暂无图片
暂无图片
暂无图片
暂无图片
暂无图片
基于融合多尺度标记信息的深度交互式图像分割 - 丁宗元, 孙权森, 王涛, 王洪元.pdf
416
13页
4次
2021-11-16
免费下载
DOI
:
issn
JournalofCom
p
uterResearchandDevelo
p
ment
(
):
,
 
稿
:
;
:
 
:
(
,
,
);
(
BK
);
(
M
)
Thisworkwassu
pp
ortedb
y
the NationalNatualScienceFoundationofChina
(
,
,
),
the Natural
ScienceFoundationofJian
g
suProvince
(
BK
),
andtheChinaPostdoctoralScienceFoundation
(
M
)
 
 
 
 
(
 
 
)
(
 
 
)
(
dz
y
hacom
)
Dee
p
InteractiveIma
g
eSe
g
mentationBasedonFusion MultiScaleAnnotation
Information
Din
g
Zon
gy
uan
,
SunQuansen
,
Wan
g
Tao
,
and Wan
g
Hon
gy
uan
(
Schoolo
f
Com
p
uterScienceandTechnolo
gy
,
Nan
j
in
g
Universit
y
o
f
ScienceandTechnolo
gy
,
Nan
j
in
g
)
(
Schoolo
f
Com
p
uterScienceandArti
f
icialIntelli
g
ence
,
Chan
g
zhouUniversit
y
,
Chan
g
zhou
,
Jian
g
su
)
Abstract  Existin
g
dee
p
interactiveima
g
e se
g
mentation al
g
orithms calculate distance ma
p
s or
Gaussian ma
p
sfortheclickannotations
,
andthenconcatenatethem withtheima
g
easthein
p
utof
networkTheinfluenceran
g
eofeachclickisthesame
,
butthe
p
ur
p
oseofeachinteractionisdifferent
Themainroleoftheearl
y
interactionisselection
,
andthelater
p
refersfinetunin
g
Tothisend
,
a
dee
p
interactiveima
g
ese
g
mentational
g
orithmfusedwith multiscalelabelinformationis
p
ro
p
osed
First
,
b
y
settin
g
differentGaussianradius
,
two
g
rou
p
sofGaussian ma
p
swithdifferentscalesare
calculatedforeachclickSecondl
y
,
b
y
fusin
g
withthesmallscaleGaussian ma
p
s
,
andsomedown
sam
p
lin
g
modulesinthebasicse
g
mentationnetworkareremoved
,
hencericherdetailedfeaturesof
tar
g
etsareextractedAtthesametime
,
inordertomaintaintheinte
g
rit
y
ofthetar
g
etse
g
mentation
results
,
anonlocalfeatureattentionmoduleis
p
ro
p
osedandthismodulefuseslar
g
escaleGaussian
ma
p
sFinall
y
,
accordin
g
tothe
p
robabilit
y
information
p
rovidedb
y
theGaussianma
p
,
a
p
robabilit
y
clicklossis
p
ro
p
osedtoenhancethese
g
mentation
p
erformanceofthetar
g
etneartheclick
p
oint
Ex
p
erimentalresultsshowthatthe
p
ro
p
osedal
g
orithm cannotonl
y
maintaintheinte
g
rit
y
ofthe
se
g
mentation
,
butalsoobtainthese
g
mentationresultsofthetar
g
etdetails
,
which
g
reatl
y
reducesthe
user􀆳sinteractionburden
Ke
y
words interactiveima
g
ese
g
mentation
;
dee
p
learnin
g
;
multiscaleannotation
;
Gaussian ma
p
s
;
p
robabilit
y
clickloss
 
 
,
,
,
,
,
,
,
,
,
,
使
,
,
,
,
,
,
:
,
,
 
;
;
;
;
 TP
  
,
,
,
,
(
线
),
,
,
,
广
[
]
,
,
Gra
p
hCut
[
]
,
GrabCut
[
]
,
Random walks
[
]
,
,
,
,
(
dee
p
interactiveob
j
ectselection
)
[
]
,
(
full
y
convolutionalneuralnetwork
,
FCN
)
[
]
,
,
(
便
,
)
,
,
,
[
]
,
使
,
,
[
]
,
,
,
,
使
,
,
:
,
,
~
,
使
,
;
,
,
使
,
,
,
,
,
[
]
,
,
使
~
,
Forte
[
]
,
,
,
[
]
,
,
,
使
,
,
,
,
,
,
Dee
p
LabV
[
]
,
Dee
p
LabV
[
]
,
,
ResNet
[
]
,
,
 
,
(
)
of 13
免费下载
【版权声明】本文为墨天轮用户原创内容,转载时必须标注文档的来源(墨天轮),文档链接,文档作者等基本信息,否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容,欢迎发送邮件至:contact@modb.pro进行举报,并提供相关证据,一经查实,墨天轮将立刻删除相关内容。

评论

关注
最新上传
暂无内容,敬请期待...
下载排行榜
Top250 周榜 月榜