I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
,
p
p
.
4
7
7
5
~
4
7
8
6
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
6
.
p
p
4
7
7
5
-
4
7
8
6
4775
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Enha
ncing
was
te
ma
na
g
ement t
hro
ug
h municipa
l so
l
id was
te
cla
ss
ificatio
n:
a
c
o
nv
o
lutiona
l neural network a
pp
ro
a
ch
M
d.
T
a
re
q
uzza
m
a
n
1
,
M
o
j
a
hid
ul Alo
m
Ak
a
s
h
1
,
Z
a
k
ir
H
o
s
s
a
in
2
,
M
d.
Sa
bb
ir
Rez
a
1
,
Sh
a
j
j
a
du
l H
a
qu
e
3
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
a
n
d
El
e
c
t
r
o
n
i
c
E
n
g
i
n
e
e
r
i
n
g
,
Jas
h
o
r
e
U
n
i
v
e
r
si
t
y
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
J
a
s
h
o
r
e
,
B
a
n
g
l
a
d
e
sh
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
a
n
d
El
e
c
t
r
o
n
i
c
E
n
g
i
n
e
e
r
i
n
g
,
U
t
t
a
r
a
U
n
i
v
e
r
si
t
y
,
D
h
a
k
a
,
B
a
n
g
l
a
d
e
s
h
3
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
D
a
f
f
o
d
i
l
I
n
t
e
r
n
a
t
i
o
n
a
l
U
n
i
v
e
r
si
t
y
,
S
a
v
a
r
,
B
a
n
g
l
a
d
e
s
h
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ap
r
1
,
2
0
2
4
R
ev
is
ed
Sep
30
,
2
0
2
5
Acc
ep
ted
Oct
16
,
2
0
2
5
Th
e
e
sc
a
latio
n
o
f
p
o
p
u
latio
n
,
e
c
o
n
o
m
ic
e
x
p
a
n
sio
n
,
a
n
d
i
n
d
u
striali
z
a
ti
o
n
h
a
s
re
su
lt
e
d
i
n
a
n
i
n
c
re
a
se
in
w
a
ste
p
ro
d
u
c
ti
o
n
.
Th
is
h
a
s
m
a
d
e
wa
ste
m
a
n
a
g
e
m
e
n
t
m
o
re
c
h
a
ll
e
n
g
in
g
a
n
d
h
a
s
re
su
lt
e
d
i
n
e
n
v
i
ro
n
m
e
n
tal
d
e
terio
ra
ti
o
n
,
n
e
g
a
ti
v
e
l
y
imp
a
c
ti
n
g
t
h
e
q
u
a
li
t
y
o
f
l
ife.
Re
c
y
c
li
n
g
,
re
d
u
c
in
g
,
a
n
d
re
u
si
n
g
a
re
v
iab
le
m
e
th
o
d
s
to
e
ra
d
ica
te
th
e
e
sc
a
latin
g
wa
ste
issu
e
,
re
q
u
iri
n
g
th
e
a
p
p
ro
p
riate
c
las
sifi
c
a
ti
o
n
o
f
m
u
n
icip
a
l
s
o
li
d
wa
ste
.
Th
is
st
u
d
y
fo
c
u
se
s
o
n
c
o
m
p
a
rin
g
si
x
a
d
v
a
n
c
e
d
wa
ste
c
las
sifica
ti
o
n
sy
ste
m
s
th
a
t
e
m
p
lo
y
a
p
re
-
train
e
d
c
o
n
v
o
lu
t
io
n
a
l
n
e
u
r
a
l
n
e
two
r
k
(CNN
)
d
e
sig
n
e
d
to
re
c
o
g
n
ize
twe
lv
e
d
isti
n
c
t
c
a
teg
o
ries
o
f
m
u
n
icip
a
l
wa
ste
.
It
h
a
s
b
e
e
n
d
e
term
in
e
d
th
a
t
Da
rk
Ne
t5
3
is
th
e
m
o
st
e
ffe
c
ti
v
e
c
las
sifier
a
m
o
n
g
th
e
se
six
m
o
d
e
ls.
To
a
ss
e
ss
th
e
e
ffe
c
ti
v
e
n
e
ss
o
f
e
a
c
h
wa
ste
c
las
sifier,
th
e
c
o
n
f
u
sio
n
m
a
tri
x
,
p
re
c
isio
n
,
re
c
a
ll
,
F
1
sc
o
re
,
th
e
a
re
a
u
n
d
e
r
t
h
e
re
c
e
iv
e
r
o
p
e
ra
ti
n
g
c
h
a
ra
c
teristic
c
u
rv
e
,
a
n
d
t
h
e
lo
ss
f
u
n
c
ti
o
n
a
re
e
x
a
m
in
e
d
.
It
h
a
s
b
e
e
n
fo
u
n
d
t
h
a
t
Da
rk
Ne
t
5
3
h
a
s
a
n
F
1
sc
o
re
o
f
9
8
.
7
%
a
n
d
v
a
li
d
a
t
io
n
a
c
c
u
ra
c
y
o
f
9
9
%
,
re
sp
e
c
ti
v
e
ly
.
T
h
e
su
g
g
e
ste
d
a
p
p
r
o
a
c
h
will
b
e
u
se
fu
l
in
p
ro
m
o
ti
n
g
g
a
rb
a
g
e
re
c
o
v
e
r
y
a
n
d
re
u
se
in
t
h
e
d
irec
ti
o
n
o
f
a
c
ircu
lar an
d
s
u
sta
in
a
b
le ec
o
n
o
m
y
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
Dar
k
Net5
3
Dee
p
lear
n
in
g
Mu
n
icip
al
s
o
lid
waste
W
aste c
la
s
s
if
icatio
n
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
:
Md
.
T
ar
eq
u
zz
am
an
Dep
ar
tm
en
t o
f
E
lectr
ical
an
d
E
lectr
o
n
ic
E
n
g
i
n
ee
r
in
g
,
J
ash
o
r
e
Un
iv
er
s
ity
o
f
Scien
ce
an
d
T
ec
h
n
o
lo
g
y
J
ash
o
r
e
-
7
4
0
8
,
B
an
g
lad
esh
E
m
ail: m
.
tar
eq
u
zz
am
a
n
@
ju
s
t.e
d
u
.
b
d
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
m
an
ag
em
e
n
t
o
f
m
u
n
icip
al
s
o
lid
waste
(
MS
W
)
h
as
em
er
g
ed
as
a
s
ig
n
if
ican
t
co
n
ce
r
n
i
n
d
ev
elo
p
in
g
n
atio
n
s
,
p
ar
ticu
lar
ly
in
B
an
g
lad
esh
,
d
u
e
to
th
e
o
n
g
o
in
g
ac
ce
ler
atio
n
o
f
ec
o
n
o
m
ic
g
r
o
wth
an
d
u
r
b
an
izatio
n
[
1
]
.
Acc
o
r
d
in
g
to
th
e
W
o
r
ld
B
an
k
in
2
0
2
0
,
u
r
b
an
izatio
n
in
B
an
g
lad
esh
is
o
n
e
o
f
th
e
q
u
ic
k
est
in
an
y
So
u
th
Asi
an
ec
o
n
o
m
y
.
Fas
t
u
r
b
an
izatio
n
h
as
led
to
a
n
o
tab
le
r
is
e
in
g
ar
b
ag
e
v
o
lu
m
e
a
n
d
its
ad
m
in
is
tr
atio
n
'
s
in
tr
icac
y
,
p
ar
ticu
lar
ly
in
d
e
n
s
ely
p
o
p
u
lated
ar
ea
s
s
u
ch
as
Dh
ak
a.
T
h
is
h
as
ad
v
er
s
ely
af
f
ec
ted
u
r
b
an
life
,
th
e
en
v
ir
o
n
m
en
t,
a
n
d
p
u
b
lic
h
ea
lth
.
Su
s
tain
ab
le
d
ev
elo
p
m
en
t
g
o
al
(
SDG)
1
1
,
wh
ich
f
o
cu
s
es
o
n
s
u
s
tain
ab
le
cities
an
d
co
m
m
u
n
ities
,
aim
s
to
ac
h
iev
e
a
s
p
ec
if
ic
o
b
jectiv
e
o
f
d
im
in
is
h
in
g
th
e
d
etr
im
en
tal
en
v
ir
o
n
m
e
n
tal
ef
f
ec
ts
p
er
in
d
iv
id
u
al
in
u
r
b
a
n
ar
ea
s
.
T
h
ese
o
b
jectiv
es
em
p
h
asize
ad
d
r
ess
in
g
air
q
u
ality
c
o
n
ce
r
n
s
an
d
en
h
an
cin
g
m
u
n
ici
p
al
an
d
o
th
e
r
waste
m
an
ag
em
en
t
p
r
ac
tices,
aim
in
g
to
ac
co
m
p
lis
h
th
ese
g
o
als
b
y
2
0
3
0
.
Ho
wev
er
,
ac
h
iev
i
n
g
SDG
tar
g
ets
is
im
p
ed
ed
b
y
in
a
d
eq
u
ate
m
a
n
ag
em
en
t p
r
ac
tices a
n
d
t
h
e
d
ev
el
o
p
m
en
t
o
f
a
s
u
b
s
tan
tial q
u
a
n
tity
o
f
s
o
lid
waste.
I
t
is
f
o
u
n
d
th
at
th
e
q
u
a
n
tity
o
f
waste
p
r
o
d
u
ce
d
in
B
an
g
lad
esh
in
cr
ea
s
ed
b
y
1
,
3
4
,
3
0
0
m
etr
ic
to
n
s
an
n
u
ally
,
f
r
o
m
1
,
0
7
,
7
8
,
4
9
7
m
etr
ic
to
n
s
in
1
9
7
0
to
1
,
4
7
,
7
8
,
4
9
7
m
etr
ic
t
o
n
s
in
2
0
1
2
[
2
]
,
[
3
]
.
I
n
2
0
1
4
,
th
e
an
n
u
al
p
r
o
d
u
ctio
n
o
f
u
r
b
an
ar
ea
s
am
o
u
n
ted
to
5
,
2
0
0
,
9
1
9
to
n
s
,
o
r
0
.
3
5
k
ilo
g
r
am
s
p
er
ca
p
i
ta
p
e
r
[
3
]
.
B
as
ed
o
n
th
e
l
ates
t a
v
ai
la
b
le
d
at
a,
i
t h
as b
e
en
o
b
s
e
r
v
e
d
t
h
at
t
h
e
a
v
e
r
a
g
e
p
e
r
ca
p
it
a
MSW
g
e
n
er
ati
o
n
v
ar
i
es a
c
r
o
s
s
s
ev
e
r
al
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
7
7
5
-
4
7
8
6
4776
m
u
n
i
ci
p
al
d
is
t
r
icts
,
b
e
twe
en
0
.
2
t
o
0
.
5
6
k
g
/c
ap
/d
a
y
[
4
]
.
Ac
co
r
d
in
g
to
a
s
t
u
d
y
c
o
n
d
u
c
te
d
i
n
2
0
1
6
–
2
0
1
7
,
Dh
a
k
a,
th
e
ca
p
ital
o
f
B
an
g
lad
esh
,
p
r
o
d
u
ce
d
a
d
aily
a
v
er
ag
e
o
f
6
4
4
8
.
3
7
3
m
etr
ic
to
n
s
o
f
MS
W
.
T
h
is
am
o
u
n
ts
to
ar
o
u
n
d
0
.
5
7
to
n
s
p
er
ca
p
ita
ea
ch
d
ay
[
5
]
.
R
o
u
g
h
ly
5
0
%
o
f
t
h
e
g
ar
b
ag
e
p
r
o
d
u
ce
d
in
Dh
ak
a
city
is
ef
f
ec
tiv
ely
co
llected
b
y
th
e
city
co
r
p
o
r
atio
n
,
wh
ile
4
0
–
6
0
% o
f
t
h
e
g
ar
b
ag
e
s
tay
s
ab
an
d
o
n
e
d
an
d
is
n
o
t d
is
p
o
s
ed
o
f
s
af
ely
.
I
t is wo
r
th
n
o
tin
g
th
at
th
is
u
n
c
o
llected
waste
h
as a
p
p
r
o
x
im
at
ely
8
0
% o
r
g
an
ic
m
ater
ial
[
4
]
.
T
h
e
a
c
cu
m
u
la
t
i
o
n
o
f
u
n
co
l
le
c
t
e
d
w
a
s
t
e
,
n
am
e
ly
p
l
a
s
t
i
c
a
n
d
p
o
l
y
e
t
h
y
l
e
n
e
m
a
t
e
r
i
a
l
s
,
u
l
t
i
m
a
t
e
ly
f
i
n
d
s
i
t
s
w
a
y
i
n
to
d
r
a
in
ag
e
s
y
s
t
e
m
s
a
n
d
w
a
te
r
s
o
u
r
c
e
s
,
o
b
s
t
r
u
c
t
i
n
g
w
a
te
r
m
o
v
e
m
en
t
i
n
s
i
d
e
t
h
e
d
r
a
i
n
s
.
F
u
r
t
h
er
m
o
r
e,
th
e
d
e
t
r
im
e
n
ta
l
i
m
p
a
c
t
o
f
w
a
s
t
e
d
i
s
p
o
s
a
l
o
n
s
o
u
r
c
e
s
o
f
wa
t
e
r
,
ag
r
i
cu
l
t
u
r
a
l
r
e
g
io
n
s
,
a
n
d
o
t
h
e
r
c
r
u
c
i
a
l
lo
c
a
t
i
o
n
s
h
a
r
m
s
b
io
d
i
v
e
r
s
i
t
y
an
d
o
v
e
r
a
l
l
w
e
l
l
-
b
e
in
g
[
6
]
.
T
h
e
an
t
i
c
ip
a
t
ed
t
i
m
e
f
o
r
p
l
a
s
t
i
c
d
e
g
r
a
d
a
t
io
n
in
th
e
s
o
i
l
i
s
a
r
o
u
n
d
4
0
0
y
e
ar
s
[
7
]
,
an
d
th
e
ty
p
i
c
a
l
t
i
m
e
f
r
a
m
e
f
o
r
t
h
e
d
e
g
r
a
d
a
t
io
n
o
f
a
g
l
a
s
s
b
o
t
t
l
e
wh
e
n
d
i
s
c
ar
d
ed
i
n
t
h
e
g
r
o
u
n
d
a
s
g
a
r
b
ag
e
i
s
a
r
o
u
n
d
o
n
e
m
i
l
l
io
n
y
e
ar
s
.
I
n
th
i
s
s
c
e
n
ar
i
o
,
t
h
e
e
n
v
ir
o
n
m
en
t
a
l
i
m
p
a
c
t
r
e
s
u
l
t
in
g
f
r
o
m
th
e
d
i
s
p
o
s
a
l
o
f
r
e
c
y
cl
i
n
g
m
a
t
er
i
a
l
s
,
s
u
ch
a
s
p
l
a
s
t
i
c,
g
l
a
s
s
,
an
d
m
e
t
a
l
,
a
s
w
a
s
t
e
i
s
s
u
b
s
t
an
t
i
a
l
a
n
d
s
h
o
u
l
d
n
o
t
b
e
u
n
d
e
r
e
s
t
im
a
t
e
d
.
T
o
a
lar
g
e
ex
ten
t,
r
ec
y
clin
g
in
f
r
astru
ctu
r
e
allo
ws
f
o
r
th
e
r
ec
ap
tu
r
e
o
f
m
ater
ials
th
at
wo
u
ld
o
th
er
wis
e
b
e
d
is
ca
r
d
ed
.
A
p
o
ten
tially
p
o
s
s
ib
le
ap
p
r
o
ac
h
f
o
r
en
h
an
cin
g
th
e
h
an
d
lin
g
o
f
waste
lev
els
an
d
tr
an
s
f
o
r
m
in
g
MSW
in
to
v
alu
ab
le
m
ater
ials
o
r
p
r
o
d
u
cts
is
th
e
im
p
lem
en
tatio
n
o
f
s
o
u
r
ce
s
ep
ar
atio
n
,
as
o
p
p
o
s
ed
to
th
e
co
n
v
en
tio
n
al
p
r
ac
tices
o
f
in
cin
er
atio
n
o
r
lan
d
f
illi
n
g
.
T
h
e
co
n
ce
p
t
o
f
s
o
u
r
ce
s
ep
ar
a
tio
n
in
v
o
lv
es
th
e
s
eg
r
eg
atio
n
o
f
MSW
in
to
d
is
t
in
ct
ca
teg
o
r
ies
b
ased
o
n
th
e
u
n
iq
u
e
p
r
o
p
er
ties
o
f
ea
c
h
item
b
ef
o
r
e
a
n
y
f
u
r
th
er
tr
ea
tm
en
t
p
r
o
ce
s
s
es
[
8
]
.
T
h
e
c
lass
if
icatio
n
o
f
MSW
h
as
s
ig
n
if
ican
ce
s
in
ce
ea
c
h
waste
ca
te
g
o
r
y
n
ec
ess
itates
a
d
is
tin
ct
ap
p
r
o
ac
h
to
its
m
an
ag
em
en
t.
T
h
e
m
eth
o
d
s
o
f
cl
ass
if
ied
ac
cu
m
u
latio
n
,
class
if
ied
s
h
ip
p
in
g
,
an
d
class
if
ied
b
u
r
ial
ar
e
im
p
lem
en
ted
b
ased
o
n
th
e
d
is
tin
ctiv
e
ch
ar
ac
ter
is
tics
o
f
all
s
o
r
ts
o
f
waste.
T
h
e
im
p
lem
en
tatio
n
o
f
g
ar
b
a
g
e
s
ep
ar
atio
n
is
a
f
u
n
d
a
m
en
tal
an
d
cr
u
cial
ap
p
r
o
a
ch
in
MSW
m
an
ag
em
en
t,
aim
in
g
to
attain
waste
r
ed
u
ctio
n
,
r
eso
u
r
c
e
u
tili
za
tio
n
,
an
d
e
n
v
ir
o
n
m
en
t
al
s
af
ety
[
9
]
.
Hen
ce
,
th
e
p
r
o
f
icien
t
ad
m
i
n
is
tr
atio
n
o
f
MSW
ca
n
s
u
b
s
tan
tia
lly
co
n
tr
ib
u
te
to
th
e
d
ev
elo
p
m
en
t
o
f
an
ec
o
lo
g
ically
s
u
s
tain
ab
le
ec
o
s
y
s
tem
.
E
f
f
ec
tiv
ely
a
d
d
r
ess
in
g
waste
ac
cu
m
u
latio
n
th
r
o
u
g
h
th
e
ad
o
p
tio
n
an
d
ap
p
licatio
n
o
f
r
ec
y
clin
g
a
n
d
r
eu
s
e
m
eth
o
d
o
lo
g
ies
m
ay
r
esu
l
t
in
th
is
.
T
h
e
o
p
tim
u
m
p
er
f
o
r
m
an
ce
o
f
r
ec
y
clin
g
s
y
s
tem
s
m
ay
b
e
ac
h
iev
ed
th
r
o
u
g
h
th
e
in
te
g
r
atio
n
o
f
tech
n
ical
ad
v
an
ce
m
en
ts
.
Ho
wev
er
,
it
s
h
o
u
ld
b
e
n
o
ted
th
at
in
th
ese
s
y
s
tem
s
,
th
e
b
r
ea
k
d
o
wn
o
f
waste
co
n
tin
u
es to
r
ely
o
n
h
u
m
an
in
v
o
l
v
em
en
t
[
6
]
,
[
1
0
]
.
Nev
er
th
eless
,
th
e
ad
v
an
ce
m
e
n
t
o
f
ar
tific
ial
in
tellig
en
ce
tech
n
o
lo
g
ies
a
n
d
t
h
e
im
p
lem
en
tatio
n
o
f
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
s
h
av
e
th
e
p
o
ten
tial
to
en
h
a
n
ce
s
y
s
tem
p
r
o
d
u
ctiv
ity
in
th
e
f
o
r
eseea
b
l
e
f
u
tu
r
e
,
s
u
r
p
ass
in
g
th
e
co
n
tr
i
b
u
tio
n
s
o
f
h
u
m
a
n
in
v
o
lv
em
en
t.
Mo
r
e
s
p
ec
if
ically
,
th
e
co
n
tr
o
l
m
ec
h
an
is
m
s
u
s
ed
b
y
th
e
h
u
m
an
b
r
ain
m
ay
b
e
s
u
cc
ess
f
u
lly
an
d
r
ap
id
ly
tr
an
s
f
o
r
m
ed
in
t
o
AI
-
en
a
b
led
m
ac
h
in
es.
I
n
th
e
co
n
te
x
t
o
f
th
is
p
ar
ticu
lar
ad
v
an
ce
m
e
n
t,
it
b
ec
o
m
es
ap
p
ar
en
t
th
at
th
e
u
s
e
o
f
r
ec
y
clin
g
s
y
s
tem
s
th
at
r
ely
o
n
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
s
f
o
r
waste
ca
teg
o
r
izatio
n
is
an
in
ev
itab
le
r
esu
lt
[
1
1
]
.
As
o
n
e
o
f
th
e
m
o
r
e
s
o
p
h
is
ticated
im
ag
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
es,
d
ee
p
lear
n
in
g
o
u
tp
e
r
f
o
r
m
s
th
e
h
u
m
an
e
y
e
in
ter
m
s
o
f
p
e
r
f
o
r
m
an
ce
an
d
y
ield
s
ac
cu
r
ate
f
in
d
in
g
s
[
1
2
]
.
I
n
th
e
s
tu
d
y
o
f
w
aste
ca
teg
o
r
izatio
n
,
Ma
o
et
a
l
.
[
1
3
]
u
ti
liz
ed
T
r
ash
N
et
d
at
aset
in
a
g
en
eti
c
a
lg
o
r
it
h
m
to
o
p
ti
m
i
ze
f
u
ll
y
-
c
o
n
n
ec
t
e
d
-
la
y
e
r
o
f
Den
s
eNe
t1
2
1
a
n
d
cl
ass
i
f
ie
d
o
n
l
y
s
i
x
ca
te
g
o
r
i
es
o
f
w
aste
,
wh
i
le
Z
h
a
n
g
e
t
a
l
.
[
1
4
]
a
p
p
li
ed
D
en
s
eN
et
1
6
9
o
n
NW
NU
-
T
R
ASH
d
a
tase
t,
ac
h
i
e
v
i
n
g
a
n
a
cc
u
r
a
cy
o
f
8
2
%.
W
a
n
g
e
t a
l.
[
1
5
]
cl
ass
i
f
ie
d
o
n
l
y
n
i
n
e
was
te
ca
t
eg
o
r
ies
wit
h
M
o
b
il
eNe
tV
3
an
d
cl
o
u
d
co
m
p
u
ti
n
g
a
t
9
4
.
2
6
%
ac
cu
r
a
cy
.
Alti
k
at
et
a
l.
[
1
6
]
f
o
u
n
d
th
at
f
i
v
e
-
la
y
e
r
e
d
co
n
v
o
lu
ti
o
n
al
n
e
u
r
al
n
etw
o
r
k
s
(
C
NN
)
w
h
er
e
f
o
u
r
-
la
y
er
m
o
d
e
ls
o
u
tp
er
f
o
r
m
ed
p
a
r
ti
cu
la
r
l
y
f
o
r
o
r
g
a
n
ic
w
aste
.
An
E
f
f
icien
tNet
-
B
0
m
o
d
el
wa
s
p
r
esen
ted
b
y
Ma
lik
et
a
l.
[
1
7
]
to
ca
teg
o
r
ize
s
p
ec
if
ic
litt
er
ca
teg
o
r
ies.
Nn
am
o
k
o
et
a
l.
[
1
8
]
d
em
o
n
s
t
r
ated
th
at
im
ag
e
r
eso
lu
tio
n
ca
n
h
a
v
e
a
co
n
s
id
er
ab
le
im
p
ac
t
o
n
p
er
f
o
r
m
an
ce
in
th
e
class
if
icatio
n
o
f
r
ec
y
clab
l
e
an
d
o
r
g
an
ic
waste,
wh
ile
Yu
d
h
an
a
a
n
d
Fah
m
i
[
1
9
]
ca
teg
o
r
ized
im
ag
es
in
to
o
r
g
an
ic
an
d
in
o
r
g
an
ic
g
ar
b
a
g
e
u
s
in
g
C
NN
.
Pit
ak
aso
et
a
l.
[
2
0
]
,
[
2
1
]
p
r
o
p
o
s
ed
a
v
a
r
iety
o
f
C
NN
m
o
d
els
o
f
a
d
u
al
en
s
em
b
le
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
wh
er
e
g
eo
m
etr
ic
ally
en
h
a
n
ce
d
p
ictu
r
es
wer
e
s
elec
ted
f
o
r
p
o
s
t
-
d
is
aster
waste
cla
s
s
if
icatio
n
.
Ho
wev
er
,
Ham
za
h
et
a
l.
[
2
2
]
u
s
ed
Fas
ter
R
-
C
N
N
to
ca
teg
o
r
ize
f
iv
e
tr
ash
ca
teg
o
r
ies
in
an
in
teg
r
ated
m
o
b
ile
ap
p
,
J
o
s
e
an
d
Sas
ip
r
ab
a
[
2
3
]
p
r
esen
ted
a
h
y
b
r
id
m
o
d
el
th
at
co
m
b
in
e
d
Fas
ter
R
-
C
N
N
with
a
co
m
p
lex
-
v
alu
ed
e
n
co
d
i
n
g
m
u
lti
-
ch
ain
s
ee
k
er
o
p
tim
izatio
n
al
g
o
r
it
h
m
(
C
MSOA)
m
eth
o
d
.
T
h
e
ef
f
ec
tiv
en
ess
o
f
en
h
an
ce
d
C
NN
m
o
d
els
was
s
h
o
wn
in
[
2
4
]
,
wh
o
ac
h
iev
e
d
an
ac
cu
r
a
cy
o
f
9
4
.
4
0
%.
W
ith
9
3
.
2
8
%
ac
c
u
r
ac
y
,
C
h
h
ab
r
a
et
a
l.
[
2
5
]
p
r
esen
t
ed
an
i
m
p
r
o
v
e
d
d
ee
p
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
DC
NN)
.
Fo
r
g
ar
b
ag
e
class
if
icatio
n
,
Pra
s
an
th
an
d
R
au
t
[
2
6
]
f
av
o
r
ed
E
f
f
icien
tN
etB
0
ab
o
v
e
alter
n
ativ
e
m
o
d
e
ls
,
with
a
9
4
.
1
5
%
ac
cu
r
ac
y
r
ate
,
an
d
Qiu
et
a
l.
[
2
7
]
ac
h
iev
ed
9
5
.
4
%
ac
cu
r
a
cy
b
y
en
h
an
ci
n
g
C
E
-
E
f
f
icien
tNetV2
with
d
ata
au
g
m
en
tatio
n
an
d
atten
tio
n
m
eth
o
d
s
.
C
N
N
m
o
d
e
ls
s
u
c
h
a
s
D
e
n
s
eN
e
t
1
6
9
,
M
o
b
i
l
e
N
e
tV
2
,
a
n
d
R
e
s
N
et
5
0
V
2
w
e
r
e
c
o
m
p
a
r
e
d
f
o
r
g
a
r
b
a
g
e
c
l
a
s
s
i
f
ic
a
t
i
o
n
[
2
8
]
.
A
b
o
o
d
a
n
d
A
l
-
T
a
l
i
b
[
2
9
]
a
s
s
es
s
e
d
V
G
G
1
6
,
I
n
c
e
p
t
i
o
n
V
3
,
M
o
b
i
l
e
Ne
t
V
2
,
an
d
E
f
f
i
c
i
e
n
t
N
e
tB
0
,
w
i
t
h
a
c
c
u
r
ac
i
es
r
a
n
g
i
n
g
f
r
o
m
9
2
%
t
o
9
7
%
.
O
n
t
h
e
o
t
h
e
r
h
a
n
d
,
J
a
y
a
r
a
m
a
n
a
n
d
L
a
k
s
h
m
i
n
a
r
a
y
a
n
a
n
[
3
0
]
cr
ea
ted
M
SW
-
N
e
t
,
a
h
i
e
r
a
r
c
h
i
c
al
m
o
d
e
l
e
m
p
l
o
y
i
n
g
C
NN
a
n
d
B
a
y
es
ia
n
-
o
p
t
i
m
i
z
e
d
M
o
b
i
l
eN
e
t
m
o
d
el
s
,
w
h
e
r
e
as
a
C
NN
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
cin
g
w
a
s
te
ma
n
a
g
eme
n
t th
r
o
u
g
h
mu
n
icip
a
l so
lid
w
a
s
te
cla
s
s
ifica
tio
n
…
(
Md
.
Ta
r
eq
u
z
z
a
ma
n
)
4777
a
n
d
G
r
a
p
h
-
L
S
T
M
m
o
d
e
l
w
i
t
h
a
9
7
.
5
%
a
c
c
u
r
a
c
y
f
o
r
s
i
x
w
a
s
t
e
c
l
a
s
s
e
s
wa
s
p
r
o
v
i
d
e
d
b
y
L
i
a
n
d
C
h
e
n
[
3
1
]
.
Usi
n
g
R
esNex
t
-
1
0
1
with
f
ed
er
ated
lear
n
in
g
,
Kh
a
n
et
a
l
.
[
3
2
]
o
b
tain
ed
an
ac
cu
r
ac
y
o
f
8
9
.
6
2
%.
T
asic
et
a
l
.
[
3
3
]
u
s
ed
C
NNs
with
Ad
aBo
o
s
t
an
d
X
GB
o
o
s
t
to
clas
s
if
y
o
r
g
an
ic
an
d
r
ec
y
clab
le
g
ar
b
ag
e
in
a
b
in
ar
y
m
an
n
e
r
with
an
ac
cu
r
ac
y
o
f
o
v
e
r
8
9
.
6
1
%.
I
s
la
m
et
a
l.
[
3
4
]
p
r
o
p
o
s
ed
a
tailo
r
ed
Den
s
eNe
t2
0
1
ar
c
h
itectu
r
e
with
an
in
teg
r
ate
d
atten
tio
n
m
ec
h
an
is
m
,
s
ettin
g
a
n
ew
b
en
ch
m
ar
k
f
o
r
a
d
ap
tab
le
waste
class
if
icatio
n
.
Pre
v
io
u
s
r
esear
ch
h
as
in
teg
r
at
ed
im
ag
e
class
if
icatio
n
tech
n
o
lo
g
ies
an
d
m
ac
h
in
e
lear
n
in
g
a
p
p
r
o
ac
h
es
to
en
h
an
ce
waste
class
if
icatio
n
.
Ho
wev
er
,
th
ese
s
tu
d
ies
ar
e
s
u
b
ject
to
ce
r
tain
co
n
s
tr
ain
ts
.
T
h
e
lim
itatio
n
s
o
f
th
e
s
tu
d
y
en
co
m
p
ass
s
ev
er
al
asp
ec
ts
.
Firs
tly
,
th
e
waste
cla
s
s
if
icatio
n
s
y
s
tem
u
n
d
er
in
v
e
s
tig
atio
n
ex
h
ib
its
a
lim
ited
n
u
m
b
er
o
f
waste
item
s
.
Seco
n
d
ly
,
th
e
ex
is
tin
g
wast
e
class
if
icatio
n
m
o
d
el'
s
ac
cu
r
a
cy
f
alls
s
h
o
r
t
o
f
th
e
d
esire
d
lev
el.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
ad
d
r
ess
es
b
o
th
o
f
th
e
s
e
p
r
o
b
lem
s
.
T
h
e
p
r
o
p
o
s
ed
waste
clas
s
if
ier
is
d
is
tin
ctiv
e
in
th
at,
u
n
lik
e
t
h
e
p
r
ev
io
u
s
class
if
ier
,
wh
ich
ca
n
ca
teg
o
r
ize
a
m
ax
im
u
m
o
f
s
ix
class
es,
th
e
p
r
o
p
o
s
ed
m
o
d
el
ca
n
class
if
y
twelv
e
class
es.
I
n
th
is
s
tu
d
y
,
s
ix
C
NN
ar
ch
itectu
r
es,
s
p
ec
if
ically
R
esn
et5
0
,
R
esn
et1
0
1
,
R
esn
et1
5
2
,
Dar
k
Net5
3
,
Go
o
g
leNe
t,
a
n
d
I
n
ce
p
t
io
n
V3
,
ar
e
e
m
p
lo
y
e
d
to
class
if
y
1
2
d
is
tin
ct
k
in
d
s
o
f
m
u
n
icip
al
waste
th
r
o
u
g
h
th
e
ex
tr
ac
tio
n
o
f
h
ig
h
-
q
u
ality
f
e
atu
r
es f
r
o
m
t
h
e
im
ag
e
o
f
waste.
2.
M
E
T
H
O
D
2
.
1
.
Sa
m
ple
co
llect
io
n
T
o
co
n
s
tr
u
ct
a
p
r
ed
ictiv
e
m
o
d
el,
it
is
im
p
e
r
ativ
e
t
o
u
n
d
e
r
g
o
a
tr
ai
n
in
g
p
r
o
ce
s
s
[
3
5
]
.
T
h
e
u
tili
za
tio
n
o
f
in
tellig
en
t
s
y
s
tem
s
in
lieu
o
f
h
u
m
an
lab
o
r
with
in
waste
m
an
ag
em
en
t
f
ac
ilit
ies
is
an
ess
en
tial
p
r
er
eq
u
is
ite
f
o
r
ac
h
ie
v
in
g
b
o
th
ec
o
n
o
m
ic
ef
f
icien
cy
a
n
d
en
s
u
r
in
g
a
s
af
e
r
ec
y
clin
g
p
r
o
ce
s
s
.
Hen
ce
,
t
h
e
o
b
jectiv
e
o
f
t
h
is
s
tu
d
y
is
to
id
en
tif
y
an
d
ac
k
n
o
wled
g
e
s
ev
er
al
p
r
e
v
alen
t
r
ec
y
clab
le
m
ater
ials
,
in
clu
d
i
n
g
p
ap
e
r
,
ca
r
d
b
o
a
r
d
,
o
r
g
an
ic
waste,
m
etal,
an
d
p
las
tic.
T
h
e
tr
ain
d
ata
s
et
h
as
b
ee
n
s
o
u
r
ce
d
f
r
o
m
Kag
g
le.
Kag
g
le
is
an
o
n
lin
e
p
latf
o
r
m
an
d
co
m
m
u
n
ity
th
at
f
ac
ilit
ates
d
ata
s
cien
ce
co
m
p
e
titi
o
n
s
an
d
s
er
v
es
as
a
g
ath
er
i
n
g
p
lace
f
o
r
d
ata
s
cien
tis
ts
an
d
m
ac
h
i
n
e
lear
n
in
g
p
r
ac
titi
o
n
er
s
.
T
h
e
d
ataset
c
o
m
p
r
is
es
a
to
tal
o
f
1
5
,
1
5
0
im
ag
es,
ea
ch
b
elo
n
g
in
g
to
o
n
e
o
f
twelv
e
d
is
tin
ct
ca
teg
o
r
ies
r
ep
r
esen
tin
g
v
ar
io
u
s
ty
p
es
o
f
m
u
n
icip
al
waste.
T
h
ese
ca
teg
o
r
ies
in
clu
d
e
p
ap
er
,
ca
r
d
b
o
a
r
d
,
o
r
g
a
n
ic
co
n
ten
t,
m
etal,
p
last
ic,
g
r
ee
n
-
g
lass
,
b
r
o
wn
g
lass
,
wh
ite
-
g
lass
,
clo
th
es,
s
h
o
es,
b
atter
ies,
an
d
tr
ash
.
2
.
2
.
Da
t
a
prepro
ce
s
s
ing
Data
p
r
e
-
p
r
o
ce
s
s
in
g
is
th
e
c
r
u
cial
s
tep
in
m
ac
h
i
n
e
lear
n
in
g
t
h
at
in
v
o
l
v
es
clea
n
in
g
,
tr
an
s
f
o
r
m
in
g
,
a
n
d
p
r
ep
ar
in
g
th
e
d
ata
to
m
ak
e
it
s
u
itab
le
f
o
r
th
e
m
o
d
el
p
r
ed
ictio
n
[
3
6
]
.
I
n
th
is
p
r
o
ject,
t
h
r
ee
ty
p
es
o
f
d
ata
tr
an
s
f
o
r
m
atio
n
,
n
am
ely
r
escalin
g
,
r
an
d
o
m
c
r
o
p
,
an
d
r
a
n
d
o
m
f
lip
,
ar
e
u
s
ed
to
p
r
o
ce
s
s
th
e
d
ata
s
et.
T
h
e
r
escalin
g
lay
e
r
s
ca
les
th
e
p
ix
el
v
alu
es
o
f
th
e
in
p
u
t
im
a
g
es
b
y
d
i
v
id
in
g
th
em
b
y
2
5
5
.
T
h
is
s
tep
en
s
u
r
es
th
at
th
e
p
ix
el
v
al
u
es
ar
e
in
th
e
r
an
g
e
o
f
[
0
,
1
]
.
T
h
e
R
an
d
o
m
C
r
o
p
lay
er
r
an
d
o
m
ly
cr
o
p
s
t
h
e
in
p
u
t
im
a
g
es
to
a
s
p
ec
if
ied
s
ize
o
f
2
2
4
×
2
2
4
p
ix
els.
T
h
is
s
tep
h
elp
s
in
d
ata
a
u
g
m
en
tatio
n
a
n
d
in
t
r
o
d
u
ce
s
v
ar
iatio
n
s
in
th
e
in
p
u
t
d
ata.
T
h
e
R
an
d
o
m
Fli
p
lay
er
r
an
d
o
m
l
y
f
lip
s
th
e
im
ag
es
h
o
r
izo
n
tally
.
T
h
is
au
g
m
en
tatio
n
tech
n
iq
u
e
f
u
r
th
e
r
en
h
an
ce
s
th
e
d
iv
er
s
ity
o
f
th
e
t
r
ain
in
g
d
ata
.
T
h
e
r
escalin
g
lay
er
d
o
es
n
o
t
ch
an
g
e
th
e
im
a
g
e
s
ize
b
u
t
o
n
ly
s
ca
les
th
e
p
ix
el
v
al
u
es.
T
h
e
R
an
d
o
m
C
r
o
p
(
tr
ain
in
g
d
ata
)
an
d
r
es
izin
g
(
test
d
ata)
lay
e
r
s
m
o
d
if
y
th
e
im
ag
e
s
ize
b
y
cr
o
p
p
in
g
o
r
r
esizin
g
,
r
esp
ec
tiv
ely
.
Af
ter
all
th
is
,
we
g
o
t
p
ix
el
s
ize
2
2
4
×
2
2
4
f
o
r
ev
er
y
im
ag
e
.
2
.
3
.
Da
t
a
pa
rt
i
t
io
nin
g
T
o
tal
o
b
s
er
v
atio
n
is
d
iv
id
e
d
i
n
to
th
r
ee
p
ar
ts
:
tr
ain
in
g
d
ata,
v
alid
atio
n
s
et,
an
d
test
d
ata.
T
r
ain
d
ata
is
u
s
ed
to
tr
ain
th
e
p
r
ed
ictiv
e
m
o
d
el,
v
alid
atio
n
d
ata
is
u
s
ed
to
v
alid
ate
th
e
m
o
d
el
,
an
d
th
e
tes
t d
ata
s
et
i
s
u
s
ed
to
ev
alu
ate
th
e
p
e
r
f
o
r
m
an
ce
o
f
t
h
e
f
in
al
m
o
d
el
[
3
7
]
.
A
to
tal
o
f
6
0
%
o
f
th
e
d
ata
is
u
s
ed
to
t
r
ain
th
e
m
o
d
el,
2
0
%
o
f
th
e
d
ata
is
u
s
ed
t
o
v
alid
ate
t
h
e
m
o
d
el
,
an
d
t
h
e
r
em
ain
i
n
g
d
ata
is
u
s
ed
to
test
th
e
f
in
al
p
r
e
d
ictiv
e
m
o
d
el.
2
.
4
.
M
o
del
t
ra
ini
ng
a
nd
t
est
ing
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
em
p
lo
y
s
p
r
e
-
tr
ain
e
d
weig
h
ts
f
r
o
m
I
m
ag
eNe
t
to
in
itialize
th
e
R
es
N
et5
0
,
R
esNet1
0
1
,
an
d
R
es
N
et1
5
2
m
o
d
els.
Su
b
s
eq
u
e
n
t
to
t
h
e
f
r
e
ez
in
g
o
f
th
e
la
y
er
s
with
in
th
e
m
o
d
els,
a
d
d
itio
n
al
lay
er
s
ar
e
i
n
co
r
p
o
r
ated
to
f
ac
il
itate
th
e
p
r
o
ce
s
s
o
f
class
if
icati
o
n
.
B
ef
o
r
e
c
o
m
p
ilin
g
an
o
v
e
r
v
iew
o
f
t
h
e
m
o
d
el'
s
ar
ch
itectu
r
e,
th
e
m
o
d
el
is
ass
em
b
led
b
y
in
c
o
r
p
o
r
atin
g
an
o
p
tim
izer
,
lo
s
s
f
u
n
ctio
n
,
an
d
m
etr
ics.
T
h
e
p
r
o
c
ess
th
a
t
is
u
s
e
d
f
o
r
ea
c
h
R
es
N
et
a
r
ch
i
te
ct
u
r
e
e
n
t
ails
u
t
ili
zi
n
g
in
p
u
t
i
m
a
g
es
wit
h
d
i
m
e
n
s
i
o
n
s
o
f
(
2
2
4
,
2
2
4
,
3
)
p
i
x
e
ls
,
wh
e
r
e
t
h
e
h
ei
g
h
t
is
2
2
4
p
ix
els
,
th
e
wi
d
t
h
is
2
2
4
p
i
x
e
ls
,
a
n
d
t
h
e
r
e
a
r
e
t
h
r
e
e
c
o
l
o
r
c
h
an
n
els
c
o
r
r
esp
o
n
d
in
g
to
t
h
e
R
GB
c
o
l
o
r
m
o
d
el
.
T
h
e
R
esN
et
m
o
d
el
a
cc
ep
ts
in
p
u
t
i
m
a
g
e
s
wit
h
d
im
e
n
s
i
o
n
s
(
2
2
4
,
2
2
4
,
3
)
an
d
p
r
o
ce
ed
s
to
p
r
o
ce
s
s
th
em
th
r
o
u
g
h
its
n
u
m
er
o
u
s
lay
er
s
.
T
h
e
R
esNet1
0
1
m
o
d
el
m
ain
tain
s
t
h
e
o
r
ig
in
al
d
im
en
s
io
n
s
o
f
th
e
in
p
u
t im
ag
es.
Fo
llo
win
g
th
e
im
p
lem
en
tati
o
n
o
f
t
h
e
R
esNet
m
o
d
el,
a
Glo
b
alAv
er
ag
ePo
o
lin
g
2
D
lay
er
is
in
co
r
p
o
r
ated
.
T
h
e
f
o
llo
win
g
l
ay
er
d
o
es
g
lo
b
al
av
er
ag
e
p
o
o
lin
g
ac
r
o
s
s
th
e
s
p
atial
d
im
e
n
s
io
n
s
o
f
t
h
e
f
ea
tu
r
e
m
ap
s
p
r
o
d
u
ce
d
b
y
th
e
p
r
ec
ed
i
n
g
lay
er
s
,
r
esu
ltin
g
i
n
a
r
ed
u
c
tio
n
o
f
th
e
s
p
atial
d
im
e
n
s
io
n
s
to
a
p
r
ed
eter
m
in
ed
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
7
7
5
-
4
7
8
6
4778
s
ize.
C
o
n
s
eq
u
en
tly
,
a
f
ea
tu
r
e
v
ec
to
r
with
d
im
en
s
io
n
s
o
f
(
1
,
1
,
2
0
4
8
)
is
o
b
tain
e
d
.
Af
ter
w
ar
d
s
,
a
d
e
n
s
e
lay
er
co
m
p
r
is
in
g
1
2
u
n
its
is
in
co
r
p
o
r
ated
af
te
r
th
e
g
lo
b
al
av
er
ag
e
p
o
o
lin
g
lay
er
.
T
h
e
f
u
lly
co
n
n
ec
ted
la
y
er
is
r
esp
o
n
s
ib
le
f
o
r
ac
q
u
ir
in
g
an
d
d
is
ce
r
n
in
g
d
is
tin
ct
p
atter
n
s
an
d
f
ea
tu
r
es
th
at
f
ac
ilit
ate
clas
s
if
y
in
g
im
ag
es
in
to
1
2
d
is
tin
ct
class
es.
T
h
e
f
ea
t
u
r
e
v
ec
to
r
o
b
tain
ed
f
r
o
m
t
h
e
p
r
ec
ed
in
g
p
h
ase
is
f
latten
ed
an
d
f
ed
in
to
th
e
d
en
s
e
lay
er
.
Fo
llo
win
g
th
at,
a
s
o
f
t
m
ax
ac
tiv
atio
n
f
u
n
ctio
n
is
in
co
r
p
o
r
ated
in
to
th
e
ac
tiv
atio
n
lay
er
af
ter
th
e
d
en
s
e
lay
er
.
T
h
e
a
f
o
r
em
e
n
tio
n
ed
lay
er
g
en
er
ates th
e
p
r
o
b
ab
ilit
y
o
f
ea
ch
class
with
in
th
e
s
et
o
f
1
2
class
es.
T
h
e
R
esNet
m
o
d
el
m
ain
tain
s
a
co
n
s
is
ten
t
im
ag
e
s
ize
th
r
o
u
g
h
o
u
t
its
ar
ch
itectu
r
e
with
o
u
t
alter
in
g
th
e
d
im
en
s
io
n
s
o
f
th
e
im
ag
es.
T
h
e
alter
atio
n
s
in
s
ize
s
o
lely
tr
an
s
p
ir
e
d
u
r
in
g
th
e
p
r
ep
r
o
ce
s
s
in
g
s
tag
es
p
r
ec
ed
in
g
th
e
m
o
d
el.
T
h
e
m
o
d
el
is
tr
ain
ed
f
o
r
2
0
ep
o
ch
s
,
af
ter
wh
ich
its
p
er
f
o
r
m
an
ce
is
d
eter
m
in
ed
.
Af
ter
th
e
m
o
d
el
tr
ain
in
g
,
th
e
test
d
ata
is
ap
p
lied
to
th
e
m
o
d
el,
a
n
d
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
m
o
d
el
is
ev
alu
at
ed
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
s
tu
d
y
em
p
lo
y
s
s
ix
d
is
tin
ct
C
NN
ar
ch
itectu
r
es
to
ca
teg
o
r
ize
twelv
e
m
u
n
icip
al
s
o
lid
d
eb
r
is
class
if
icatio
n
s
ac
cu
r
ately
.
T
h
e
m
o
d
el
is
tr
ain
ed
u
s
in
g
th
e
s
t
o
ch
asti
c
g
r
ad
ien
t
d
escen
t
m
o
m
en
tu
m
m
eth
o
d
f
o
r
o
p
tim
izatio
n
,
with
an
in
itial
lear
n
in
g
r
ate
o
f
0
.
0
0
1
.
T
h
e
in
itial
lear
n
in
g
r
ate
was
s
et
at
a
s
m
all
n
u
m
b
er
s
in
ce
a
h
ig
h
v
alu
e
m
ig
h
t
lea
d
th
e
r
at
e
to
d
iv
er
g
e
o
r
p
r
o
d
u
ce
wo
r
s
e
-
th
an
-
id
ea
l
r
esu
lts
.
Nev
er
th
e
less
,
th
e
s
to
ch
asti
c
g
r
ad
ien
t
d
escen
t
p
r
o
ce
s
s
ten
d
s
to
f
lu
ctu
ate
wh
ile
f
o
llo
win
g
th
e
t
r
ajec
to
r
y
o
f
th
e
s
teep
e
s
t
d
r
o
p
to
war
d
s
th
e
o
p
tim
al
s
o
lu
tio
n
.
On
e
m
et
h
o
d
to
d
ec
r
ea
s
e
t
h
is
o
s
cillatio
n
is
b
y
i
n
clu
d
in
g
m
o
m
en
tu
m
p
ar
t
s
in
th
e
p
a
r
am
eter
u
p
d
ate
[
3
8
]
.
T
h
e
m
o
m
en
tu
m
p
ar
am
eter
in
th
is
s
tu
d
y
is
ass
ig
n
ed
a
co
n
s
tan
t v
alu
e
o
f
0
.
9
.
I
n
th
is
s
tu
d
y
,
d
ata
n
o
r
m
aliza
tio
n
is
p
er
f
o
r
m
e
d
p
r
io
r
to
th
e
u
s
e
o
f
tr
ain
in
g
d
ata
to
p
r
ev
en
t
o
v
er
f
itti
n
g
.
No
r
m
aliza
tio
n
is
a
cr
u
cial
a
n
d
ef
f
ec
tiv
e
tech
n
iq
u
e
f
o
r
r
ed
u
cin
g
g
en
er
aliza
tio
n
er
r
o
r
s
an
d
d
ea
lin
g
with
o
v
er
f
itti
n
g
in
d
ee
p
lear
n
in
g
m
o
d
els
ca
u
s
ed
b
y
v
ar
io
u
s
p
ar
am
eter
s
.
I
n
th
is
s
tu
d
y
,
th
e
L
2
r
eg
u
lar
izatio
n
is
ad
ju
s
ted
at
1
0
e
-
4
to
lim
it
th
e
g
r
ad
ien
t
v
alu
e
a
n
d
p
r
o
v
id
e
a
th
r
esh
o
ld
f
o
r
th
e
g
r
ad
ien
t.
T
h
e
ev
alu
atio
n
o
f
ea
ch
waste
class
if
ier
in
v
o
lv
es
an
al
y
zin
g
th
e
co
n
f
u
s
io
n
m
atr
ix
,
ar
ea
u
n
d
er
t
h
e
r
ec
eiv
er
o
p
er
atin
g
ch
ar
ac
te
r
is
tic
cu
r
v
e,
p
r
ec
is
io
n
,
r
ec
all,
F1
s
co
r
e,
an
d
l
o
s
s
f
u
n
ctio
n
t
o
d
eter
m
in
e
th
eir
ef
f
icac
y
.
3
.
1
.
Co
nfusi
o
n m
a
t
rice
s
Fig
u
r
e
1
illu
s
tr
ates
th
e
co
n
f
u
s
io
n
m
atr
ix
,
with
Dar
k
Net5
3
i
n
Fig
u
r
e
1
(
a)
,
Go
o
g
leNe
t
in
Fig
u
r
e
1
(
b
)
,
I
n
ce
p
tio
n
V3
in
Fig
u
r
e
1
(
c)
,
R
esNet5
0
in
Fig
u
r
e
1
(
d
)
,
R
esNet1
0
1
in
Fig
u
r
e
1
(
e)
,
an
d
R
esNet1
5
2
in
Fig
u
r
e
1
(
f
)
[
3
9
]
,
[
4
0
]
,
all
o
f
wh
ich
wer
e
u
tili
ze
d
in
th
e
p
er
f
o
r
m
an
ce
ev
alu
atio
n
o
f
th
e
cl
ass
if
icatio
n
m
o
d
els.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
is
also
ev
alu
ated
f
o
r
th
e
r
est
o
f
th
e
m
o
d
els.
T
h
e
r
o
ws
o
f
th
e
m
atr
ices
co
n
tain
th
e
p
r
ed
icted
v
al
u
es
f
o
r
th
e
f
o
llo
win
g
ca
teg
o
r
ies:
ca
r
d
b
o
ar
d
,
g
l
ass
,
m
etal,
p
ap
er
,
p
last
ic,
an
d
d
eb
r
is
.
T
h
e
co
l
u
m
n
s
o
f
th
e
m
atr
ices c
o
n
tain
th
e
ac
t
u
al
v
alu
es c
o
r
r
esp
o
n
d
i
n
g
to
t
h
e
s
am
e
ca
teg
o
r
ies.
I
n
Fig
u
r
e
1
(
a)
,
t
h
e
d
ia
g
o
n
al
lin
e
r
ef
lects
th
e
ac
cu
r
ate
class
if
icatio
n
o
f
tr
ash
p
h
o
to
s
,
wh
ile
th
e
v
alu
es
elsewh
er
e
in
d
icate
th
e
n
u
m
b
e
r
o
f
waste
im
ag
es
th
at
ar
e
n
o
t
co
r
r
ec
tly
class
if
ied
.
I
n
Fig
u
r
e
1
(
a)
,
3
6
5
b
atter
y
tr
ash
p
h
o
t
o
s
ar
e
a
p
p
r
o
p
r
iately
lab
eled
as
b
atter
ies,
b
u
t
f
o
u
r
p
ap
er
s
am
p
les,
f
o
u
r
m
etal
s
am
p
les,
an
d
o
n
e
s
h
o
e
ca
teg
o
r
y
s
am
p
le
ar
e
w
r
o
n
g
ly
lab
eled
as
b
atter
ies.
I
t
is
f
o
u
n
d
th
at
th
e
c
o
n
f
u
s
io
n
m
atr
i
x
f
o
r
Dar
k
Net5
3
,
I
n
ce
p
tio
n
V3
,
Go
o
g
leNe
t,
R
esNet5
0
,
R
esNet1
0
1
,
an
d
R
esNet1
5
2
ca
n
co
r
r
ec
tly
s
o
r
t
u
p
to
6
1
4
4
,
6
1
2
1
,
5
9
5
4
,
2
9
8
0
,
2
9
9
3
,
a
n
d
2
9
9
3
p
ictu
r
es
,
r
esp
ec
tiv
ely
.
T
h
e
p
r
ec
is
io
n
o
f
p
r
ed
ictin
g
e
ac
h
class
m
ay
b
e
d
eter
m
in
e
d
u
s
in
g
th
e
co
n
f
u
s
io
n
m
at
r
ix
.
Fig
u
r
e
2
illu
s
tr
ates
a
co
m
p
ar
ativ
e
co
m
p
ar
is
o
n
o
f
s
ev
e
r
al
ca
teg
o
r
izat
io
n
m
eth
o
d
s
b
ased
o
n
th
ei
r
p
r
ec
is
io
n
v
alu
es.
A
h
ig
h
er
p
r
ec
is
io
n
s
co
r
e
in
d
icate
s
th
at
th
e
m
o
d
el
ca
n
ac
cu
r
a
tely
ca
teg
o
r
ize
th
e
m
o
s
t
p
o
s
i
tiv
e
class
if
icatio
n
s
with
in
a
ce
r
tain
ca
teg
o
r
y
.
T
h
e
g
r
ap
h
d
em
o
n
s
tr
ates
th
at
t
h
e
Dar
k
Net5
3
m
o
d
el
ac
cu
r
at
ely
id
en
tifie
s
m
o
s
t
ca
teg
o
r
ies.
Dar
k
Net5
3
d
em
o
n
s
tr
ates
ef
f
ec
tiv
e
class
if
icatio
n
in
eig
h
t
o
u
t
o
f
th
e
twelv
e
ca
te
g
o
r
ies.
I
n
t
h
e
f
o
u
r
r
em
ain
in
g
ca
teg
o
r
ies,
I
n
ce
p
tio
n
V3
o
u
tp
er
f
o
r
m
s
Dar
k
Net5
3
is
p
er
f
o
r
m
in
g
well.
Nev
er
t
h
eless
,
th
e
ac
cu
r
ac
y
o
f
D
ar
k
Net5
3
f
o
r
th
ese
f
o
u
r
ca
te
g
o
r
ies is
m
ar
g
in
ally
in
f
er
io
r
t
o
th
at
o
f
I
n
ce
p
tio
n
V
3
.
3.
2.
Rec
eiv
er
o
pera
t
o
r
cha
r
a
ct
er
is
t
ic
a
nd
a
re
a
un
der
t
he
curv
e
T
h
e
r
ec
eiv
er
o
p
er
ato
r
c
h
ar
ac
te
r
is
tic
(
R
O
C
)
cu
r
v
e
is
a
to
o
l
th
at
r
esear
ch
er
s
h
av
e
u
tili
ze
d
to
ass
es
s
th
e
p
r
ed
ictiv
e
p
o
wer
o
f
a
m
o
d
el
[
4
1
]
.
T
h
e
co
n
f
u
s
io
n
m
at
r
ix
m
ay
b
e
u
s
ed
to
o
b
tain
th
e
R
OC
cu
r
v
e.
T
h
e
ar
ea
u
n
d
er
t
h
e
cu
r
v
e
(
AUC)
is
b
en
ea
th
th
e
co
o
r
d
in
ate
a
x
is
an
d
th
e
R
OC
cu
r
v
e
[
4
2
]
.
T
h
e
ty
p
i
ca
l
r
a
n
g
e
f
o
r
A
UC
is
0
.
5
t
o
1
.
T
h
e
ac
cu
r
a
c
y
r
a
te
o
f
a
c
lass
i
f
i
er
is
h
ig
h
er
i
f
its
AU
C
is
l
a
r
g
er
.
C
o
n
ce
r
n
s
a
b
o
u
t
b
i
n
a
r
y
cl
ass
i
f
icat
io
n
ar
e
p
a
r
ti
c
u
la
r
l
y
am
en
a
b
le
t
o
t
h
e
a
p
p
lic
ati
o
n
o
f
th
e
AUC
-
R
OC
c
u
r
v
e
.
Ho
w
ev
er
,
b
y
u
ti
liz
in
g
o
n
e
-
to
-
m
a
n
y
te
c
h
n
iq
u
es,
it
is
p
o
s
s
ib
l
e
to
ex
te
n
d
its
s
c
o
p
e
t
o
e
n
c
o
m
p
ass
o
t
h
er
class
i
f
i
ca
t
io
n
is
s
u
es
t
h
a
t
in
v
o
lv
e
m
a
n
y
class
es
[
4
3
]
.
T
h
is
s
tu
d
y
ex
am
in
es
twe
lv
e
d
is
ti
n
ct
class
if
ic
ati
o
n
s
o
f
g
a
r
b
a
g
e
,
a
n
d
Fi
g
u
r
e
3
ill
u
s
t
r
at
es
t
h
e
R
OC
c
u
r
v
es
,
wit
h
Da
r
k
Net
5
3
i
n
Fi
g
u
r
e
3
(
a
)
,
Go
o
g
l
e
Net
i
n
F
ig
u
r
e
3
(
b
)
,
I
n
ce
p
t
io
n
V
3
i
n
F
ig
u
r
e
3
(
c
)
,
R
e
s
Net
5
0
i
n
Fi
g
u
r
e
3
(
d
)
,
R
esNe
t
1
0
1
in
Fi
g
u
r
e
3
(
e)
,
a
n
d
R
es
Ne
t1
5
2
i
n
Fi
g
u
r
e
3
(f)
.
On
e
co
m
m
o
n
u
s
e
o
f
R
OC
an
aly
s
is
is
ev
alu
atin
g
class
if
icat
io
n
alg
o
r
ith
m
p
er
f
o
r
m
a
n
ce
b
y
d
em
o
n
s
tr
atin
g
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
cin
g
w
a
s
te
ma
n
a
g
eme
n
t th
r
o
u
g
h
mu
n
icip
a
l so
lid
w
a
s
te
cla
s
s
ifica
tio
n
…
(
Md
.
Ta
r
eq
u
z
z
a
ma
n
)
4779
co
r
r
elatio
n
b
etwe
en
tr
u
e
p
o
s
itiv
e
an
d
f
alse
p
o
s
itiv
e
[
4
4
]
.
T
h
er
e
is
a
way
to
e
v
alu
ate
th
e
ef
f
ec
tiv
en
ess
o
f
t
h
e
ca
teg
o
r
izatio
n
u
s
in
g
t
h
e
AUC.
A
h
ig
h
er
AUC
in
d
icate
s
m
o
r
e
p
r
ec
is
e
ca
teg
o
r
izatio
n
.
B
y
an
aly
zin
g
th
e
R
OC
cu
r
v
e
o
f
s
ev
e
r
al
class
if
ier
s
,
it
is
ev
id
en
t
th
at
th
e
AUC
is
g
r
ea
ter
f
o
r
th
e
Dar
Net5
3
class
if
ier
th
an
f
o
r
o
th
er
class
if
icatio
n
m
o
d
els.
Hen
ce
,
Dar
k
Net5
3
h
as
s
u
p
er
io
r
p
r
ed
i
ctiv
e
ca
p
ab
ilit
ies
co
m
p
ar
e
d
to
th
e
o
th
e
r
m
o
d
el
s
.
Ad
d
itio
n
ally
,
it
is
n
o
ted
th
at
wh
en
co
m
p
a
r
in
g
Dar
k
Net5
3
t
o
o
th
er
m
o
d
els,
th
e
tr
u
e
p
o
s
itiv
e
r
ate
is
g
r
ea
ter
ac
r
o
s
s
all
ca
teg
o
r
y
ty
p
es.
C
o
n
ce
r
n
in
g
th
e
en
tire
n
u
m
b
er
o
f
p
o
s
itiv
e
s
am
p
les,
th
e
class
if
ie
r
p
r
o
p
er
ly
id
e
n
tifie
s
th
e
m
o
s
t p
o
s
itiv
e
s
am
p
les.
As a
r
esu
lt,
th
e
m
o
d
el'
s
s
en
s
itiv
it
y
is
h
ig
h
.
(
a)
(
b
)
(
c
)
(
d
)
(
e
)
(
f
)
Fig
u
r
e
1
.
C
o
n
f
u
s
io
n
m
atr
i
x
f
o
r
(
a)
Dar
k
Net5
3
,
(
b
)
Go
o
g
leN
et
,
(
c
)
I
n
ce
p
tio
n
V3
,
(
d
)
R
esNe
t5
0
,
(
e
)
R
esNet1
0
1
,
an
d
(
f
)
R
esNet1
5
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
7
7
5
-
4
7
8
6
4780
Fig
u
r
e
2
.
Th
e
p
r
ec
is
io
n
o
f
d
if
f
er
e
n
t m
o
d
els f
o
r
d
if
f
er
en
t c
las
s
es
(
a)
(
b
)
(
c
)
(
d
)
(
e
)
(
f
)
Fig
u
r
e
3
.
R
OC
f
o
r
(
a
)
Dar
k
Ne
t5
3
,
(
b
)
Go
o
g
leNe
t
,
(
c
)
I
n
ce
p
ti
o
n
V3
,
(
d
)
R
esNet5
0
,
(
e
)
R
esNet1
0
1
,
a
n
d
(
f
)
R
esNet1
5
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
cin
g
w
a
s
te
ma
n
a
g
eme
n
t th
r
o
u
g
h
mu
n
icip
a
l so
lid
w
a
s
te
cla
s
s
ifica
tio
n
…
(
Md
.
Ta
r
eq
u
z
z
a
ma
n
)
4781
3
.
3
.
Acc
ura
cy
a
nd
lo
s
s
curv
e
I
n
ad
d
i
t
io
n
,
t
h
e
a
c
cu
r
a
cy
p
l
o
t
a
n
d
t
h
e
tr
a
i
n
in
g
l
o
s
s
c
u
r
v
e
w
e
r
e
u
t
i
l
iz
e
d
to
a
s
s
e
s
s
t
h
e
p
er
f
o
r
m
an
c
e
o
f
th
e
v
ar
i
o
u
s
c
l
a
s
s
i
f
i
c
at
i
o
n
m
o
d
e
l
s
.
W
h
en
l
ea
r
n
i
n
g
m
o
d
el
s
a
r
e
in
t
r
o
d
u
c
ed
to
a
n
e
w
t
as
k
,
l
e
ar
n
in
g
c
u
r
v
e
s
a
r
e
th
o
u
g
h
t
t
o
b
e
u
s
e
f
u
l
t
o
o
l
s
f
o
r
t
r
ac
k
in
g
th
e
i
r
s
u
c
c
e
s
s
.
L
ea
r
n
in
g
c
u
r
v
e
s
ar
e
a
m
a
t
h
e
m
a
ti
c
a
l
d
e
s
cr
i
p
t
io
n
o
f
t
h
e
l
e
a
r
n
i
n
g
p
r
o
c
e
s
s
t
h
a
t
h
a
p
p
e
n
s
a
s
a
r
e
s
u
l
t
o
f
t
a
s
k
r
e
p
e
ti
t
i
o
n
[
4
5
]
.
T
h
r
o
u
g
h
o
u
t
t
h
e
t
r
a
i
n
i
n
g
p
r
o
c
e
s
s
o
f
a
m
a
c
h
in
e
l
ea
r
n
i
n
g
m
o
d
e
l,
i
t
i
s
p
o
s
s
i
b
le
t
o
a
s
s
e
s
s
t
h
e
m
o
d
e
l's
p
r
e
s
e
n
t
s
t
a
t
e
a
t
ea
c
h
i
t
e
r
a
t
io
n
o
f
t
h
e
t
r
a
in
i
n
g
a
l
g
o
r
i
t
h
m
.
T
h
e
m
o
d
e
l
's
p
e
r
f
o
r
m
a
n
c
e
m
ay
b
e
a
s
s
e
s
s
e
d
b
y
e
v
a
l
u
a
t
i
n
g
i
t
o
n
th
e
t
r
a
in
i
n
g
d
a
t
a
s
e
t
,
p
r
o
v
i
d
in
g
i
n
s
i
g
h
t
s
i
n
to
i
t
s
l
e
a
r
n
i
n
g
c
a
p
a
b
i
l
i
t
i
e
s
.
A
d
d
i
t
i
o
n
a
l
ly
,
i
t
m
ay
b
e
a
s
s
e
s
s
e
d
u
s
in
g
a
s
e
p
a
r
a
te
v
a
l
i
d
a
t
io
n
d
a
t
a
s
et
t
h
a
t
i
s
n
o
t
i
n
c
lu
d
e
d
in
th
e
tr
a
i
n
in
g
d
a
t
a
s
e
t.
T
h
e
m
o
d
e
l
's
g
en
e
r
a
l
iz
i
n
g
ab
i
l
i
t
i
e
s
ar
e
s
h
o
w
n
b
y
a
n
a
l
y
z
in
g
t
h
e
v
a
l
i
d
a
t
io
n
d
a
t
a
s
e
t.
B
a
s
e
d
o
n
th
e
a
cc
u
r
a
c
y
c
u
r
v
e,
i
t
i
s
ev
i
d
e
n
t
th
a
t
a
l
l
m
o
d
e
l
s
ex
h
ib
i
t
s
t
r
o
n
g
p
e
r
f
o
r
m
an
c
e
o
n
t
h
e
t
r
a
in
i
n
g
d
a
t
a.
T
h
e
ir
v
a
l
i
d
a
t
i
o
n
a
n
d
t
r
a
in
i
n
g
a
c
cu
r
ac
y
a
r
e
p
r
o
g
r
e
s
s
i
v
e
l
y
im
p
r
o
v
in
g
,
s
u
g
g
e
s
t
i
n
g
th
at
t
h
e
l
e
a
r
n
i
n
g
m
o
d
e
l
ef
f
ec
t
i
v
e
ly
f
i
t
s
t
h
e
tr
a
i
n
in
g
d
a
t
a
.
T
h
e
a
c
c
u
r
a
c
y
cu
r
v
e
o
f
D
a
r
k
N
e
t
5
3
i
s
r
e
p
r
e
s
e
n
t
ed
i
n
F
i
g
u
r
e
4
.
T
h
e
a
c
cu
r
a
cy
c
u
r
v
e
f
o
r
I
n
c
ep
t
i
o
n
V
3
,
R
e
s
N
e
t
5
0
,
R
e
s
N
e
t
1
0
1
,
R
e
s
N
e
t
1
5
2
,
a
n
d
G
o
o
g
le
N
e
t
c
a
n
b
e
f
o
u
n
d
in
F
i
g
u
r
e
s
5
(
a)
t
o
5
(
e
)
.
I
t
i
s
f
o
u
n
d
t
h
a
t
t
h
e
v
a
li
d
a
t
i
o
n
a
c
cu
r
ac
i
e
s
o
f
Go
o
g
le
N
e
t
,
D
a
r
k
N
e
t
5
3
,
I
n
c
ep
t
i
o
n
V
3
,
R
e
s
N
e
t
5
0
,
R
e
s
N
e
t
1
0
1
,
a
n
d
R
e
s
N
e
t
1
5
2
a
r
e
d
e
t
e
r
m
in
e
d
t
o
b
e
9
5
.
9
2
%
,
9
9
%
,
9
8
.
6
1
%
,
9
4
.
2
9
%
,
9
4
.
2
1
%
,
a
n
d
9
5
.
1
0
%
c
o
r
r
e
s
p
o
n
d
in
g
l
y
.
Fig
u
r
e
4
.
T
h
e
ac
cu
r
ac
y
cu
r
v
e
f
o
r
Dar
k
Net5
3
T
h
e
l
o
s
s
c
u
r
v
e
i
s
a
n
o
t
h
e
r
m
e
t
r
i
c
f
o
r
e
v
a
l
u
a
t
i
n
g
t
h
e
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
'
s
e
f
f
i
c
a
c
y
.
T
h
e
l
o
s
s
c
u
r
v
e
f
o
r
D
a
r
k
N
e
t
5
3
,
I
n
c
e
p
t
i
o
n
V
3
,
G
o
o
g
l
e
N
e
t
,
R
e
s
N
e
t
5
0
,
R
e
s
N
e
t
1
0
1
,
a
n
d
R
e
s
N
e
t
1
5
2
c
a
n
b
e
f
o
u
n
d
i
n
F
i
g
u
r
e
s
6
(
a
)
t
o
6
(
f
)
.
T
h
e
lo
s
s
cu
r
v
e
o
f
all
m
o
d
els
d
em
o
n
s
tr
ates
a
co
n
s
is
ten
t
d
ec
r
ea
s
e
in
tr
ain
in
g
an
d
v
alid
atio
n
l
o
s
s
as
th
e
n
u
m
b
er
o
f
iter
atio
n
s
r
is
es.
T
h
e
lo
s
s
c
u
r
v
e
f
o
r
th
e
Go
o
g
leNe
t,
Dar
k
Net5
3
,
an
d
I
n
ce
p
tio
n
V3
m
o
d
els
ex
h
ib
its
r
ed
u
ce
d
tr
ain
in
g
an
d
v
alid
atio
n
lo
s
s
es
co
m
p
ar
ed
t
o
th
e
o
th
er
m
o
d
els
.
Ad
d
itio
n
ally
,
it
is
ev
id
en
t
th
at
th
e
tr
ain
in
g
er
r
o
r
is
s
m
aller
th
an
th
e
v
alid
atio
n
er
r
o
r
,
an
d
th
e
d
is
cr
ep
an
c
y
b
etwe
en
th
e
v
alid
atio
n
an
d
tr
ai
n
in
g
er
r
o
r
s
is
m
in
im
al.
T
h
ese
i
n
d
ica
to
r
s
s
u
g
g
est
t
h
a
t
th
e
m
o
d
e
ls
ar
e
n
o
t
s
u
f
f
er
in
g
f
r
o
m
o
v
e
r
f
itt
i
n
g
.
T
a
b
l
e
1
p
r
es
en
ts
a
c
o
m
p
ar
ati
v
e
co
m
p
a
r
is
o
n
o
f
s
i
x
d
is
ti
n
ct
MS
W
class
if
ie
r
s
.
T
ab
le
1
d
em
o
n
s
tr
ates
th
at
Dar
k
Net5
3
ex
h
i
b
its
g
r
ea
ter
r
ec
all
an
d
p
r
ec
is
io
n
v
alu
es.
T
h
e
r
ec
all
v
alu
e
o
f
9
8
.
7
%
in
d
icate
s
th
at
th
e
m
o
d
el
ca
n
ac
cu
r
ately
ca
teg
o
r
ize
9
8
.
7
%
o
f
tr
ash
item
s
in
to
a
ce
r
tain
ca
teg
o
r
y
.
R
o
u
g
h
ly
1
.
3
%
o
f
ce
r
tain
t
r
ash
is
m
is
clas
s
if
ied
in
to
o
th
er
ca
teg
o
r
ies.
T
h
e
h
ar
m
o
n
ic
m
ea
n
o
f
ac
cu
r
ac
y
a
n
d
r
ec
all
is
r
ep
r
esen
te
d
b
y
th
e
F1
s
co
r
e
[
4
6
]
.
E
lev
ated
v
alu
e
s
o
f
th
e
F
-
m
ea
s
u
r
e
,
wh
ich
r
an
g
e
f
r
o
m
ze
r
o
to
o
n
e,
in
d
icate
s
u
p
er
io
r
class
if
icat
io
n
p
er
f
o
r
m
an
ce
.
T
h
e
F1
s
co
r
e
o
f
Dar
k
Net5
3
is
s
ig
n
if
ican
tly
h
ig
h
,
as
s
h
o
wn
in
T
ab
le
1
.
As
a
r
esu
lt,
DAr
k
Net5
3
ex
h
i
b
its
s
u
p
er
io
r
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
co
m
p
ar
ed
to
th
e
o
th
er
m
o
d
el
s
.
T
ab
le
1
.
C
o
m
p
a
r
ativ
e
an
aly
s
is
b
etwe
en
d
if
f
e
r
en
t p
r
ed
ictiv
e
m
o
d
els
C
r
i
t
e
r
i
a
D
a
r
k
N
e
t
5
3
(
%)
I
n
c
e
p
t
i
o
n
V
3
(
%)
G
o
o
g
l
e
N
e
t
(
%)
R
e
sN
e
t
5
0
(
%)
R
e
sN
e
t
1
0
1
(
%)
R
e
sN
e
t
1
5
2
(
%)
A
c
c
u
r
a
c
y
99
9
8
.
6
1
9
5
.
9
2
9
4
.
2
9
9
4
.
2
1
9
5
.
1
0
P
r
e
c
i
s
i
o
n
9
8
.
8
9
7
.
9
4
9
4
.
7
5
9
4
.
1
6
9
4
.
9
1
9
4
.
9
1
R
e
c
a
l
l
9
8
.
7
1
9
8
.
2
0
9
3
.
9
4
94
95
95
F
1
S
c
o
r
e
9
8
.
8
9
8
.
1
9
4
.
3
4
94
95
95
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
7
7
5
-
4
7
8
6
4782
(
a)
(
b
)
(
c)
(
d
)
(
e)
Fig
u
r
e
5
.
Acc
u
r
ac
y
c
u
r
v
e
f
o
r
(
a)
I
n
ce
p
tio
n
V3
,
(
b
)
R
esNet5
0
,
(
c)
R
esNet1
0
1
(
d
)
R
esNet1
5
2
,
an
d
(
e)
Go
o
g
leNe
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
E
n
h
a
n
cin
g
w
a
s
te
ma
n
a
g
eme
n
t th
r
o
u
g
h
mu
n
icip
a
l so
lid
w
a
s
te
cla
s
s
ifica
tio
n
…
(
Md
.
Ta
r
eq
u
z
z
a
ma
n
)
4783
(
a)
(
b
)
(
c)
(
d
)
(
e)
(f)
Fig
u
r
e
6
.
L
o
s
s
cu
r
v
e
f
o
r
(
a
)
D
ar
k
Net5
3
,
(
b
)
I
n
ce
p
tio
n
V3
,
(
c)
Go
o
g
leNe
t,
(
d
)
R
esNet5
0
,
(
e)
R
esNet1
0
1
,
an
d
(
f
)
R
esNet1
5
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
7
7
5
-
4
7
8
6
4784
4.
CO
NCLU
SI
O
N
T
h
is
r
esear
ch
s
tu
d
y
p
r
esen
ted
an
ef
f
ec
tiv
e
waste
class
if
icat
io
n
m
o
d
el
ca
teg
o
r
izin
g
twelv
e
d
is
tin
ct
f
o
r
m
s
o
f
MSW
.
Six
d
is
tin
ct
m
o
d
els
f
o
r
class
if
y
in
g
g
ar
b
ag
e
ar
e
p
r
esen
ted
,
an
d
a
th
o
r
o
u
g
h
s
tu
d
y
is
co
n
d
u
cted
to
co
m
p
ar
e
th
em
.
T
h
e
p
e
r
f
o
r
m
an
ce
o
f
th
ese
s
ix
class
if
ier
s
is
ass
e
s
s
ed
b
ased
o
n
th
e
co
n
f
u
s
io
n
m
atr
ix
,
R
OC
,
ac
cu
r
ac
y
c
u
r
v
e,
lo
s
s
cu
r
v
e,
p
r
ec
is
io
n
-
r
ec
all,
an
d
F
1
s
co
r
e.
I
t
h
as
b
ee
n
d
eter
m
in
e
d
th
at
Dar
k
Net5
3
,
wh
ich
h
as
a
v
alid
atio
n
ac
cu
r
ac
y
o
f
9
9
%
,
is
th
e
m
o
s
t
ef
f
ec
tiv
e
class
if
ier
am
o
n
g
t
h
ese
s
ix
m
o
d
els.
Ho
wev
er
,
g
ar
b
ag
e
ex
h
ib
its
m
an
y
s
h
ap
es,
en
c
o
m
p
ass
in
g
v
ar
io
u
s
s
o
r
ts
,
co
lo
r
s
,
an
d
tex
tu
r
es.
Ad
d
itio
n
ally
,
th
e
m
eth
o
d
s
o
f
r
u
b
b
is
h
co
llectio
n
d
if
f
e
r
s
ig
n
if
ican
tly
f
r
o
m
t
h
o
s
e
u
s
ed
f
o
r
r
eg
u
lar
co
m
m
o
d
ities
.
Du
e
to
th
is
f
a
cto
r
,
th
e
s
u
g
g
ested
m
o
d
el
will
ex
h
ib
it
s
u
b
p
ar
p
er
f
o
r
m
an
ce
wh
e
n
ap
p
lied
to
ac
tu
al
p
h
o
to
s
th
at
o
v
e
r
la
p
with
ea
ch
o
t
h
er
.
Simu
ltan
eo
u
s
ly
,
wh
ile
ca
teg
o
r
izatio
n
is
u
n
d
en
ia
b
ly
a
cr
u
cial
s
tag
e,
th
e
ac
tu
ato
r
'
s
ef
f
ec
ti
v
en
ess
m
u
s
t
also
b
e
co
n
s
id
er
ed
f
o
r
ac
h
ie
v
in
g
ac
tu
al
au
to
m
ated
s
o
r
tin
g
.
Su
b
s
eq
u
en
t
r
esear
ch
will
in
v
o
lv
e
ev
al
u
atin
g
th
e
s
u
g
g
ested
m
o
d
el
o
n
m
o
r
e
d
atasets
an
d
d
ev
elo
p
in
g
an
e
f
f
ec
tiv
e
ac
tu
at
o
r
to
en
h
an
ce
th
e
o
v
e
r
all
ac
cu
r
ac
y
an
d
ef
f
icien
cy
o
f
ca
teg
o
r
izatio
n
.
F
UNDING
I
NF
O
R
M
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
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Md
.
T
ar
eq
u
zz
am
an
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Mo
jah
id
u
l A
lo
m
Ak
ash
✓
✓
✓
✓
✓
✓
✓
✓
✓
Z
ak
ir
Ho
s
s
ain
✓
✓
✓
✓
✓
Md
.
Sab
b
ir
R
ez
a
✓
✓
✓
✓
✓
✓
Sh
ajjad
u
l H
aq
u
e
✓
✓
✓
✓
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
Data
u
tili
ze
d
to
s
u
p
p
o
r
t th
e
o
u
tco
m
e
o
f
th
is
s
tu
d
y
ad
d
r
ess
ed
with
in
th
e
ar
ticle
.
RE
F
E
R
E
NC
E
S
[
1
]
X
.
Y
a
o
e
t
a
l
.
,
“
R
e
d
u
c
t
i
o
n
p
o
t
e
n
t
i
a
l
o
f
G
H
G
e
mi
ssi
o
n
s
f
r
o
m
m
u
n
i
c
i
p
a
l
so
l
i
d
w
a
s
t
e
i
n
c
i
n
e
r
a
t
i
o
n
f
o
r
p
o
w
e
r
g
e
n
e
r
a
t
i
o
n
i
n
B
e
i
j
i
n
g
,
”
J
o
u
rn
a
l
o
f
C
l
e
a
n
e
r P
r
o
d
u
c
t
i
o
n
,
v
o
l
.
2
4
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
c
l
e
p
r
o
.
2
0
1
9
.
1
1
8
2
8
3
.
[
2
]
M
.
A
s
h
i
k
u
z
z
a
m
a
n
a
n
d
M
.
H
.
H
o
w
l
a
d
e
r
,
“
S
u
s
t
a
i
n
a
b
l
e
so
l
i
d
w
a
st
e
ma
n
a
g
e
me
n
t
i
n
B
a
n
g
l
a
d
e
s
h
:
i
ss
u
e
s
a
n
d
c
h
a
l
l
e
n
g
e
s,
”
i
n
S
u
s
t
a
i
n
a
b
l
e
W
a
st
e
M
a
n
a
g
e
m
e
n
t
:
C
h
a
l
l
e
n
g
e
s
i
n
D
e
v
e
l
o
p
i
n
g
C
o
u
n
t
ri
e
s
,
P
e
n
n
sy
l
v
a
n
i
a
,
U
S
A
:
I
G
I
G
l
o
b
a
l
,
2
0
2
0
,
p
p
.
3
5
–
55
,
d
o
i
:
1
0
.
4
0
1
8
/
9
7
8
1
7
9
9
8
0
1
9
8
6
.
c
h
0
0
2
.
[
3
]
S
.
S
h
a
ms
,
J
.
N
.
S
a
h
u
,
S
.
M
.
S
.
R
a
h
man
,
a
n
d
A
.
A
h
s
a
n
,
“
S
u
st
a
i
n
a
b
l
e
w
a
st
e
ma
n
a
g
e
me
n
t
p
o
l
i
c
y
i
n
B
a
n
g
l
a
d
e
s
h
f
o
r
r
e
d
u
c
t
i
o
n
o
f
g
r
e
e
n
h
o
u
se
g
a
ses
,
”
S
u
st
a
i
n
a
b
l
e
C
i
t
i
e
s
a
n
d
S
o
c
i
e
t
y
,
v
o
l
.
3
3
,
p
p
.
1
8
–
2
6
,
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
scs
.
2
0
1
7
.
0
5
.
0
0
8
.
[
4
]
D
.
T.
Jeri
n
e
t
a
l
.
,
“
A
n
o
v
e
r
v
i
e
w
o
f
p
r
o
g
r
e
ss
t
o
w
a
r
d
s
i
m
p
l
e
me
n
t
a
t
i
o
n
o
f
s
o
l
i
d
w
a
s
t
e
m
a
n
a
g
e
me
n
t
p
o
l
i
c
i
e
s
i
n
D
h
a
k
a
,
B
a
n
g
l
a
d
e
sh
,
”
H
e
l
i
y
o
n
,
v
o
l
.
8
,
n
o
.
2
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
2
.
e
0
8
9
1
8
.
[
5
]
H
.
M
a
h
m
u
d
,
“
C
l
i
m
a
t
e
c
h
a
n
g
e
a
n
d
m
u
n
i
c
i
p
a
l
s
o
l
i
d
w
a
s
t
e
ma
n
a
g
e
m
e
n
t
i
n
D
h
a
k
a
M
e
g
a
c
i
t
y
i
n
B
a
n
g
l
a
d
e
s
h
,
”
i
n
C
l
i
m
a
t
e
C
h
a
n
g
e
i
n
B
a
n
g
l
a
d
e
s
h
:
A
C
r
o
ss
-
D
i
s
c
i
p
l
i
n
a
r
y
Fr
a
m
e
w
o
rk
,
C
h
a
m
,
S
w
i
t
z
e
r
l
a
n
d
:
S
p
r
i
n
g
e
r
,
2
0
2
1
,
p
p
.
1
3
5
–
155
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
030
-
75825
-
7.
[
6
]
N
.
J.
G
.
J.
B
a
n
d
a
r
a
a
n
d
J
.
P
.
A
.
H
e
t
t
i
a
r
a
t
c
h
i
,
“
E
n
v
i
r
o
n
m
e
n
t
a
l
i
m
p
a
c
t
s
w
i
t
h
w
a
s
t
e
d
i
s
p
o
s
a
l
p
r
a
c
t
i
c
e
s
i
n
a
s
u
b
u
r
b
a
n
mu
n
i
c
i
p
a
l
i
t
y
i
n
S
r
i
La
n
k
a
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
n
v
i
ro
n
m
e
n
t
a
n
d
W
a
s
t
e
M
a
n
a
g
e
m
e
n
t
,
v
o
l
.
6
,
n
o
.
1
/
2
,
p
p
.
1
0
7
–
1
1
6
,
2
0
1
0
,
d
o
i
:
1
0
.
1
5
0
4
/
I
JEW
M
.
2
0
1
0
.
0
3
3
9
8
7
.
[
7
]
A
.
C
h
a
m
a
s
e
t
a
l
.
,
“
D
e
g
r
a
d
a
t
i
o
n
r
a
t
e
s
o
f
p
l
a
s
t
i
c
s
i
n
t
h
e
e
n
v
i
r
o
n
m
e
n
t
,
”
A
C
S
S
u
st
a
i
n
a
b
l
e
C
h
e
m
i
st
ry
&
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
8
,
n
o
.
9
,
p
p
.
3
4
9
4
–
3
5
1
1
,
F
e
b
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
2
1
/
a
c
ssu
s
c
h
e
me
n
g
.
9
b
0
6
6
3
5
.
[
8
]
T.
Tr
u
s
h
n
a
e
t
a
l
.
,
“
I
n
t
e
r
v
e
n
t
i
o
n
s
t
o
p
r
o
m
o
t
e
h
o
u
s
e
h
o
l
d
w
a
s
t
e
se
g
r
e
g
a
t
i
o
n
:
a
s
y
st
e
ma
t
i
c
r
e
v
i
e
w
,
”
H
e
l
i
y
o
n
,
v
o
l
.
1
0
,
n
o
.
2
,
F
e
b
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
4
.
e
2
4
3
3
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.