge
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
e
p l
e
ar
ni
ng ar
c
hi
te
c
tu
r
e
s
f
or
l
oc
at
io
n and ide
nt
if
ic
at
io
n i
n s
t
or
age
s
y
s
te
m
s
(
A
nny
A
s
tr
id
E
s
pi
ti
a C
ubi
ll
os
)
593
O
bj
e
c
t
de
te
c
ti
on
c
or
r
e
s
ponds
to
lo
c
a
ti
ng
a
ll
pos
it
io
ns
of
obj
e
c
ts
to
be
ve
r
if
ie
d
in
a
n
e
nt
r
y
th
a
t
is
de
li
m
it
e
d i
n boxe
s
a
nd l
a
b
e
le
d a
c
c
or
di
ng t
o c
a
te
gor
y t
o w
hi
c
h
i
t
be
lo
ngs
[
9]
. A
va
r
ie
ty
of
a
ppl
ic
a
ti
on
s
e
m
pl
oy
de
e
p
le
a
r
ni
ng
f
or
I
ndus
tr
y
4.0,
s
uc
h
a
s
c
om
put
e
r
vi
s
io
n
-
ba
s
e
d
s
upe
r
m
a
r
ke
t
m
e
r
c
ha
ndi
s
e
m
a
na
ge
m
e
nt
us
in
g
C
N
N
,
Y
O
L
O
v3,
a
nd
K
e
r
a
s
[
10]
.
G
iv
e
n
a
dva
nc
e
m
e
nt
of
de
e
p
le
a
r
ni
ng
m
ode
ls
a
nd
a
r
ti
f
ic
ia
l
in
te
ll
ig
e
nc
e
te
c
hnol
ogi
e
s
,
r
e
a
l
-
ti
m
e
obj
e
c
t
de
te
c
ti
on
m
e
th
ods
a
r
e
a
n
a
r
e
a
of
s
tu
dy
to
im
pr
ove
th
e
ir
e
f
f
e
c
ti
ve
ne
s
s
a
nd
pot
e
nt
ia
l
f
or
pr
a
c
ti
c
a
l
a
ppl
ic
a
ti
ons
, i
n
c
lu
di
ng obje
c
ts
w
it
h g
e
om
e
tr
ic
di
ve
r
s
it
y a
nd
va
r
ie
ty
of
opt
ic
a
l
pr
ope
r
ti
e
s
[
11]
.
T
hi
s
is
r
e
f
le
c
te
d
in
m
a
na
ge
m
e
nt
of
c
om
pone
nt
s
in
s
uppl
y
c
ha
in
s
in
a
ut
om
ot
iv
e
in
dus
tr
y
a
nd
m
a
nuf
a
c
tu
r
in
g
in
dus
tr
y,
f
o
r
e
xa
m
pl
e
,
to
de
te
c
t
a
s
s
e
m
bl
y
pa
r
ts
[
12]
.
A
ls
o,
in
id
e
nt
if
ic
a
ti
on
o
f
obj
e
c
ts
w
it
h
de
f
or
m
e
d
pa
r
ts
us
in
g
g
e
om
e
tr
ic
r
e
s
tr
ic
ti
on
a
nd
pe
na
lt
y
[
13]
a
nd
in
c
ons
tr
uc
ti
on
in
dus
tr
y
by
c
a
te
gor
iz
in
g
obj
e
c
ts
by w
or
ke
r
, m
a
te
r
ia
l,
m
a
c
hi
ne
a
nd de
s
ig
n
[
14]
.
O
bj
e
c
t
de
te
c
ti
on
p
e
r
f
or
m
a
nc
e
ha
s
im
pr
o
ve
d
t
hr
o
ugh
u
s
e
of
l
a
t
e
s
t
ge
ne
r
a
t
io
n
m
od
e
l
s
b
a
s
e
d
on
bo
th
s
i
ngl
e
-
s
t
a
g
e
a
n
d
t
w
o
-
s
ta
ge
d
e
e
p
le
a
r
n
in
g
[
12]
,
[
1
5]
.
A
lt
h
oug
h
f
a
c
il
it
i
e
s
o
f
s
y
s
t
e
m
s
b
a
s
e
d
on
in
du
s
tr
y
4
.0
,
t
od
a
y
th
e
y
c
a
n
h
e
l
p
pr
e
v
e
n
t
a
n
d/
o
r
c
o
r
r
e
c
t
r
i
s
k
s
it
u
a
ti
on
s
i
n
w
or
k
e
r
s
[
16]
,
e
v
e
n
m
u
s
c
ul
o
s
ke
l
e
t
a
l
r
is
k
s
f
r
om
r
e
p
e
ti
ti
v
e
ta
s
k
s
r
e
m
a
in
hi
gh
[
1
7]
.
O
ne
of
f
a
c
to
r
s
of
t
hi
s
ty
pe
o
f
r
is
k i
s
t
a
s
k
of
s
t
or
in
g
go
od
s
,
f
or
e
x
a
m
pl
e
s
t
a
c
ki
ng
b
ox
e
s
,
a
ta
s
k
th
a
t
c
a
n
b
e
a
ut
om
a
te
d.
T
he
r
e
f
o
r
e
,
th
e
o
bj
e
c
ti
v
e
of
th
i
s
a
r
ti
c
le
i
s
t
o
d
e
s
ig
n
a
nd
e
v
a
l
ua
te
two
p
a
tt
e
r
n
r
e
c
o
gni
t
io
n
m
e
t
hod
s
f
or
id
e
nt
if
y
in
g
box
e
s
in
i
nd
us
tr
i
a
l
e
nv
ir
on
m
e
nt
s
,
w
hi
c
h
c
a
n
b
e
m
a
ni
pul
a
t
e
d
by
r
o
bot
ic
a
g
e
n
ts
i
n
a
s
e
c
on
d
pha
s
e
.
F
or
th
is
,
c
onv
ol
u
ti
o
n
a
l
n
e
tw
or
k
s
a
r
e
us
e
d
w
it
h
d
e
te
c
t
io
n
of
r
e
g
io
n
-
b
a
s
e
d
c
o
nv
ol
u
ti
o
na
l
ne
ur
a
l
n
e
tw
or
k
s
(
R
C
N
N
)
r
e
gi
o
n
s
[
18]
,
b
a
s
e
d
on
m
ul
ti
l
e
v
e
l
d
e
t
e
c
ti
on
of
pa
tt
e
r
n
s
,
a
t
f
ir
s
t
l
e
v
e
l
t
he
d
e
t
e
c
ti
on
of
bo
xe
s
a
nd
a
t
s
e
c
on
d
le
v
e
l
t
h
e
de
te
c
t
io
n
of
p
a
r
ti
c
u
la
r
c
h
a
r
a
c
t
e
r
i
s
t
ic
s
t
h
a
t
a
l
lo
w
t
he
m
t
o
be
c
l
a
s
s
if
i
e
d
.
T
hi
s
a
r
ti
c
le
is
di
vi
de
d
in
to
f
our
s
e
c
ti
on
s
,
th
e
f
ir
s
t
w
it
h
th
e
in
t
r
oduc
ti
on
to
th
e
s
ta
te
of
th
e
a
r
t
a
nd
pr
opos
e
d de
ve
lo
pm
e
nt
. T
he
s
e
c
ond
s
e
c
ti
on pr
e
s
e
nt
s
t
he
t
w
o pr
opos
e
d m
e
th
ods
. T
he
t
hi
r
d s
e
c
ti
on pr
e
s
e
nt
s
t
he
a
na
ly
s
is
a
nd di
s
c
us
s
io
n of
r
e
s
ul
ts
.
F
in
a
ll
y t
he
c
onc
lu
s
io
ns
r
e
a
c
h
e
d a
r
e
pr
e
s
e
nt
e
d.
2.
M
E
T
H
O
D
I
n
or
de
r
to
id
e
nt
if
y
a
nd
lo
c
a
te
a
pr
oduc
t
(
box
)
w
it
hi
n
a
s
to
r
a
ge
a
r
e
a
s
uc
h
a
s
a
s
he
lf
,
s
ta
r
ti
ng
poi
nt
is
to
e
s
ta
bl
is
h a
t
e
s
t
s
c
e
na
r
io
i
n w
hi
c
h r
e
gi
on
-
ba
s
e
d de
e
p l
e
a
r
ni
ng ne
twor
ks
w
il
l
be
us
e
d, s
te
p
s
t
o f
ol
lo
w
a
r
e
s
how
n
gr
a
phi
c
a
ll
y
in
F
ig
ur
e
1,
f
or
th
is
,
th
e
m
os
t
c
onve
ni
e
nt
pr
e
de
f
in
e
d
a
r
c
hi
te
c
tu
r
e
is
s
e
le
c
te
d
a
c
c
or
di
ng
to
c
ha
r
a
c
te
r
is
ti
c
s
to
be
di
s
c
r
im
in
a
te
d.
S
ubs
e
que
nt
ly
,
s
ol
ut
io
ns
a
r
e
de
s
ig
ne
d
th
a
t
r
e
s
pond
to
w
ha
t
is
r
e
qui
r
e
d:
id
e
nt
if
y e
a
c
h box a
nd i
ts
t
ype
.
T
o t
r
a
in
ne
twor
k, da
t
a
ba
s
e
i
s
bui
lt
, ne
twor
k a
r
c
h
it
e
c
tu
r
e
i
s
de
te
r
m
in
e
d, t
r
a
in
in
g
pa
r
a
m
e
te
r
s
a
r
e
s
e
le
c
te
d
a
s
th
e
m
o
s
t
a
ppr
opr
ia
te
opt
im
iz
e
r
,
a
nd
va
li
da
ti
on
te
s
ts
a
r
e
pe
r
f
or
m
e
d
to
id
e
nt
if
y
a
nd
doc
um
e
nt
e
r
r
or
s
.
F
in
a
ll
y,
th
e
m
ode
l
is
r
e
f
in
e
d
a
nd
th
e
m
e
tr
ic
s
o
f
pr
oc
e
s
s
in
g
ti
m
e
a
nd
le
ve
l
of
a
c
c
ur
a
c
y
in
f
in
a
l
r
e
c
ogni
ti
on a
r
e
c
a
lc
ul
a
te
d t
o gi
ve
a
r
e
c
om
m
e
nd
a
ti
on i
n t
hi
s
r
e
ga
r
d.
F
ig
ur
e
1. M
e
th
odol
ogy f
lo
w
c
ha
r
t
3.
R
E
S
U
L
T
S
A
N
D
D
I
S
C
U
S
S
I
O
N
C
ons
id
e
r
in
g
r
obus
t
de
v
e
lo
pm
e
nt
s
in
pr
e
de
f
in
e
d
C
N
N
-
ba
s
e
d
a
r
c
hi
te
c
tu
r
e
s
,
s
uc
h
a
s
R
e
s
N
e
t
-
18
[
18]
,
[
19]
,
R
e
s
N
et
-
50
[
20]
,
R
e
s
N
e
t
-
101
[
21]
,
a
nd
s
im
il
a
r
,
it
is
de
c
id
e
d
to
us
e
a
tr
a
ns
f
e
r
m
ode
l
of
le
a
r
ni
ng.
F
o
r
th
is
c
a
s
e
,
di
s
c
r
im
in
a
ti
on
of
s
to
r
a
ge
boxe
s
la
be
le
d
f
or
c
la
s
s
if
ic
a
ti
o
n
a
nd
lo
c
a
te
d
on
a
s
he
lf
is
r
e
qui
r
e
d.
F
or
it
s
e
va
lu
a
ti
on,
a
s
im
ul
a
ti
on
of
th
r
e
e
-
le
ve
l
s
to
r
a
ge
s
he
lv
e
s
a
nd
box
e
s
la
be
le
d
w
it
h
th
e
l
e
tt
e
r
s
A
,
B
, C
w
il
l
be
us
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
.
1
4
, N
o.
1
,
F
e
br
ua
r
y
202
5
:
592
-
601
594
a
s
s
how
n
in
F
ig
ur
e
2.
T
he
boxe
s
w
il
l
be
of
di
f
f
e
r
e
nt
ty
pe
s
th
a
t
a
ll
ow
,
if
ne
c
e
s
s
a
r
y,
to
ha
ve
di
f
f
e
r
e
nt
pr
oduc
ts
in
te
r
na
ll
y
or
s
a
m
e
pr
oduc
t
in
di
f
f
e
r
e
nt
qua
nt
it
ie
s
,
u
s
in
g
va
r
io
us
s
iz
e
s
of
boxe
s
.
T
a
ki
ng
in
to
a
c
c
ount
th
e
lo
w
num
be
r
of
c
ha
r
a
c
te
r
is
ti
c
s
to
be
di
s
c
r
im
in
a
te
d,
it
i
s
de
c
id
e
d
to
us
e
a
tr
a
ns
f
e
r
le
a
r
ni
ng
m
ode
l
to
c
e
nt
r
a
li
z
e
th
e
de
s
ig
n
in
th
e
m
os
t
f
a
vor
a
bl
e
gl
oba
l
a
lg
or
it
hm
ic
s
tr
uc
tu
r
e
a
nd
not
in
th
e
ne
twor
k
a
r
c
hi
te
c
tu
r
e
.
F
or
th
is
c
a
s
e
,
tr
a
in
in
g
ba
s
e
d
on
th
e
R
e
s
N
e
t
-
18
a
r
c
hi
te
c
tu
r
e
[
18]
,
is
c
hos
e
n,
a
s
ha
ll
ow
c
la
s
s
if
ic
a
ti
on
a
nd
lo
c
a
li
z
a
ti
on
a
r
c
hi
te
c
tu
r
e
t
ha
t
r
e
duc
e
s
t
he
c
om
put
a
ti
ona
l
c
os
t.
F
ig
ur
e
2. S
to
r
a
ge
s
he
lf
e
xa
m
pl
e
S
in
c
e
t
he
a
i
m
i
s
to
l
o
c
a
t
e
e
a
c
h
bo
x
on
t
h
e
s
he
lf
a
nd
s
u
bs
e
qu
e
nt
ly
id
e
nt
if
y
th
e
t
yp
e
of
b
ox
,
tw
o
p
a
r
ti
c
ul
a
r
s
o
lu
ti
on
s
a
r
e
pr
op
os
e
d.
F
ir
s
t
a
tr
e
e
s
tr
u
c
t
ur
e
w
he
r
e
two
R
e
s
N
e
t
-
1
8
a
r
c
h
it
e
c
tu
r
e
s
a
r
e
tr
a
in
e
d
[
22]
,
t
he
to
p
n
od
e
t
ha
t
w
il
l
d
e
t
e
c
t
b
ox
e
s
a
nd
a
f
te
r
th
a
t
th
e
ne
t
w
or
k of
t
he
bo
tt
o
m
n
od
e
t
ha
t
w
i
ll
de
te
c
t
th
e
t
yp
e
of
b
ox
.
S
e
c
ond
s
ol
ut
i
on
us
e
s
s
a
m
e
n
e
tw
or
k
w
i
th
jo
i
nt
tr
a
in
i
ng
of
b
ox
e
s
a
nd
b
ox
t
yp
e
l
a
b
e
l
,
w
he
r
e
ne
two
r
k
out
put
,
w
it
h
b
ox
id
e
nt
if
i
c
a
t
io
n,
is
f
e
d
b
a
c
k
f
or
l
a
t
e
r
id
e
n
ti
f
i
c
a
ti
on
of
bo
x
ty
pe
,
a
nd
a
g
a
in
unt
il
a
ll
t
ype
s
of
bo
x
e
s
pr
e
s
e
n
t
a
r
e
l
a
b
e
l
e
d
(
e
m
ul
a
ti
ng
a
n
our
o
bor
o
c
y
c
l
e
)
.
F
ig
ur
e
s
3(
a
)
a
n
d
3(
b)
r
e
s
p
e
c
ti
ve
ly
il
l
u
s
tr
a
te
s
f
lo
w
c
ha
r
t
of
pr
op
os
e
d
s
ol
ut
i
on
s
.
(
a
)
(
b)
F
ig
ur
e
3. F
lo
w
c
ha
r
t
of
(
a
)
t
r
e
e
s
tr
uc
tu
r
e
a
nd (
b)
our
obor
o
s
tr
uc
t
ur
e
F
or
tr
a
in
in
g
of
e
a
c
h
s
ol
ut
io
n,
a
da
ta
ba
s
e
is
us
e
d
th
a
t
is
bui
lt
ba
s
e
d
on
de
s
ir
e
d
obj
e
c
ti
ve
.
F
ig
ur
e
4
il
lu
s
tr
a
te
s
pa
r
t
of
da
ta
ba
s
e
of
box
s
to
r
a
ge
s
he
lv
e
s
,
w
he
r
e
box
e
s
of
di
f
f
e
r
e
nt
ty
pe
s
a
nd
s
iz
e
s
a
r
e
us
e
d.
E
a
c
h
im
a
ge
ha
s
a
r
e
s
ol
ut
io
n of
1
,
000
×
1
,
000 pixe
ls
a
nd 100 im
a
ge
s
of
t
hi
s
t
ype
a
r
e
u
s
e
d. F
ig
ur
e
5 pr
e
s
e
nt
s
a
por
ti
on
of
da
ta
ba
s
e
of
boxe
s
w
it
h
di
s
c
r
im
in
a
ti
on
la
be
ls
,
in
w
hi
c
h
c
a
s
e
boxe
s
of
ty
pe
A
,
B
,
or
C
w
il
l
be
r
e
c
ogni
z
e
d
w
it
hout
di
s
c
r
im
in
a
ti
ng
th
e
ir
s
iz
e
or
s
ha
p
e
.
E
a
c
h
im
a
ge
ha
s
a
r
e
s
ol
ut
io
n
of
224
×
224
pi
xe
l
s
(
R
e
s
N
e
t
-
18
in
put
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
e
p l
e
ar
ni
ng ar
c
hi
te
c
tu
r
e
s
f
or
l
oc
at
io
n and ide
nt
if
ic
at
io
n i
n s
t
or
age
s
y
s
te
m
s
(
A
nny
A
s
tr
id
E
s
pi
ti
a C
ubi
ll
os
)
595
a
nd
100
im
a
ge
s
of
th
is
ty
pe
a
r
e
us
e
d.
F
ig
ur
e
6
s
how
s
ne
twor
k
a
r
c
hi
te
c
tu
r
e
us
e
d,
w
he
r
e
it
is
im
por
ta
nt
to
hi
ghl
ig
ht
t
ha
t
R
e
s
N
e
t
-
18 ha
s
18 l
a
ye
r
s
w
it
h hyp
e
r
pa
r
a
m
e
te
r
s
de
f
in
e
d a
s
t
he
s
i
z
e
of
f
il
te
r
s
a
nd c
onvolut
io
n a
nd
pool
in
g
ope
r
a
ti
ons
[
8]
,
[
23]
.
T
he
r
e
f
or
e
,
in
c
onc
e
pt
of
tr
a
ns
f
e
r
le
a
r
ni
ng,
onl
y
th
e
ty
pe
of
ne
twor
k
tr
a
in
in
g
pa
r
a
m
e
te
r
s
is
de
te
r
m
in
e
d, a
s
i
ll
us
tr
a
te
d i
n T
a
bl
e
1.
F
ig
ur
e
4. D
a
ta
ba
s
e
e
xa
m
pl
e
of
boxe
s
on s
to
r
a
ge
s
he
lf
F
ig
ur
e
5. D
is
c
r
im
in
a
ti
on l
a
be
le
d boxe
s
da
ta
ba
s
e
F
ig
ur
e
6. R
e
s
N
e
t
-
18
a
r
c
hi
te
c
tu
r
e
I
n
th
is
c
a
s
e
,
im
a
ge
s
a
r
e
r
e
s
iz
e
d
to
224×
224
pi
xe
ls
be
c
a
us
e
it
is
in
put
s
iz
e
of
th
e
ne
twor
k
(
224×
224×
3)
.
R
e
s
ol
ut
io
n
r
e
duc
ti
on
g
e
ne
r
a
te
s
noi
s
e
in
tr
a
in
in
g
th
a
t
m
a
ke
s
A
d
a
m
opt
im
iz
e
r
pr
e
s
e
nt
b
e
tt
e
r
pe
r
f
or
m
a
nc
e
th
a
n
one
ba
s
e
d
on
s
to
c
ha
s
ti
c
gr
a
di
e
nt
de
s
c
e
nt
(
S
G
D
)
[
24]
.
F
ig
u
r
e
s
7
to
9
s
how
r
e
s
ul
ts
of
tr
a
in
in
g
ne
twor
ks
on
a
c
om
put
e
r
w
it
h a
n R
T
X
-
4070 G
P
U
w
it
h 8G
B
of
m
e
m
o
r
y.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
.
1
4
, N
o.
1
,
F
e
br
ua
r
y
202
5
:
592
-
601
596
T
a
bl
e
1.
T
r
a
in
in
g pa
r
a
m
e
te
r
s
P
a
r
a
m
e
t
e
r
S
e
t
t
i
ng va
l
ue
T
r
a
i
ni
ng O
pt
i
ons
a
da
m
M
i
ni
B
a
t
c
h S
i
z
e
1
I
ni
t
i
a
l
L
e
a
r
n R
a
t
e
10e
-
6
M
a
x E
poc
hs
100
V
a
l
i
da
t
i
on D
a
t
a
P
r
e
pr
oc
e
s
s
e
d T
r
a
i
ni
ng D
a
t
a
V
a
l
i
da
t
i
on F
r
e
que
nc
y
1
,
054
F
ig
ur
e
7. F
ir
s
t
tr
e
e
ne
twor
k (
t
r
a
in
in
g boxe
s
)
F
ig
ur
e
8. S
e
c
ond tr
e
e
ne
twor
k (
tr
a
in
in
g l
a
be
ls
)
F
ig
ur
e
9. O
ur
obor
o m
ode
l
ne
twor
k t
r
a
in
in
g
(
tr
a
in
in
g boxe
s
a
nd
la
be
ls
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
e
p l
e
ar
ni
ng ar
c
hi
te
c
tu
r
e
s
f
or
l
oc
at
io
n and ide
nt
if
ic
at
io
n i
n s
t
or
age
s
y
s
te
m
s
(
A
nny
A
s
tr
id
E
s
pi
ti
a C
ubi
ll
os
)
597
A
c
om
pa
r
is
on
of
tr
a
in
in
g
r
e
s
ul
ts
is
pr
e
s
e
nt
e
d
in
T
a
bl
e
2,
it
is
ob
s
e
r
ve
d
th
a
t
s
um
of
tr
a
in
in
g
ti
m
e
of
th
e
tr
e
e
m
ode
l
ne
twor
ks
is
34.65
m
in
ut
e
s
,
e
a
c
h
w
it
h
it
s
r
e
s
pe
c
ti
ve
da
ta
ba
s
e
(
boxe
s
a
nd
la
be
ls
)
.
F
or
th
e
our
obo
r
o
m
ode
l
ne
twor
k,
a
n
in
te
gr
a
te
d
da
ta
b
a
s
e
of
boxe
s
a
nd
l
a
be
ls
i
s
us
e
d,
tr
a
in
in
g
to
ok
40.43
m
in
ut
e
s
.
F
or
bot
h
c
a
s
e
s
,
pr
e
c
is
io
n
le
ve
ls
gr
e
a
t
e
r
th
a
n
95%
a
r
e
obt
a
in
e
d
in
le
a
r
ni
ng.
R
oo
t
m
e
a
n
s
qua
r
e
e
r
r
or
(
R
M
S
E
)
is
a
n
in
di
c
a
to
r
to
e
va
lu
a
te
qua
li
ty
of
pr
e
di
c
ti
ons
f
or
a
pa
r
ti
c
ul
a
r
da
ta
s
e
t
a
nd
no
t
be
twe
e
n
da
ta
s
e
ts
,
s
in
c
e
it
de
pe
nds
on
s
c
a
le
us
e
d,
it
s
how
s
to
w
h
a
t
e
xt
e
nt
pr
e
di
c
ti
ons
va
r
y
f
r
om
tr
ue
va
lu
e
s
u
s
in
g
E
uc
li
de
a
n
di
s
ta
nc
e
,
th
e
r
e
f
or
e
,
va
lu
e
s
c
lo
s
e
to
z
e
r
o
a
r
e
de
s
ir
e
d,
in
th
is
c
a
s
e
,
it
is
obs
e
r
ve
d
th
a
t
tr
a
in
in
g
is
m
or
e
a
c
c
ur
a
te
f
or
th
e
our
obor
o
m
ode
l
w
he
n
c
om
pa
r
e
d
w
it
h
a
ve
r
a
ge
v
a
lu
e
of
th
e
tr
e
e
m
od
e
l.
A
lm
a
la
q
a
nd
E
dw
a
r
ds
[
25]
c
om
pa
r
e
s
a
c
c
ur
a
c
y
r
e
s
ul
ts
us
in
g
R
M
S
E
f
or
va
r
io
us
a
ppl
ic
a
ti
ons
us
in
g
C
N
N
.
T
a
bl
e
2
a
ls
o
s
how
s
th
a
t
in
to
ta
l
7
,
600
it
e
r
a
ti
ons
w
e
r
e
c
a
r
r
ie
d
ou
t
in
t
he
t
r
e
e
m
ode
l
ne
twor
k w
hi
le
t
he
our
obor
o m
ode
l
ne
twor
k r
e
qui
r
e
d 8
,
300 f
or
i
ts
t
r
a
in
in
g.
T
a
bl
e
2.
T
r
a
in
in
g r
e
s
ul
ts
R
e
s
ul
t
s
T
r
e
e
m
ode
l
O
ur
obor
o m
ode
l
B
oxe
s
L
a
be
l
s
T
ot
a
l
/
A
ve
r
a
ge
V
a
l
i
da
t
i
on R
M
S
E
0.1075
0.075616
0.091558
0.059665
T
r
a
i
ni
ng t
i
m
e
19,35 m
i
nut
e
s
15,3 m
i
nut
e
s
34,65 m
i
nut
e
s
40,43 m
i
nut
e
s
I
t
e
r
a
t
i
ons
3
,
900
3
,
700
7
,
600
8
,
300
A
f
te
r
tr
a
in
in
g,
va
li
da
ti
on
te
s
ts
a
r
e
c
a
r
r
ie
d
out
w
it
h
va
r
io
us
r
e
s
ul
ts
s
uc
h
a
s
th
os
e
pr
e
s
e
nt
e
d
in
F
ig
ur
e
10. E
r
r
or
s
a
r
e
obs
e
r
ve
d i
n t
he
c
la
s
s
if
ic
a
ti
on
a
s
s
how
n
F
i
gur
e
10(
a
)
w
he
r
e
t
w
o of
t
he
ni
ne
boxe
s
a
r
e
not
la
be
le
d
c
or
r
e
c
tl
y,
th
is
e
r
r
or
is
de
r
iv
e
d
f
r
om
m
ode
l
us
e
d,
f
or
t
he
our
obor
o
c
a
s
e
,
a
s
w
il
l
be
e
xpl
a
in
e
d
la
te
r
.
A
not
he
r
e
r
r
or
i
s
e
vi
de
nt
i
n non
-
de
te
c
ti
on of
l
a
be
l
a
s
s
how
n
F
ig
ur
e
10(
b
)
. F
in
a
ll
y,
F
ig
ur
e
10
(
c
)
e
xpos
e
s
one
of
s
e
ve
r
a
l
c
a
s
e
s
of
c
or
r
e
c
t
c
la
s
s
if
ic
a
ti
ons
.
(
a
)
(
b)
(
c
)
F
ig
ur
e
10
.
O
ur
o
bor
o
s
tr
uc
tu
r
e
r
e
s
ul
ts
(
a
)
c
l
a
s
s
if
i
c
a
ti
o
n
w
it
h
e
r
r
or
s
,
(
b)
u
n
c
l
a
s
s
if
i
e
d
,
a
n
d
(
c
)
c
or
r
e
c
t
c
l
a
s
s
if
i
c
a
ti
on
F
ig
ur
e
11
s
how
s
c
a
us
e
of
m
is
la
be
li
ng,
va
li
da
ti
ng
w
it
h
s
e
ve
r
a
l
e
xa
m
pl
e
s
,
is
e
vi
de
nt
ly
de
r
iv
e
d
f
r
om
th
e
s
a
m
e
our
obor
o
m
ode
l.
T
ha
t
is
,
w
h
e
n
r
e
-
e
nt
e
r
in
g
box
d
e
te
c
ti
on
im
a
ge
f
r
om
f
ir
s
t
ti
m
e
to
ne
twor
k,
a
s
m
od
e
l
is
tr
a
in
e
d
w
it
h
box
d
a
ta
ba
s
e
,
it
a
ls
o
s
e
e
k
s
to
id
e
nt
if
y
th
e
s
e
.
F
ig
ur
e
11(
a
)
s
how
s
a
box
ty
pe
c
l
a
s
s
if
ic
a
ti
on
e
r
r
or
,
w
hi
le
F
ig
ur
e
11(
b)
s
how
s
a
box
id
e
nt
if
ic
a
ti
on
e
r
r
or
.
S
iz
e
of
box
it
us
e
s
f
or
obj
e
c
t
de
te
c
ti
on,
bi
a
s
e
d
to
th
e
m
in
im
um
of
la
be
l,
it
s
e
e
s
pa
r
ts
of
s
a
m
e
box
a
s
a
ne
w
box,
th
is
e
r
r
or
w
he
n
id
e
nt
if
i
e
d
is
e
li
m
in
a
te
d
w
it
h
s
of
twa
r
e
f
il
te
r
in
g
in
id
e
nt
if
ic
a
ti
on
of
la
be
l
ty
pe
,
d
e
le
ti
ng
e
ve
r
yt
hi
ng
th
a
t
is
m
a
r
ke
d
a
s
a
box
a
nd
e
s
ta
bl
is
hi
ng
c
yc
le
s
in
de
te
c
ti
on, one
ge
ne
r
a
l
f
or
t
he
box a
nd ne
xt
f
or
t
he
l
a
be
l
in
t
he
o
ur
obor
o m
ode
l.
E
r
r
or
of
no
n
-
d
e
t
e
c
ti
o
n of
l
a
b
e
l
oc
c
ur
s
be
c
a
u
s
e
t
r
a
in
in
g b
ox t
yp
e
i
s
t
a
k
e
n
a
t
2
24
×
2
24
pi
x
e
l
s
in
da
t
a
ba
s
e
.
W
h
e
n
e
a
c
h
bo
x
i
s
d
e
t
e
c
t
e
d
a
n
d
e
xt
r
a
c
te
d
f
r
o
m
or
i
gi
n
a
l
1,
000
×
1
,00
0
-
pi
xe
l
im
a
g
e
t
o
pr
e
s
e
r
v
e
r
e
s
ol
u
ti
o
n,
i
t
h
a
s
s
i
z
e
s
gr
e
a
te
r
t
ha
n
2
24
×
2
24
pi
xe
ls
, s
o
it
m
us
t
b
e
r
e
s
i
z
e
d t
o
n
e
tw
o
r
k
i
npu
t
s
i
z
e
. R
e
c
ta
ngu
la
r
b
ox
e
s
, w
h
e
n
r
e
s
iz
e
d
,
m
odi
f
y s
h
a
p
e
a
n
d w
id
th
of
l
a
b
e
l
,
m
a
k
in
g
it
no
t
l
e
g
ib
l
e
i
n s
o
m
e
c
a
s
e
s
,
w
hi
c
h w
a
s
s
ol
v
e
d
b
y e
nt
e
r
i
ng t
h
e
s
e
c
a
s
e
s
in
to
da
ta
ba
s
e
.
F
i
gur
e
12
il
l
us
tr
a
te
s
t
he
w
a
y
in
w
hi
c
h
t
he
e
r
r
or
w
a
s
e
v
id
e
n
t
.
I
n
t
hi
s
c
a
s
e
,
a
n
e
x
a
m
pl
e
w
a
s
ta
ke
n
f
r
om
d
a
t
a
b
a
s
e
to
va
l
id
a
t
e
c
o
r
r
e
c
t
c
la
s
s
i
f
ic
a
t
io
n
.
F
ig
ur
e
12(
a
)
a
n
d
l
a
t
e
r
it
i
s
r
e
s
iz
e
d
s
li
gh
tl
y
s
m
a
ll
e
r
a
nd
w
he
n
e
v
a
lu
a
t
e
d
th
r
ou
gh
ne
tw
or
k,
it
d
oe
s
not
di
s
ti
n
gui
s
h
l
a
b
e
l
a
n
d
th
e
r
e
f
or
e
do
e
s
n
ot
c
l
a
s
s
if
y
b
ox
a
s
s
h
ow
n
F
i
gur
e
12(
b)
.
F
ig
ur
e
13
r
e
la
te
s
de
t
e
c
ti
on
f
or
c
a
s
e
of
th
e
tr
e
e
-
ty
pe
m
ode
l,
th
e
r
e
a
r
e
two
s
c
e
na
r
io
s
,
th
e
f
ir
s
t
th
a
t
s
ho
w
s
a
n
e
xa
m
pl
e
of
c
or
r
e
c
t
ope
r
a
ti
on
in
F
ig
ur
e
13(
a
)
a
nd
th
e
s
e
c
ond
w
he
r
e
th
e
a
bs
e
nc
e
of
c
la
s
s
if
ic
a
ti
on
is
e
vi
de
nt
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
.
1
4
, N
o.
1
,
F
e
br
ua
r
y
202
5
:
592
-
601
598
a
s
s
how
n
in
F
ig
ur
e
13(
b)
,
th
is
e
r
r
o
r
w
a
s
s
ol
ve
d
in
s
a
m
e
w
a
y
a
s
e
r
r
or
of
non
-
de
te
c
ti
on
of
th
e
ou
r
obor
o
m
ode
l
la
be
l
s
in
c
e
i
t
w
a
s
due
t
o s
a
m
e
c
a
u
s
e
. O
nc
e
a
dj
us
tm
e
nt
s
ha
ve
be
e
n m
a
de
t
o de
te
c
te
d e
r
r
or
s
, r
e
f
in
e
d m
ode
ls
a
r
e
e
va
lu
a
te
d
us
in
g
c
onf
us
io
n
m
a
tr
ic
e
s
th
a
t
s
how
box
s
or
ti
ng
r
e
s
ul
ts
f
or
e
a
c
h
m
od
e
l,
a
s
s
e
e
n
in
F
ig
ur
e
14.
M
ode
l
1 i
n F
ig
ur
e
14(
a
)
s
how
s
c
or
r
e
s
ponds
t
o t
w
o t
r
e
e
ne
twor
k
s
a
nd mode
l
2 i
n F
ig
u
r
e
14(
b)
t
o t
he
our
obo
r
o
ty
pe
,
th
e
c
la
s
s
e
s
c
or
r
e
s
pond
to
c
or
r
e
c
t
la
be
li
ng
of
e
a
c
h
ty
p
e
o
f
box
A
=
1,
B
=
2
a
nd
C
=
3.
T
a
bl
e
3
il
lu
s
tr
a
te
s
s
ta
ti
s
ti
c
a
l
r
e
s
ul
ts
of
c
onf
us
io
n m
a
tr
ix
m
e
tr
ic
s
of
e
a
c
h m
ode
l
us
e
d, i
t
is
e
vi
de
nt
t
ha
t
m
ode
l
1 (
tr
e
e
)
ha
s
a
hi
ghe
r
pr
e
c
is
io
n,
how
e
ve
r
,
bot
h
m
ode
ls
pr
e
s
e
nt
a
hi
gh
pr
e
c
is
io
n
th
a
t
e
x
c
e
e
ds
90%
.
T
he
r
e
i
s
a
s
ig
ni
f
ic
a
nt
di
f
f
e
r
e
nc
e
in
th
e
c
l
a
s
s
if
ic
a
ti
on
a
nd
lo
c
a
li
z
a
ti
on
ti
m
e
of
e
a
c
h
m
ode
l,
f
or
m
ode
l
1
it
is
6.8565
s
e
c
onds
,
w
hi
le
m
ode
l
2
(
our
obor
o)
i
s
4.884 s
e
c
onds
, a
lm
os
t
two s
e
c
onds
m
or
e
.
(
a
)
(
b)
F
ig
ur
e
11.
C
la
s
s
if
ic
a
ti
on e
r
r
or
:
(
a
)
m
is
c
la
s
s
if
ic
a
ti
on
a
nd (
b)
ne
twor
k
de
te
c
ti
ons
e
r
r
or
(
a
)
(
b)
F
ig
ur
e
12. De
te
c
ti
on e
r
r
or
due
t
o s
iz
e
(
a
)
224×
224 pixe
ls
a
nd (
b
)
220×
220 pixe
ls
(
a
)
(
b)
F
ig
ur
e
13. T
r
e
e
s
tr
uc
tu
r
e
de
te
c
ti
on
(
a
)
c
or
r
e
c
t
c
la
s
s
if
ic
a
ti
on a
nd
(
b)
c
la
s
s
if
ic
a
ti
on
w
it
hout
de
te
c
ti
on
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
e
p l
e
ar
ni
ng ar
c
hi
te
c
tu
r
e
s
f
or
l
oc
at
io
n and ide
nt
if
ic
at
io
n i
n s
t
or
age
s
y
s
te
m
s
(
A
nny
A
s
tr
id
E
s
pi
ti
a C
ubi
ll
os
)
599
(a
)
(
b)
F
ig
ur
e
14. C
onf
us
io
n m
a
tr
ic
e
s
(
a
)
tr
e
e
m
ode
l
a
nd (
b)
our
obor
o
m
ode
l
T
a
bl
e
3. C
onf
us
io
n m
a
tr
ix
m
e
tr
ic
s
P
a
r
a
m
e
t
e
r
T
r
e
e
m
ode
l
(
1)
O
ur
obor
o m
ode
l
(
2)
P
r
e
c
i
s
i
on
0.96875
0.90909
R
e
c
a
l
l
0.93939
0.90909
F1
-
s
c
or
e
0.95385
0.90909
I
n ge
ne
r
a
l,
m
a
in
f
a
ls
e
pos
it
iv
e
s
t
ha
t
oc
c
ur
a
r
e
due
t
o
c
ha
nge
s
i
n r
e
s
ol
ut
io
n of
l
e
tt
e
r
s
us
e
d a
s
l
a
be
ls
t
o
di
s
c
r
im
in
a
te
th
e
ty
pe
of
box.
T
he
r
e
duc
ti
on
in
r
e
s
ol
ut
io
n,
a
s
m
e
nt
io
ne
d,
ge
ne
r
a
te
s
a
lo
s
s
of
c
ha
r
a
c
te
r
is
ti
c
s
th
a
t,
in
th
is
c
a
s
e
,
c
a
us
e
s
le
tt
e
r
s
to
be
c
onf
us
e
d
m
a
in
ly
w
it
h
B
,
a
s
in
di
c
a
te
d
by
th
e
c
onf
us
io
n
m
a
tr
ic
e
s
.
T
hi
s
c
a
s
e
is
c
a
r
r
ie
d
out
in
r
e
c
ta
ngul
a
r
boxe
s
th
a
t
pr
e
s
e
nt
a
gr
e
a
te
r
lo
s
s
of
c
ha
r
a
c
te
r
is
ti
c
s
w
he
n
th
e
y
a
r
e
r
e
s
iz
e
d
to
th
e
s
iz
e
of
ne
two
r
k
in
put
.
A
s
f
ut
ur
e
r
e
s
e
a
r
c
h,
th
e
c
ons
tr
uc
ti
on
of
a
lg
or
it
hm
s
is
pr
opos
e
d
th
a
t
a
ll
ow
not
onl
y
th
e
id
e
nt
if
ic
a
ti
on
of
boxe
s
a
nd
th
e
ir
ty
pe
s
,
but
a
l
s
o
a
c
hi
e
ve
th
e
in
te
r
pr
e
ta
ti
on
of
th
e
ty
pi
c
a
l
im
a
ge
s
us
e
d
a
s
gui
da
n
c
e
on t
he
ha
ndl
in
g t
ha
t
s
houl
d b
e
gi
ve
n t
o t
he
boxe
s
f
or
t
he
ir
m
a
ni
pul
a
ti
on a
nd s
ubs
e
que
nt
ly
t
he
i
nt
e
r
pr
e
ta
ti
on of
m
or
e
de
ta
il
e
d
la
be
l
s
lo
c
a
te
d
on
pr
oduc
ts
.
P
r
ogr
e
s
s
c
oul
d
a
ls
o
be
m
a
de
to
w
a
r
ds
e
xpa
ndi
ng
th
e
s
c
ope
by
c
ons
id
e
r
in
g t
he
ha
ndl
in
g of
boxe
s
a
nd pr
oduc
ts
by r
obot
ic
a
g
e
n
ts
f
or
t
he
pr
e
pa
r
a
ti
on o
f
or
de
r
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2252
-
8938
I
nt
J
A
r
ti
f
I
nt
e
ll
,
V
ol
.
1
4
, N
o.
1
,
F
e
br
ua
r
y
202
5
:
592
-
601
600
4.
C
O
N
C
L
U
S
I
O
N
I
t
is
c
onc
lu
de
d
th
a
t
th
e
two
s
ol
ut
io
ns
de
s
ig
ne
d
a
r
e
f
ul
ly
f
unc
ti
o
na
l
a
nd
a
ll
ow
th
e
obj
e
c
ti
ve
o
f
lo
c
a
ti
ng
a
nd
id
e
nt
if
yi
ng
pr
oduc
ts
s
to
r
e
d
in
la
be
le
d
boxe
s
on
s
he
lv
e
s
to
be
m
e
t
w
it
h
a
n
a
c
c
ur
a
c
y
gr
e
a
te
r
th
a
n
90%
a
nd
w
it
h pr
oc
e
s
s
in
g t
im
e
s
of
l
e
s
s
t
ha
n
7 s
e
c
ond
s
, i
n both c
a
s
e
s
, w
hi
c
h f
or
a
n a
ppl
ic
a
ti
on i
n a
r
e
a
l
e
nvi
r
onm
e
nt
i
s
a
m
a
na
ge
a
bl
e
r
e
s
pon
s
e
ti
m
e
.
D
e
s
pi
te
e
xi
s
ti
ng
di
f
f
e
r
e
nc
e
s
in
tr
a
i
ni
ng
ti
m
e
s
,
f
or
pr
a
c
ti
c
a
l
pur
pos
e
s
th
e
s
e
do
not
tu
r
n
out
to
be
r
e
le
va
nt
,
a
s
a
r
e
pr
e
c
is
io
n
a
nd
a
gi
li
ty
in
r
e
s
pon
s
e
ti
m
e
a
s
c
r
it
e
r
ia
f
or
s
e
le
c
ti
ng
th
e
m
o
s
t
c
onve
ni
e
nt
m
ode
l
in
l
ig
ht
of
e
nd us
e
r
'
s
obj
e
c
ti
ve
s
,
w
hi
c
h doe
s
a
ll
ow
us
t
o
c
onc
lu
de
w
hi
c
h m
ode
l
w
il
l
be
m
os
t
c
onve
ni
e
nt
in
a
ppl
ic
a
ti
on
e
nvi
r
onm
e
nt
.
I
t
is
e
vi
de
nt
th
a
t,
a
lt
hough
c
om
p
ut
a
ti
ona
l
c
os
t
is
hi
ghe
r
w
he
n
us
in
g
two
tr
e
e
ne
twor
k a
r
c
hi
te
c
tu
r
e
s
, by doubli
ng us
e
of
r
e
s
our
c
e
s
, pr
e
c
is
io
n a
c
hi
e
ve
d i
s
gr
e
a
te
r
, due
t
o s
pe
c
ia
li
z
a
ti
on of
t
he
pa
tt
e
r
n
le
a
r
ni
ng
of
e
a
c
h
ne
twor
k.
W
he
r
e
it
c
a
n
be
c
onc
lu
de
d
th
a
t
te
m
por
a
l
a
c
c
um
ul
a
ti
on
c
a
n
f
a
vor
us
e
of
th
e
our
obor
o a
r
c
hi
te
c
tu
r
e
, s
in
c
e
f
or
e
a
c
h
e
xe
c
ut
io
n t
w
o
s
e
c
ond
s
a
d
va
nt
a
ge
i
t
pr
e
s
e
nt
s
ove
r
t
he
t
r
e
e
m
ode
l
m
us
t
be
a
dde
d.
A
C
K
N
O
WL
E
D
G
E
M
E
N
T
S
P
r
oduc
t
de
r
iv
e
d
f
r
om
th
e
r
e
s
e
a
r
c
h
pr
oj
e
c
t
ti
tl
e
d
"
M
e
jo
r
a
m
ie
nt
o
de
pr
oc
e
s
os
de
c
ont
r
ol
de
in
ve
nt
a
r
io
y
a
li
s
ta
m
ie
nt
o
de
pe
di
dos
e
n
e
l
m
a
r
c
o
de
la
in
dus
tr
ia
4.0"
I
N
V
-
I
N
G
-
3970
f
in
a
nc
e
d
by
th
e
vi
c
e
-
r
e
c
to
r
f
o
r
r
e
s
e
a
r
c
h of
t
he
U
ni
ve
r
s
id
a
d
M
il
it
a
r
N
ue
va
G
r
a
na
da
, y
e
a
r
2024.
R
E
F
E
R
E
N
C
E
S
[
1]
F
.
E
.
H
a
bi
b,
H
.
B
noua
c
hi
r
,
M
.
C
he
r
gui
,
a
nd
A
.
A
m
m
oum
ou,
“
I
ndus
t
r
y
4.0
c
onc
e
pt
s
a
nd
i
m
pl
e
m
e
nt
a
t
i
on
c
ha
l
l
e
nge
s
:
l
i
t
e
r
a
t
ur
e
r
e
vi
e
w
,”
2022
9t
h
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
W
i
r
e
l
e
s
s
N
e
t
w
o
r
k
s
and
M
o
bi
l
e
C
om
m
uni
c
at
i
ons
,
W
I
N
C
O
M
2022
,
2022,
doi
:
10.1109/
W
I
N
C
O
M
55661.2022.9966456.
[
2]
X
.
X
i
n,
S
.
L
.
K
e
oh,
M
.
S
e
ve
gna
ni
,
M
.
S
a
e
r
be
c
k,
a
nd
T
.
P
.
K
hoo,
“
A
da
pt
i
v
e
m
ode
l
ve
r
i
f
i
c
a
t
i
on
f
or
m
odul
a
r
i
z
e
d
i
ndus
t
r
y
4.
0
a
ppl
i
c
a
t
i
ons
,”
I
E
E
E
A
c
c
e
s
s
, vol
. 10, pp. 125353
–
125364, 2022, doi
:
10.1109/
A
C
C
E
S
S
.2022.3225399.
[
3]
L
.
L
.
Y
a
dl
a
,
R
.
A
.
M
udr
a
ga
da
,
H
.
P
oka
,
a
nd
T
.
V
i
gne
s
h,
“
S
m
a
r
t
m
a
nuf
a
c
t
ur
i
ng
i
n
i
ndus
t
r
i
e
s
us
i
ng
i
nt
e
r
ne
t
o
f
t
hi
ngs
,”
2023
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
C
o
m
put
e
r
C
om
m
uni
c
at
i
on
and
I
nf
or
m
at
i
c
s
,
I
C
C
C
I
2023
,
2023,
doi
:
10.1109/
I
C
C
C
I
56745.2023.10128206.
[
4]
G
.
W
a
ng,
D
.
L
i
,
a
nd
H
.
S
ong,
“
A
f
or
m
a
l
a
na
l
yt
i
c
a
l
f
r
a
m
e
w
or
k
f
or
i
ot
-
ba
s
e
d
pl
ug
-
a
nd
pl
a
y
m
a
nuf
a
c
t
ur
i
ng
s
ys
t
e
m
c
ons
i
de
r
i
ng
pr
oduc
t
l
i
f
e
-
c
yc
l
e
de
s
i
gn
c
os
t
,”
I
E
E
E
T
r
ans
ac
t
i
ons
on
I
ndus
t
r
i
al
I
nf
or
m
at
i
c
s
,
vol
.
19,
no.
2,
pp.
1647
–
1654,
2023,
doi
:
10.1109/
T
I
I
.2022.3192681.
[
5]
B
.
W
u,
“
M
ot
i
on
c
ont
r
ol
a
l
gor
i
t
hm
f
or
a
ut
om
a
t
i
c
w
e
l
di
ng
of
c
om
pl
e
x
i
nt
e
r
s
e
c
t
i
ng
l
i
ne
j
oi
nt
s
ba
s
e
d
on
de
e
p
l
e
a
r
ni
ng,”
2023
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
M
e
c
hat
r
oni
c
s
,
I
oT
and
I
ndus
t
r
i
al
I
nf
or
m
at
i
c
s
,
I
C
M
I
I
I
2023
,
pp.
352
–
356,
2023,
doi
:
10.1109/
I
C
M
I
I
I
58949.2023.00073.
[
6]
S
. V
a
dda
di
, V
. S
r
i
ni
va
s
,
N
. A
.
R
e
ddy, H
.
G
i
r
i
s
h, D
.
R
a
j
ki
r
a
n, a
nd A
.
D
e
vi
pr
i
ya
, “
F
a
c
t
or
y i
nve
nt
or
y a
ut
om
a
t
i
on us
i
ng i
ndu
s
t
r
y
4.0
t
e
c
hnol
ogi
e
s
,”
2022
I
E
E
E
I
A
S
G
l
obal
C
onf
e
r
e
nc
e
on
E
m
e
r
gi
ng
T
e
c
hnol
og
i
e
s
,
G
l
obC
onE
T
2022
,
pp.
734
–
738,
2022,
doi
:
10.1109/
G
l
obC
onE
T
53749.2022.9872416.
[
7]
S
.
D
ong,
P
.
W
a
ng,
a
nd
K
.
A
bba
s
,
“
A
s
ur
ve
y
on
de
e
p
l
e
a
r
ni
ng
a
nd
i
t
s
a
ppl
i
c
a
t
i
ons
,”
C
om
put
e
r
Sc
i
e
nc
e
R
e
v
i
e
w
,
vol
.
40,
2021,
doi
:
10.1016/
j
.c
os
r
e
v.2021.100379.
[
8]
S
.
A
l
ba
w
i
,
T
.
A
.
M
oha
m
m
e
d,
a
nd
S
.
A
l
-
Z
a
w
i
,
“
U
nde
r
s
t
a
ndi
ng
of
a
c
onv
ol
ut
i
ona
l
ne
ur
a
l
ne
t
w
or
k,”
P
r
oc
e
e
di
ngs
of
2017
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
E
ngi
ne
e
r
i
ng
and
T
e
c
hnol
ogy
,
I
C
E
T
2017
,
vol
.
2018,
pp.
1
–
6,
2017,
doi
:
10.1109/
I
C
E
ngT
e
c
hnol
.2017.8308186.
[
9]
D
.
E
r
ha
n,
C
.
S
z
e
ge
dy,
A
.
T
os
he
v,
a
nd
D
.
A
ngue
l
ov,
“
S
c
a
l
a
bl
e
obj
e
c
t
de
t
e
c
t
i
o
n
us
i
ng
de
e
p
ne
ur
a
l
ne
t
w
or
ks
,”
P
r
oc
e
e
di
ngs
of
t
he
I
E
E
E
C
om
put
e
r
Soc
i
e
t
y
C
onf
e
r
e
nc
e
on
C
om
put
e
r
V
i
s
i
on
and
P
at
t
e
r
n
R
e
c
ogni
t
i
on
,
pp.
2155
–
2162,
2014,
doi
:
10.1109/
C
V
P
R
.2014.276.
[
10]
H
.
T
.
D
o
a
nd
V
.
C
.
P
ha
m
,
“
D
e
e
p
l
e
a
r
ni
ng
ba
s
e
d
goods
m
a
na
ge
m
e
nt
i
n
s
upe
r
m
a
r
ke
t
s
,”
J
our
nal
of
A
dv
anc
e
s
i
n
I
nf
or
m
at
i
on
T
e
c
hnol
ogy
, vol
. 12, no. 2, pp. 164
–
168, 2021, doi
:
10.12720/
j
a
i
t
.12.2.164
-
168.
[
11]
A
.
B
ör
ol
d,
M
.
T
e
uc
ke
,
A
.
R
us
t
,
a
nd
M
.
F
r
e
i
t
a
g,
“
D
e
e
p
l
e
a
r
ni
ng
-
ba
s
e
d
obj
e
c
t
r
e
c
ogni
t
i
on
f
or
c
ount
i
ng
c
a
r
c
om
pone
nt
s
t
o
s
uppor
t
ha
ndl
i
ng
a
nd
pa
c
ki
ng
pr
oc
e
s
s
e
s
i
n
a
ut
om
ot
i
ve
s
uppl
y
c
ha
i
n
s
,”
I
F
A
C
-
P
ape
r
s
O
nL
i
ne
,
vol
.
53,
pp.
10645
–
10650,
2020,
doi
:
10.1016/
j
.i
f
a
c
ol
.2020.12.2828.
[
12]
N
.
D
.
N
guye
n,
T
.
D
o,
T
.
D
.
N
go,
a
nd
D
.
D
.
L
e
,
“
A
n
e
va
l
ua
t
i
on
o
f
de
e
p
l
e
a
r
ni
ng
m
e
t
hods
f
or
s
m
a
l
l
obj
e
c
t
de
t
e
c
t
i
on,”
J
our
nal
of
E
l
e
c
t
r
i
c
al
and C
om
put
e
r
E
ngi
ne
e
r
i
ng
, vol
. 2020, 2020, doi
:
10.1155/
2020/
318
9691.
[
13]
W
.
O
uya
ng
e
t
al
.
,
“
D
e
e
pI
D
-
N
e
t
:
obj
e
c
t
de
t
e
c
t
i
on
w
i
t
h
de
f
or
m
a
bl
e
pa
r
t
ba
s
e
d
c
onvol
ut
i
ona
l
ne
ur
a
l
ne
t
w
or
ks
,”
I
E
E
E
T
r
ans
ac
t
i
ons
on P
at
t
e
r
n A
nal
y
s
i
s
and M
ac
hi
ne
I
nt
e
l
l
i
ge
nc
e
, vol
. 39, no. 7, pp. 1320
–
1334, 2017, doi
:
10.1109/
T
P
A
M
I
.2016.2587642.
[
14]
R
.
D
ua
n,
H
.
D
e
ng,
M
.
T
i
a
n,
Y
.
D
e
ng,
a
nd
J
.
L
i
n,
“
S
O
D
A
:
A
l
a
r
ge
-
s
c
a
l
e
ope
n
s
i
t
e
obj
e
c
t
de
t
e
c
t
i
on
da
t
a
s
e
t
f
or
de
e
p
l
e
a
r
ni
ng
i
n
c
ons
t
r
uc
t
i
on,”
A
ut
om
at
i
on i
n C
ons
t
r
uc
t
i
on
, vol
. 142, 2022, doi
:
10.1016/
j
.a
ut
c
on.2022.104499.
[
15]
A
.
S
ha
r
m
a
,
T
.
M
i
s
hr
a
,
J
.
K
uka
de
,
A
.
G
ol
w
a
l
ka
r
,
a
nd
H
.
T
om
a
r
,
“
O
bj
e
c
t
de
t
e
c
t
i
on
us
i
ng
T
e
ns
or
F
l
ow
,”
I
C
T
A
nal
y
s
i
s
and
A
ppl
i
c
at
i
ons
, pp. 343
–
352, 2023, doi
:
10.1007/
978
-
981
-
99
-
6568
-
7_31.
[
16]
F
. P
i
l
a
t
i
e
t
al
.
, “
O
pe
r
a
t
or
5.0:
e
nh
a
nc
i
ng t
he
phy
s
i
c
a
l
r
e
s
i
l
i
e
nc
e
of
w
or
ke
r
s
i
n a
s
s
e
m
bl
y l
i
ne
s
,”
2023
I
E
E
E
I
nt
e
r
nat
i
onal
W
or
k
s
hop
on
M
e
t
r
ol
ogy
f
or
I
ndus
t
r
y
4.0
and
I
oT
,
M
e
t
r
oI
nd4.0
and
I
oT
2023
,
pp.
177
–
182,
2023,
doi
:
10.1109/
M
e
t
r
oI
nd4.0I
oT
57462.2023.10180145.
[
17]
Y
.
W
.
C
ha
n,
T
.
H
.
H
ua
ng,
Y
.
T
.
T
s
a
n,
W
.
C
.
C
ha
n,
C
.
H
.
C
h
a
ng,
a
nd
Y
.
T
e
T
s
a
i
,
“
T
he
r
i
s
k
c
l
a
s
s
i
f
i
c
a
t
i
on
of
e
r
gonom
i
c
m
us
c
ul
os
ke
l
e
t
a
l
di
s
or
de
r
s
i
n
w
or
k
-
r
e
l
a
t
e
d
r
e
pe
t
i
t
i
ve
m
a
nua
l
ha
ndl
i
ng
ope
r
a
t
i
ons
w
i
t
h
de
e
p
l
e
a
r
ni
ng
a
ppr
oa
c
he
s
,
”
2020
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on P
e
r
v
as
i
v
e
A
r
t
i
f
i
c
i
al
I
nt
e
l
l
i
ge
nc
e
, I
C
P
A
I
2020
, pp. 268
–
271, 202
0, doi
:
10.1109/
I
C
P
A
I
51961.2020.00057.
[
18]
K
.
H
e
,
X
.
Z
ha
ng,
S
.
R
e
n,
a
nd
J
.
S
un,
“
I
de
nt
i
t
y
m
a
ppi
ng
s
i
n
d
e
e
p
r
e
s
i
du
a
l
ne
t
w
o
r
ks
,”
C
om
put
e
r
V
i
s
i
on
–
E
C
C
V
2016
(
E
C
C
V
2016)
,
pp. 630
–
645, 2016, doi
:
10.1007/
978
-
3
-
319
-
46493
-
0_38.
[
19]
F
. R
a
m
z
a
n
e
t
a
l
.
,
“
A
de
e
p
l
e
a
r
ni
n
g a
pp
r
oa
c
h
f
o
r
a
ut
om
a
t
e
d
di
a
gn
os
i
s
a
nd
m
u
l
t
i
-
c
l
a
s
s
c
l
a
s
s
i
f
i
c
a
t
i
on
of
a
l
z
he
i
m
e
r
’
s
d
i
s
e
a
s
e
s
t
a
ge
s
us
i
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
A
r
ti
f
I
nt
e
ll
I
S
S
N
:
2252
-
8938
D
e
e
p l
e
ar
ni
ng ar
c
hi
te
c
tu
r
e
s
f
or
l
oc
at
io
n and ide
nt
if
ic
at
io
n i
n s
t
or
age
s
y
s
te
m
s
(
A
nny
A
s
tr
id
E
s
pi
ti
a C
ubi
ll
os
)
601
r
e
s
t
i
n
g
-
s
t
a
t
e
f
M
R
I
a
nd
r
e
s
i
d
ua
l
ne
u
r
a
l
ne
t
w
o
r
ks
,”
J
o
ur
na
l
of
M
e
di
c
al
Sy
s
t
e
m
s
,
vol
. 4
4,
no
. 2
,
202
0,
do
i
:
1
0.1
00
7/
s
1
09
16
-
0
19
-
1
475
-
2.
[
20]
T
.
N
.
V
.
S
.
P
r
a
ve
e
n,
D
.
S
i
va
t
hm
i
ka
,
G
.
J
a
hna
vi
,
a
nd
J
.
B
ol
l
e
du,
“
A
n
i
n
-
de
pt
h
e
xpl
or
a
t
i
on
of
R
e
s
N
e
t
-
50
f
or
c
om
pl
e
x
e
m
ot
i
on
r
e
c
ogni
t
i
on
t
o
unr
a
ve
l
i
ng
e
m
ot
i
ona
l
s
t
a
t
e
s
,”
2023
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
A
dv
anc
e
m
e
nt
i
n
C
om
put
at
i
on
and
C
om
put
e
r
T
e
c
hnol
ogi
e
s
, I
nC
A
C
C
T
2023
, pp. 322
–
326, 2023, doi
:
10.1109/
I
nC
A
C
C
T
575
35.2023.10141774.
[
21]
D
.
R
.
S
oum
ya
,
D
.
L
.
K
.
R
e
ddy,
A
.
N
a
ga
r
,
a
nd
A
.
K
.
R
a
j
poot
,
“
E
nha
nc
i
ng
br
a
i
n
t
um
or
di
a
gnos
i
s
:
ut
i
l
i
z
i
ng
R
e
s
N
e
t
-
101
on
M
R
I
i
m
a
ge
s
f
or
de
t
e
c
t
i
on,”
V
i
T
E
C
oN
2023
-
2nd
I
E
E
E
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
V
i
s
i
on
T
ow
ar
d
s
E
m
e
r
gi
ng
T
r
e
nds
i
n
C
o
m
m
uni
c
at
i
on
and N
e
t
w
or
k
i
ng T
e
c
hnol
ogi
e
s
, P
r
oc
e
e
di
ngs
, 2023, doi
:
10.1109/
V
i
T
E
C
oN
58111.2023.10157378.
[
22]
G
.
K
a
ur
,
N
.
S
ha
r
m
a
,
R
.
C
ha
uha
n,
S
.
K
ukr
e
t
i
,
a
nd
R
.
G
upt
a
,
“
E
ye
di
s
e
a
s
e
c
l
a
s
s
i
f
i
c
a
t
i
on
us
i
ng
R
e
s
N
e
t
-
18
de
e
p
l
e
a
r
ni
ng
a
r
c
hi
t
e
c
t
ur
e
,”
2023
2nd
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on
F
ut
ur
i
s
t
i
c
T
e
c
hnol
ogi
e
s
,
I
N
C
O
F
T
2023
,
2023,
doi
:
10.1109/
I
N
C
O
F
T
60753.2023.10425690.
[
23]
A
.
C
ha
udhur
i
,
“
H
i
e
r
a
r
c
hi
c
a
l
m
odi
f
i
e
d
f
a
s
t
R
-
C
N
N
f
or
obj
e
c
t
de
t
e
c
t
i
on,”
I
n
f
or
m
at
i
c
a
,
vol
.
45,
no.
7,
pp.
67
–
81,
2021,
doi
:
10.31449/
i
nf
.v45i
7.3732.
[
24]
S
.
D
a
hi
ya
,
T
.
G
ul
a
t
i
,
a
nd
D
.
G
upt
a
,
“
P
e
r
f
or
m
a
nc
e
a
na
l
ys
i
s
of
de
e
p
l
e
a
r
ni
ng
a
r
c
hi
t
e
c
t
ur
e
s
f
or
pl
a
nt
l
e
a
ve
s
di
s
e
a
s
e
de
t
e
c
t
i
on,”
M
e
as
ur
e
m
e
nt
:
Se
ns
o
r
s
, vol
. 24, 2022, doi
:
10.1016/
j
.m
e
a
s
e
n.2022.100581.
[
25]
A
.
A
l
m
a
l
a
q
a
nd
G
.
E
dw
a
r
ds
,
“
A
r
e
vi
e
w
of
de
e
p
l
e
a
r
ni
ng
m
e
t
hods
a
ppl
i
e
d
on
l
oa
d
f
or
e
c
a
s
t
i
ng,”
16t
h
I
E
E
E
I
nt
e
r
nat
i
onal
C
onf
e
r
e
nc
e
on M
ac
hi
ne
L
e
ar
ni
ng and A
ppl
i
c
at
i
ons
, I
C
M
L
A
2017
, vol
. 2017, pp. 511
–
516,
2017, doi
:
10.1109/
I
C
M
L
A
.2017.0
-
110.
B
I
O
G
R
A
P
H
I
E
S
O
F
A
U
T
H
O
R
S
Anny
Astrid
Espitia
Cubillos
p
erformed
her
undergraduate
studie
s
in
Industrial
Engineering
in
the
Universidad
Militar
Nueva
Granada
in
2002
a
nd
M.Sc.
in
Industrial
Engineering
from
the
Universidad
de
Los
Andes
in
2006.
She
is
an
Associate
Professor
on
Industrial
Engineering
Program at
Universidad
M
ilitar Nuev
a Grana
da
, Bogotá, Colombia
.
He
can be cont
acted at em
ail:
anny.espi
tia@
unimil
itar.edu.
co
.
Robinson
Jiménez
-
Moreno
is
an
Electronic
Engineer
graduated
fro
m
Universidad
Distrital
Franc
isco
José
de
Caldas
in
2002.
He
rece
ived
a
M.Sc
.
in
Engine
ering
from
Universidad
Nacional
de
Colombia
in
2012
and
Ph.D.
in
Engineering at
Universidad
Distrital
Franc
isco
José
de
Caldas
in
2018.
His
curre
nt
workin
g
as
assista
nt
p
rofes
sor
of
Univer
sida
d
Militar
Nueva
Granada
and
researc
h
focuses
on
the
use
of
convolutio
nal
neural
networks
for
object
recognition
and
image
processing
for
robotic
applications
s
uch
as
human
-
machine
interactio
n.
He can be contacted at email:
robinson.j
imenez@
unimili
tar.edu.c
o
.
Esperanza
Rodríguez
Carmona
was born in Pereira (Co
lombia).
S
he performed
her
undergraduate
studies
in
Mechanical
Engineering
(1997)
from
the
Universidad
Tecnológica
de Pereira (UTP) and M
aster'
s degree in university teaching from the
Universidad de La Salle,
Bogotá,
Colombia
.
She
is
an
Associate
Professor
on
Industrial
E
ngineering
program
at
Universidad
Militar
Nueva
Granada,
Bogotá,
Colombia.
He
can
be
contacted
at
email:
esperanza.rodr
iguez@
unimil
itar.edu.
co
.
Evaluation Warning : The document was created with Spire.PDF for Python.