salient objects in clutter: bringing salient object...
TRANSCRIPT
Salient Objects in Clutter: Bringing Salient
Object Detection to the Foreground
,Ï�¸e�wÍ5ÔNµ
òwÍ5ÔNuÿí�#pÝ
��²1, §²²1, 4ñô1,
pÿu1, ûøQ1, and Ali Borji2
1 O�ÅÆ�§Hm�ƧU9§¥I2 O�ÅÀúïÄ¥%§¥ÃÛp��Æ, c�=õ, ÃÛp�, {I
http://mmcheng.net/SOCBenchmark/
Abstract. ·��éwÍ5ÔNuÿ£SOD¤�.JÑ�«nܵ�
�Y"·��©Û(½ykSODêâ8�3Xî�O �))§�b
�z�ã�3$,Ï�¸¥�¹����²wâÑ�wÍé�"3ykê
â8þ?1µ���O �¬���k?�SOD�.ÐyÑ�Ú�p5
U", §�A^uy¢.�F~|µ�§ù�.E,��ØU-<
÷¿"�â·�éwÍ5êâ8�©Û,Äk(½���¡Ú²ï�ê
â8AT÷v�7�'�:",�§·�JÑ��#�p�þêâ8¿�
#�c�wÍ5ÄOÿÁ"AO/§·��SOC£,Ï�¸e�wX
é�¤êâ8�¹5gF~ÔNaO�wÍÚ�wÍÔN�ã�"Ø
ÔNaO�I5�§z�wÍã�Ñ��XU�Ny¢.|µ¥£
O]Ô5��'á5"��§·�3êâ8þé�«�{?1Äuá5�
5Uµ�"
Keywords: wÍ5ÔNuÿ· wÍ5ÄOÿÁ· êâ8· á5
1 {0
�©Ì�©ÛwÍ5ÔNuÿ£SOD¤�?Ö"ÀúwÍ5�3��<
aÀúXÚÀJÀú|µ�,�f8�Uå" SODKýuuÿ|µ¥áÚ
�õ5¿å�ÔN§,�Å��/J�ÔN�Ó+"SOD�`:3u§3Nõ
O�ÅÀú?Ö¥þk2��A^§�)µÀú�l [4]§ã�u¢ [14, 16]§O
�Åã/Æ [9]§SNa��ã�à} [45]ÚfiÒ�©� [18, 39,40]"
2 Deng-Ping Fan et al.
�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ �wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ �wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ �wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ�wÍ
�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ�ÙÔ
HO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SO
�����������������
HO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SOHO,OC,SC,SO
�÷��÷��÷��÷��÷��÷��÷��÷��÷��÷��÷��÷��÷��÷��÷��÷��÷�
HO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SCHO,SC
û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤û¤
HO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,AC
ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ
g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�g1�
OC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SCOC,SC
����������������� <<<<<<<<<<<<<<<<<
HO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SCHO,OC,OV,SC
�����������������<<<<<<<<<<<<<<<<<
OVOVOVOVOVOVOVOVOVOVOVOVOVOVOVOVOV
ÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅÃÅ
�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý�ý
HO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SCHO,OC,SC
<<<<<<<<<<<<<<<<<
BO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OCBO,OC
�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶�¶¶
HO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OVHO,OC,OV
<<<<<<<<<<<<<<<<<
HO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,ACHO,OC,AC
�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð�Ð
O�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅO�ÅHOHOHOHOHOHOHOHOHOHOHOHOHOHOHOHOHO
)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M)P�>M
����������������������������������
Fig. 1. SOCêâ8¥���ã��)�wÍÔNã�£111¤ÚwÍÔNã�£12
1�14 1¤"éuwÍÔNã�§·�Jø¢~?ý�ã£ØÓôÚL«ØÓ¢~¤!
ÔNá5ÚaOI\"k'·�êâ8��õ`²§�ë�Ö¿á�"
�©�ó�Ì�É�ü�*�éu"Äk§yk�SODêâ8 [2, 5, 10,
11, 23, 26, 29, 32, 43, 44] 3êâÂ8L§½êâ�þ�¡�3"�"äN ó§
�õêêâ8b�ã��¹����wÍÔN§Ïd§�¿ïØ�¹wÍÔ
N�ã�"·�òd¡�êâÀJ �"d§ykêâ8Ì��¹äkü�
ÔN�ã�½?u$,Ï�¸¥�õ�ÔN£Ï~´<¤"ùêâ8ØU¿©
�Ny¢.¥ã��E,5§3y¢.¥§|µÏ~�¹õ�,Ï�ÔN"
ù����(JB´3ykêâ8þÔö�Ly�Z��.A�®²�±¦5
U���Ú£~X§3�õêêâ8þ§F -measure> 0.9 ¤�§�3y¢|µ
þ�Ly%Ã{-<÷¿£~X§Table 3¥F -measure < 0.45 ¤"ù´Ï�3
�cêâ8þÔöÑ5��.�\ ���n��|µ§¤±��§�A^u
y¢.¥�|µ�§Ùk�5�U¬É�4��f"Ïd§�)ûd¯K§
k7��ï��C¢S^��êâ8"
Ùg§3y3�êâ8þ·��U©Û�.��N5U§ùêâ8Ñ"
y�Ny¢|µ¥¤¡��]Ô��«á5"Ïd§Ú\ùá5kÏu1¤�
�\/)SOD¯K§2¤ïÄSOD�.�`":§3¤éuØÓ�A^5`§
lØÓ��Ý�*/µd�.�5U§Ùµd(J�U´ØÓ�"
�Ä�þãü�¯K§·��2��z"·��Ì��z´�ï��#
�p�þSODêâ8§òÙ·¶�SOC£Salient Objects in Clutter¤"î8�
òwÍ5ÔNuÿí�#pÝ 3
�§SOC´���¢~?SODêâ8§§�¹5g80õ�~�aO�6,000Üã
�"§�ykêâ8�ØÓ�?3un��¡µ1¤wÍÔNäkaO5º§�
^uÃXfiÒSOD?Ö�a�#.ïħ2¤�¹�wÍã�§¦Têâ8�
�Cý¢.|µ§¿�'ykêâ8�ä]Ô5§3¤wÍÔNäk�Ný¢
.¥¡��A½�¹�á5§~X$ÄE¤��§ñÚ,Ï��µ"Ï
d§·��SOCêâ8 �ykêâ8�y¢.|µ�m��ɧ¿Jø
�Ün�ÄOÿÁ£� Fig. 1¤"
d§·��éA«�k?�òÈ ²�ä£CNN¤�.?1nܵ
� [8, 15, 17, 23, 24, 28, 31, 36, 38, 48–51]"�µ��.5U§·�Ú\n�µ
��I5Ýþuÿ(J�«��q5!©����°Ý±9(J�(��q5"
d§·�JøÄuá5�5Uµ�"ùá5#N·���\/n)�.¿
�?�Ú�Ñäkdå�ïÄ��"·��&§Têâ8ÚÄOÿÁ¬é�
5�SODïħAO´éu¡�A^��.mu�)�~��K�"��êâ
8Ú©Ûóäò�¤�ú¯uÙ"
2 �'ó�
3�!¥§·�ò{�?Ø@yk�!�SOD?Ö�O�êâ8§cÙ
lI5a.!z�ã��wÍÔNêþ!ã�êþÚã��þ��¡Ðm"·
��ò£�ÄuòÈ ²�ä�SOD�."
(a) �ã (b) Å�� (c) ¢~ (d) ©�
Fig. 2. �c�SODêâ8=ÏL±�£b¤wÍÔN�Å��Ó+5I5ã�"ØÓu
5g£d¤MSCOCOÔN©�êâ8 [27]£Têâ8¥ÔNØ�½´wÍ�¤§�©�ó
��:3ué£c¤wÍÔN�¢~?�©�"
2.1 êâ8
@Ïêâ8�oÛ�uã��êþ§�oÉ�uwÍÔN�I5�þ"~
X§êâ8MSRA-A [29]ÚMSRA-B [29]¥�wÍÔNÄ�´±I½µ�/
ª?1I5"ASD [1]ÚMSRA10K [11]3zÜã�¥�õ��¹��wÍÔ
4 Deng-Ping Fan et al.
N§ SED2 [2] êâ83üÜã�¥�¹ü�ÔN�=�¹100�ã�"�
Jpêâ8��þ§Cc5ïÄ<m©Â8äk�éE,Ú,Ï�µä
kõ�ÔN�êâ8"ùêâ8�)DUT-OMRON [44]!ECSSD [43]!
Judd-A [5] ÚPASCAL-S [26]"��c�êâ8�'§ùêâ83I5�þ
Úã�êþ�¡��U?"êâ8HKU-IS [23]§XPIE [41]ÚDUTS [37]Ï
LÂ8äkõ�wÍÔN��þÅ��I5ã�£ Fig. 2 £b¤¤5�Ñù"
:", §¦��Ñ�wÍÔN§¿�vkJø¢~?£Fig. 2£c¤¤�wÍ
ÔNI5"Ød�§ [19]�ïÄ<Â8��6kÜ{ü��µã�£�õ
ê´X«nã�¤5L«�wÍ�|µ§�duý¢|µ�E,§ÏdTêâ
8Øv±�Ný¢|µ"ILSO [22]êâ8�¹¢~?wÍé�I5§�ÙI5
�>.E��o÷§XFig. 5£a¤¤«"
o ó�§ykêâ8Ì�8¥3äk{ü�µ��ßwÍÔN�ã�þ"
�Ä�ykêâ8�þãÛ�5§I����¹äk�wÍÔN!/r|µ0
�«n±9äká5�wÍÔN��bCý¢|µ�êâ8±^uT+���
5ïÄ"ù��êâ8�±�ÏïÄö�\ÉSOD�.�f:Ú`³"
2.2 �.
·��â?Ö�êþ5é�k?�SOD�Ý ²�ä�.?1©a"
ü?Ö�.���8I´uÿã�¥�wÍÔN"3LEGS [36]¥§ÛÜ
&EÚ�Ûé'Ý©Odü�ØÓ��ÝòÈ ²�äÓP§,�ò§�KÜ
±)¤wÍã"3©z [51]¥§Zhao�<�SODJÑ��õþe©�ÝÆS
µe£MC¤"Li�< [23]£MDF¤JѦ^l�ÝòÈ ²�ä¥J�õº
ÝA�5�ÑwÍã"Li�< [24]JÑ���Ýé'�ä£DCL¤§§Ø=�
Ä��&E§�ò©�?O�Ú�KÜ��ä¥"Lee�< [15]£ELD¤�Ä
lòÈ ²�ä¥J��p?A�ÚÃó�O�A�"Liu�< [28]£DHS¤
�O��ü�ã��ä§Ù¥���)��oÑ� ~ýÿã§,�´
,���äéýÿã�[!?1[z¿©�ÚÅÚ/éýÿã?1þæ�"
Long�< [30]JÑ�«�òÈ�ä£FCN¤§T�ä¦�8��ýÿ¯K�
à�àÔöC��1"RFCN [38]¦^��E��òÈ�äòo÷�ý
ÿã��wÍ5�k�&E§¿±�«Å�ã��ªU?��)¤�ýÿ
ã"DISC [8]µe�JÑ^u[âÝã��wÍ5O�"|^ü�æU�òÈ
²�ä5©O¼�o÷Ú[âÝ�wÍã"IMC [48]ÏL�òÈ�ä3ØÓ
�¡þ�ÜwÍ5�¢"§´�«�±k�/|^ÆS���Â�¢Úp
�«�ÚOêâ5¼�°(>��SOD�."�C§�SJÑ�«äkáë
�£DSS¤���e� [17]"Hou�<V\lp?OA��ÄuHED [42]e
òwÍ5ÔNuÿí�#pÝ 5
Table 1. ÄuòÈ ²�ä�SOD�."·�òù�.©�ü?Ö£S-T¤Úõ
?Ö£M-T¤Training Setµµµ MB´MSRA-Bêâ8 [29]"MK´MSRA-10K [11] ê
â8" ImageNetêâ8´� [34]"D´DUT-OMRON [44]êâ8"H´HKU-IS [23]ê
â8"P´PASCAL-S [26]êâ8"P2010´PASCAL VOC 2010�©�êâ8 [12]"
Base Modelµµµ VGGNet§ResNet-101§AlexNet§GoogleNet´¤Äu��."FCNµµµ
�.´Ä¦^�òÈ�ä"Spµµµ�.´Ä¦^���"Proposalµµµ�.´Ä¦^8IJ
�"Edgeµµµ�.´Ä¦^>�½Ó+&E
No Model Year Pub #Training Training Set Base Model FCN Sp Proposal Edge
S-T
1 LEGS [36] 2015 CVPR 3,340 MB + P — × × X ×2 MC [51] 2015 CVPR 8,000 MK GoogLeNet × X × ×3 MDF [23] 2015 CVPR 2,500 MB — × X × X4 DCL [24] 2016 CVPR 2,500 MB VGGNet X X × ×5 ELD [15] 2016 CVPR 9,000 MK VGGNet × X × ×6 DHS [28] 2016 CVPR 9,500 MK+D VGGNet × × × ×7 RFCN [38] 2016 ECCV 10,103 P2010 — X X × X8 DISC [8] 2016 TNNLS 9,000 MK — × X × ×9 IMC [48] 2017 WACV 6,000 MK ResNet-101 X X × ×10 DSS [17] 2017 CVPR 2,500 MB VGGNet X × × X11 NLDF [31] 2017 CVPR 2,500 MB VGGNet X × × ×12 AMU [49] 2017 ICCV 10,000 MK VGGNet X × × X13 UCF [50] 2017 ICCV 10,000 MK — X × × ×
M-T
1 DS [25] 2016 TIP 10,000 MK VGGNet X X × ×2 WSS [37] 2017 CVPR 456K ImageNet VGGNet X X × ×3 MSR [22] 2017 CVPR 5,000 MB + H VGGNet X × X X
��$?OA��ë�§¿�Te�¢yûÐ�5U"NLDF [31]�ÜÛ
ÜÚ�ÛA�§¿3IO�����¥O\>.���±Ôöà�à�ä"
AMU [49]´��Ï^�àÜõ?òÈA��µe"§òoÑ��ÂÚ�[�
A�N�8¤�õ�©EÇ¥",�§§g·A/ÆSXÛòz�©EÇ�A
�ãÚýÿ�wÍã�|ÜA�ã�(Ü"UCF [50]�JÑ�8�´JpwÍ
5uÿ�°�5ÚO(5"¦�3A½�òÈ���Ú\#/¤�¿ï�§
±�ï��Ø(½�SÜA�ü�8Ü"d§¦�3k��·Üþæ��{
��JÑ#O��¿ï�§±~�)èì�ä¥�òÈö��Ú���"
õ?Ö�.8c�)n«�{§DS§WSSÚMSR"DS [25]�.ïá�
�õ?Ö�ÆS�Y§T�.�^u&¢wÍ5uÿÚ�Âã�©��m�S
3�'5§§����òÈ�ä�¥�&E±)¤ÔNa��k�A�"�C§
Wang�< [37]JÑ��¶�WSS��.§T�.mu�«¦^ã�?I\
?1wÍ5uÿ�fiÒÆS�{"Äk§¦��ÓÔöcµín�ä£FIN¤
Ú�òÈ�ä?1ã�©a",�§¦�¦^cµín�äS�CRF5\r�
mI\��? ýÿwÍã"MSR [22]�õºÝ|ÜàaÚÄuMAP� [47]f
8`zµe�(Ü��^uwÍ«�uÿÚwÍÔNÓ+uÿ§T�.¦^n
�®��äk��ëê�VGG�ä6Ú^u3ØÓºÝþKÜ(J�ÆS5¿
å�.§¦�ïÄöU¼�ûÐ�ýÿ(J"
6 Deng-Ping Fan et al.
·��âJÑ�SOCêâ8é�þ�k?�ÄuòÈ ²�ä��.
£�Table 1¤?1ÄOÿÁ§uy�c�.�3�¯K¿�Ñ�5�ïÄ
��"
3 JÑ�êâ8
3�!¥§·�ò0�#��3�[�Ný¢.|µ�äk]Ô5
�SOCêâ8"5gSOC��~ã�XFig. 1¤«"d§'uSOC�aOÚá
5�ÚO©OX Fig. 4ÚFig. 6¤«"Äuykêâ8�`:Ú":§·�(½
�¡Ú²ï�êâ8AT÷v�Ô�'��¡"
1) �wÍÔN��3"A�¤k�ykSODêâ8Ñb�ã��¹���
�wÍÔN¿¿ïØ�¹wÍÔN�ã�"�´§ù«b�´��êâÀJ
��Lun�z��½"3ý¢|µ��½¥§ã�¿Øo´�¹wÍÔN"
~X§�ý/��µã�§XU�§úÚ«n��Ø�¹wX�ÔN [6]"
�wÍÔN½�µ/��0�UÓâ��|µ§Ïdî��wÍÔN��
U �"Xia�< [41]ÏL�ä�o´wÍÔNÚ�oØ´wÍÔN§JÑk
?�SOD�.§`²�wÍÔNéínwÍÔN�'�"ùL²�wÍÔN
ATÚSOD¥�wÍÔNÉ�Ó�À"�¹�½êþ��wÍÔNã�¬¦
�êâ8��Cý¢|µ§Ó��¦�SOD?ÖC��ä]Ô5"Ïd§·�
ò/�wÍÔN0½Â�/vkwÍÔN�ã�½äk/��05��ã�"
X [6, 41]¥¤ã§/��0aO�)£a¤�8©Ù��qÔN§£b¤/G�§
Ú£c¤vk�Â�«�§©OXFig. 3 £a¤- £c¤¤«"
(a) (b) (c)
Fig. 3. ��wÍã��«~
Äu�wÍÔN�½Â§·�lDTD [21]êâ8¥Â8783�«nã�"
�´Lõ�5§lpé�ÚÙ¦êâ8¥Â82217Ìã�§�)41§U
òwÍ5ÔNuÿí�#pÝ 7
�§<+§ûAÚNõÙ¦a.�ý¢|µ [26, 27, 32, 35]"·��&§B\v
��wÍÔN¬��5�ïÄó�m8��kF"���"
2) ã��êþÚaO"��êþ�ã�éuÓPy¢.|µ�õ�5Ú
´L5�'�"d§�þ�êâ�±4SOD�.;�L[Ü¿Or�zU
å"�d§·�Â85g80õ�aO�6,000Üã�§Ù¥�¹3,000Ü�kw
ÍÔN�ã�Ú3,000ÜvkwÍÔN�ã�"·�òêâ8©�Ôö8§�y
8ÚÿÁ8§'~�6µ2µ2"�(�ú²5§ÿÁ8جuÙ§ ´ÏL·
���ÕJø3�ÿÁ3"Fig. 4£a¤w«z�aO�wÍÔN�êþ"§L
²/<0aOÓé�'~§ù´Ün�§Ï�<�Ï~�Ù¦é��åÑy3
F~|µ¥"
3) wÍÔN��Û/ÛÜôÚé'"X [26]¥¤ã§â�/wÍ0�cµÚ
�µ��Û/ÛÜé'Ýk'"Ïdu�wÍÔN´Ä´uuÿ´�~��"
éuz�ÔN§·�©OO�cµÚ�µ�RGBôÚ��ã",�§|^χ2å
l5ÿþü���ã�m�ål"�ÛÚÛÜôÚé'Ý©Ù©OXFig. 4£b¤
Ú£c¤¤«"�ILSO�'§·��SOC¥äk$�ÛôÚé'ÝÚÛÜôÚé
'Ý�ÔNÓâ���'~"
4) wÍÔN� �"¥% ��@�´wÍ5uÿêâ8¥K����
��� [3, 20, 26]"Fig. 4 £d¤Ð«Ñ�|ã�9ÙU\ã"�±wѧ�
,wÍ�ÔN uØÓ� �§�´U\ãE,L²ù|ã�´�3¥% �
�"±c�ÄOÿÁÏ~æ^ù«ØO(��ª5©ÛwÍÔN� �©Ù"
�;�ù«Ø�y�§·�±�Fig. 4£e¤¥ü�þroÚrm�ÚO�¹§
Ù¥ro Úrm ©OL«ÔN¥%ÚÔN¥��£>�¤:lã�¥%kõ�"
òroÚrmرã�é���Ý���±?18�z§¦�ro, rm ∈ [0, 1]"lù
ÚOê⥧·��±*�êâ8¥�wÍÔNØÉ¥% ��K�"
5) wÍÔN���"z�wÍÔN¢~����½Â�ÔN¡ÈÓã�
o¡È�'~ [26]"XFig. 4£g¤¤«§�=k�¢~?ILSOêâ8 [22]�'§
SOC¥�wÍÔN����Cz���2�"d§SOC¥�¥.ÔNäk�
p�'~"
6) p�þ�wXé�I\"�X [17]¥¤5¿��§3ECSSDêâ8£ä
k1,000�ã�¤þ�Ôö#N�.¼�'Ù¦êâ8£~X§MSRA10K§ä
k10,000�ã�¤¼��Ð�(J"ùL²Ø5��§êâ8�þ�´�
��Ï�"�¼��þp�þ�ã�§·�lMSCOCOêâ8 [27]¥�Å
ÀJã�§ù´���.�ý¢.êâ8§Ù¥�ÔN^õ>/I5£~X§
oÑI5¤"p�þI53JpSOD�.�O(5�¡�åX'��^ [1]"�
3 http://dpfan.net/SOCBenchmark/
8 Deng-Ping Fan et al.
0-0.05
0.05
0.15
0.250.35
0.45
0.1 0.3 0.5 0.7 0.9
0 0.2 0.4 0.6 0.8 1
0.050.10.150.20.25
0.05
0.1
0.15
0.2
0 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1
0.04
0.08
0.12
0.16
0.2
SOCILSO
SOC
ILSO
�aO�ÔNêþ£LOG ºÝ¤
100
101
102
103
(a)
(b)
ILSO
SOC
'~
�ÛôÚé'Ý
(c)
ILSO
SOC
'~
ÛÜôÚé'Ý
(d) (e) �©Ù
'~
rormrorm
(f)
1ìCz£AC¤
,Ï�¸£CL¤
(g)
ILSO
SOC
¢~º�
'~
Fig. 4. £a¤·��SOCêâ8¥z�aOI5�¢~�êþ"£b§c¤�ÛôÚé'
ÝÚÛÜôÚé'Ý�ÚOêâ"£d¤5g·��êâ89ÙU\ã��|wÍã"£e¤
SOC¥�wÍÔN� �©Ù"£f¤ØÓá5�Àú~f"£g¤SOCÚILSO [22]�¢~
��©Ù"
d§·�¦^Å���I55#IPêâ8"aquͶ�SOD?Ö��Ä
OÿÁêâ8 [1,2,11,19,22,23,29,32,37,41,43]§·�vk¦^úĤ��"·
�æ�õ�Ú½5Jøp�þ�5º"ùÚ½�)ü��ãµ£i¤·��
¦5�*¯¦^I½µIP¦�@�3z�ã�¥��wÍ�ÔN"£ii¤�3
�õê£≥ 3¤*¯3wÍ5þ¿��Ó�ÔN£I½µ�IOU> 0.8¤"31�
òwÍ5ÔNuÿí�#pÝ 9
(a) ILSO (b) SOC
(c) MSCOCO (d) SOC
Fig. 5. ��CJÑ�êâ8�'�£a¤¢~?êâ8ILSO [22] £^ØëY�oÑ>
.I5¤§£c¤MSCOCOêâ8 [27] £^õ>/I5¤§·��£b§d¤SOCêâ8¥
�I5>.�²w§�þ�p"
Table 2. wÍÔNã��á5�LÚ�A£ã"ÏL*ykêâ8�A�§·�o
(ùá5"�Àú«~�±3Fig. 1ÚFig. 4 £f¤¥é�"k'�õ«~§�ë
�Ö¿á�"
á5£ã
AC 111ìììCCCzzz ÔN«�¥²w�1ìCz"
BO ���ÔÔÔNNN ÔN¡ÈÚã�¡È�'��u0.5
CL ,,,ÏÏÏ ÔN±��cµÚ�µ«�äk�q�ôÚ§·�ò�ÛôÚé'Ý��u0.2ÛÜôÚé'Ý�
�u0.9�ã�IP�,Ïã�"
HO ÉÉÉ���ÔÔÔNNN dÀúþÕA/Ø�q�Ü©|¤�ÔN"
MB $$$ÄÄÄ��� du�Ž$Ä�ËĦ�ÔNäk��>."
OC ñññ ÔN�Ü©½�Üñ
OV ���ÑÑÑÀÀÀ��� ÔN�Ü©«��Ñã�>."
SC ///GGGEEE,,,555 ÔNk'Xn[|��a�E,�>.§'XÄÔ��ÚÉ
SO ���ÔÔÔNNN ÔN¡ÈÚã�¡È�'Ç�u0.1"
�ã��§·�k3,000�^I½µI5�wÍÔNã�"31��㧷��
âI½µ�J«?�ÚÃóIPwÍÔN�Å��Ó+"�5¿§·�k10¶
��öë���Ú½±��u�I5��þ"��§·��33,000Üä
kp�þ�¢~?IPwÍÔN�ã�"XFig. 5£b§d¤¤«§·��Ô
N>.�I5´°(!�ßÚ²w�"3I5L§¥§·��V\��
3MSCOCOêâ8¥IP�#aO£~X§O�Åw«ì§lf§�Þ¤ [27]
7) äká5�wXé�"3êâ8¥ã��á5&EkÏuïÄö�*µ
��.3ØÓa.�ëêþ�5U"§�#Né�.�}�¹�u�"�d§
·�½Â�|á55L«ý¢|µ¥¡��A½�¹§~X$Ä�§ñ
10 Deng-Ping Fan et al.
AC BO CL HO MB OC OV SC SO
AC
BO
CL
HO
MB
OC
OV
SC
SO
ACBO
CL
HO
MB
OC
OV
SC SO
Fig. 6. �µSOCêâ8¥wÍã��á5©Ù"��¥�z�êiL«ã��Ñyg
ê"mµÄuÑyªÇ±��á5�m�Ì��6'X"ë����°ÝL«á5éÙ
§á5��6�p"
Ú,Ï��µ£Table 2¥o(¤"�5¿§Ï�ùá5Ø´ÕÓ�§¤±��
ã��±¦^õ�á5?1I5§"
É [33]�éu§XFig. 6�>¤«§·�Ðyêâ8ã¡á5�©Ù�
¹"SOa.äk��'~´Ï�°(�¢~?£~X§Fig. 2¥��¥û¤�
I5"Ï�y¢.�|µdØÓÀúAÚ�á�|¤§¤±HOa.Óé�'
~"MB a.3Àªv¥'·�ã��~�§�k��¬Ñy3·�ã�¥"
Ïd§MBa.3·��êâ8¥Ó�é���'~"duý¢ã�Ï~�¹
õ�á5§�d·�3Fig. 6mý�âÑy�ªÇw«á5�m�Ì��6'
X"~X§�¹NõÉ�ÔN�|µ�Uäk�þ*d{¿/¤E,�m(
��ÔN"Ïd§HOa.�OCa.§OVa.ÚSO a.äkr�65"
4 ÄOÿÁ�.
3�!¥§·�3SOCêâ8þ¥y16�SOD�.�µ�(J"A�¤
kÄuòÈ ²�ä�äk�L5�SOD�.Ñ?1µ�"�´§du,
�.��èØúm§Ïd·�éda�.Ø��Ä"d§�õê�.Ñvk
�é�wÍÔNuÿ?1`z"Ïd§ú²å�§·��¦^SOCêâ8�ÿ
Á85µ�SOD�."·�3Sec. 4.1¥£ãµ��I"SOCêâ8��N�
.5U�Sec. 4.2ÚTable 3 ¥§ �é��á5�5Uµ�(J£~X§1ì
Czá5þ�5ULy¤�Sec. 4.3ÚTable 4 ¥"·�úmµ���¿�3
�Õþ¬Jø3�µ�ÿÁ"
òwÍ5ÔNuÿí�#pÝ 11
4.1 µ��I
3riÒµ�µe¥§�½dSOD�.)¤�ýÿãM9ý�ù�G§·
�MF"uµ��Iw�·�ľ´=�«�.U)¤�Z(J"3ùp§
·�3SOCêâ8þ¦^n«ØÓ�µ��I5µ�SOD�."
Å��°Ýε "«��q5µ��{Ø�ÄýKa�wÍ5©Ù"��Ö
Ö¿§·��O�MÚG�m�8�z£[0,1]¤��²þýéØ�£MAE¤§ò
Ù½Â�µ
ε =1
W ×H
W∑x=1
H∑y=1
||M(x, y)−G(x, y)||, (1)
Ù¥WÚH©O´ã��°ÝÚpÝ"
«��q5F "�ÿþüÜã¡��«�����§Ý§·�¦
^F -measure§T�{½ÂXeµ
F =(1 + β2)Precision×Recallβ2Precision+Recall
, (2)
Ù¥β2 = 0.3d [1]JÑ¿^u²ïð£ÇÚ°Ý", §3O�ð£ÇÚ
°Ý�§F -measure¿vkéçÚ£�"ݤý�ã?1½Â"3ù«�¹
e§ØÓ�cµã¬���Ó�(J£0¤§ ùw,´ØÜn�"l �±�
Ñ(اF -measureØ·Üuµ��wÍÔNuÿ�(J"
nþ§εÚFùü��IÑ´ÄuÅ���O��ª§Ïd²~�Ñ(��
q5"1�ÀúïÄL²<aÀúXÚé|µ(��~¯a [13]"3NõA^
¥§·�Ï"SOD�.�uÿ(JU�3ÔN�(�&E"
(��q5S "Fan�< [13] JÑ�S-measureÓ��Ä«�ÚÔNü��
gþ��q55µ�uÿ(J�(��q5"Ïd§·��¦^S-measure5
µ�MÚG �m�(��q5"I�AO5¿�´·�é�e5�N5ULy
�µ�Ú©Û´ÄuS-measure?1�"
4.2 �IÚO
�¼��N(J§·�éµ��I�©ê�²þ�"�η ∈ {F, ε, S}§Ýþ²þ�©�úª�µ
Mη(D) =1
|D|∑I∈D
η(Ii), (3)
Ù¥η(Ii)´�.3ã�êâ8DSã�Iiþ�µ��©"
12 Deng-Ping Fan et al.
Table 3. n«�IeSOD�.�5U"F�L«��q5§ε´²þýéØ�§S ´(
��q5"↑�Lêi�p�Ч��½,"µ�(J�âª(3)3SOCêâ8þÏLO
��Ñ"Sall, Fall, εall ©OL«^S, F, ε�I5L«��N5ULy"\oL«�Ф
1"
ü?Ö õ?Ö
���III LEGS MC MDF DCL AMU RFCN DHS ELD DISC IMC UCF DSS NLDF DS WSS MSR
[36] [51] [23] [24] [24] [38] [28] [15] [8] [48] [50] [17] [31] [25] [37] [22]
Fall ↑ .276 .291 .307 .339 .341 .435 .360 .317 .288 .352 .333 .341 .352 .347 .327 .380
Sall ↑ .677 .757 .736 .771 .737 .814 .804 .776 .737 .664 .657 .807 .818 .779 .785 .819
εall ↓ .230 .138 .150 .157 .185 .113 .118 .135 .173 .269 .282 .111 .104 .155 .133 .113
ü?Öµéuü?Ö�.§3��SOCêâ8þ5ULy£Table 3¥
�Sall¤�Z��.´NLDF [31] (MS = 0.818)§Ùg´RFCN [38] (MS =
0.814)"MDF [23]ÚAMU [49]¦^>��¢5J,wÍãJ��O(Ý�%
�U��n��8I"�¦^ã��ÛÜ«�&E§MC [51]§MDF [23]§
ELD [15]ÚDISC [8]}Á¦^����{òã�©�¤ê�«�§,�lù
«�¥J�A�§�ù´��E, Ñ��"�?�ÚJp5U§UCF [50]§
DSS [17]§NLDF [31]ÚAMU [49]|^�òÈ�ä5UõSOD�.�5U£Table
4¥�£Ssal¤¤"Ù¦��{ÃXDCL [24]ÚIMC [48] K}Áò�����ò
È�ä(Üå5�ï��r���."d§RFCN [38]ò�)>�Ú���
�ü��'�¢|Ü��òÈ�ä¥? 3��êâ8þ¼�ûÐ�5U
£MF = 0.435, MS = 0.814¤"
õ?Öµ�þã�.ØÓ§MSR [22]¦^n����'�Ú½�uÿ¢~
?wÍÔNµ�OwÍã§uÿwÍÔNÓ+§±9£OwÍÔN�¢~"§
Mï��õºÝwÍ5uÿ�ä§�±¢y�p5U£Sall¤"Ù¦ü�õ?
Ö�.DS [25]ÚWSS [37]Ó�|^©�Ú©a(J)¤wÍãl ¼�·Ý�
5UJ,"���J�´§¦+WSS´�«fiÒ�õ?Ö�.§�§E,�
±¢y�Ù¦�iÒ�ü?Ö�.���5U"Ïd§ÄufiÒÚõ?Ö�
�.�U´�5�ïÄ��"
4.3 Äuá5�µ�
XSec. 3ÚTable 2¤«§·��wÍã�©�á5"z�á5�Ly¢
.|µ�wÍ5uÿ¥�3�]Ô5¯K"ùá5#N·�£OäkÌ�5
A�£~X§,Ï�¸��3¤�ã�8ܧ§�éuSOD�.�5U)º±
9XÛòSOD�¡�A^�?Ö�'é´�~��"~X§sketch2photoA
òwÍ5ÔNuÿí�#pÝ 13
Table 4. 3SOCwÍ5ÔNfêâ8þÄuá5�5ULy"éuz���.§©ê
éAu3A½á5�¤kÿÁã�þ�(��q5MS£�Sec. 4.1¤�²þ�§©ê�
p5ULy�Ч\oL«�p¤1§²þwÍÔNuÿ5USsal31�1ÏL(��
q5S¥y§+Ú−©OL«�²þ��'�e�5UO\Ú~�"
ü?Ö õ?Ö
ááá555LEGSMC MDF DCL AMU RFCNDHSELDDISCIMC UCF DSS NLDFDS WSSMSR
[36] [51] [23] [24] [24] [38] [28] [15] [8] [48] [50] [17] [31] [25] [37] [22]
Ssal .607 .619 .610 .705 .705 .709 .728 .664 .629 .679 .678 .698 .714 .719 .676 .748
AC .625 .631 .614 .734 .736 .744 .745 .673 .644 .702 .714 .726 .737 .764 .691 .789
BO .509 .490 .461− .610 .569 .540 .590 .576 .517 .701+ .636 .496− .568 .685 .566 .667
CL .620 .635 .566 .699 .708 .714 .743 .658 .635 .696 .704 .677− .713 .729 .678 .756
HO .666 .666 .648 .745 .755 .759 .766 .706 .681 .715 .744 .748 .755 .756 .707 .777
MB .543− .603 .615 .693 .706 .715 .722 .639 .600 .689 .682 .695 .685 .711 .641 .757
OC .609 .617 .608 .708+ .725+ .711 .716 .658 .630 .672 .701+ .689 .709 .725+ .672 .740
OV .548 .584 .568 .699 .708+ .687 .706 .637 .573 .693+ .685+ .665 .688 .722+ .624 .743
SC .608 .620 .669+ .738 .731 .735 .763 .688 .653 .690 .722+ .746+ .745 .724 .677 .773
SO .573− .601 .621 .691 .685 .698 .713 .644 .614 .648− .650 .696− .703 .696 .659 .730
^ [7]�U�3�ÔNþäkûÐ5U��.§ ù�±ÏLÄuá5�5Uµ
��{5EO"
(J3Table 4¥§·�w«�«SOD�.3A½á5L��êâ8f8
þ�5U"du�Ì��§3±eÜ©¥§·�=ÀJ��L5á5?1?
�Ú©Û"�õ[!�±3Ö¿á�¥é�"
�ÔN£BO¤ �ÔN��ÅåléC�§²~¬Ñy�ÔN£BO¤|µ§
Ïd3ã¡¥�±�Ù/w����©i½ãY", 3ù«�¹e§��u
'5ÛÜ&E��.ò�îØ�§�����5U��£~X§DSS [17]��
28.9 �5U§MC [51] ��20.8 �5U±9RFCN [38]��23.8 �5
U¤", §IMC [48] �.�5ULyÑ�þ,3.2%"3�\)T�.�6
§�§·��Ñ��Ün�)º§=IMC¦^oÑýÿã5L��§¿|
^LÝ©��ã�5Ö¿(�&E§l 3BOa.ã�þ¼�-<÷¿�(
J"�´§LÝ©��ã�Ã{�Ö"��[!§Ïd¬��da�.3SOa
.ã�þ�5Ueü4.6%"
�ÔN£SO¤éu¤kSOD�.5`§£OSOa.Ñ´����(J
�¯K"3daã�þ¤k�.Ñ��5Ueü£~X§lDSS [17]�e
ü0.3%�LEGS [36]�eü5.6%¤§Ï�3òÈ ²�ä�eæ�ÏméN´�
À�ÔN"DSS [17]´����3SOa.ã�þ5U=kÑ�eü��.§
§3BOa.ã�þ�5U����£28.9%¤"MDF [23]¦^õºÝ���ã
����ä�Ñ\§Ïd§UéÐ/�3�ÔN�[!", §du��
14 Deng-Ping Fan et al.
����k�§MDFEÃ{k�/a��Û�§��3BOa.ã�þÑy
��£O�}VÇ"
ñ£OC¤3ñ|µ¥§ÔN�Ü©ñ"Ïd§SOD�.I�Ó¼�
Û�±�ÖÔN&E�Ø��"�d§DS [25]ÚAMU [49] |^eæ�L§¥
�õºÝA�)¤KÜwÍã¶UCF [50]JÑ�«Ø(½�ÆSÅ�5ÆS
Ø(½�òÈA�"¤kù�{ÑÁã¼��¹�ÛÚÛÜA��wÍã"
ØѤ�§ù�{3OCa.þ����Ø���J"Äuþã©Û§·
��uyùn��.3I��õ�Â&E�|µþ�5ULy�~ЧXAC§
OVÚCLa."
É�ÔN£HO¤a.|µ3y¢)¹¥é~�"3HOa.�ã�þØÓ
�.�5U©O'Ù3oêâ8þ�²þ5Uk¤J,§Ä�?u3.9 % �9.7
%�m�ÅÄþ"·�~¦ù´Ï�HOa.Óoêâ8�'~��§�*þ
¦SOD�.�·Üù�á5"ù�(J3,«§Ýþy¢·�3Fig. 6¥�Ú
Oêâ"
5 ?ØÚ(Ø
â·�¤�§ù�ó�JÑ8c��ºÝ��éÄuòÈ ²�ä�w
Í5ÔNuÿ�.�5Uµ��Y"·��©Û�ÑykSODêâ8¥�3
�î�êâÀJ �"ù«�O ����k?�SOD�{3ykêâ8þ
?1µ��A���AC�Ú�p5ULy§�3A^uy¢.F~|µ
�E,��ØU-<÷¿"Äu·��©Û§·�Äk(½�¡Ú²ï�ê
â8AT÷v�7 ���¡"·�Äk�ïp�þ�SODêâ8SOC"§
�¹5gF~)¹�g,�¸¥���Cý¢�¸�wÍÔNã�"SOCêâ
8ò�X�m�í£ uÐÚO�§¿ò3õ���þÿ°ïÄ��15§~
X§wÍÔN�aê [46]§¢~?wÍ5ÔNuÿ [22]§ÄufiÒ�wXé
�uÿ [37]��",�§���\/)SOD¯K§ïÄSOD�{�`":§
¿3ØÓ�*:Ú�¦e�*/µ��.5U§·�JÑ�|á5£~X§
*Cz¤"��§·�3SOCêâ8þéykSOD�.?1Äuá5�5U
µ�§µ��(J��5��.muÚ�.'�m8¿÷F"�#��"
��
�ïÄ��I[g,�ÆÄ7£NO.61620106008,61572264¤§I[�c
<â|±Oy§U9½#Ñ�cÆög,�ÆÄ7£NO.17JCJQJC43700¤§u
�M#ïÄOy�|±"
òwÍ5ÔNuÿí�#pÝ 15
References
1. Achanta, R., Hemami, S., Estrada, F., Susstrunk, S.: Frequency-tuned salient re-
gion detection. In: CVPR. pp. 1597–1604. IEEE (2009)
2. Alpert, S., Galun, M., Basri, R., Brandt, A.: Image Segmentation by Probabilistic
Bottom-Up Aggregation and Cue Integration. In: CVPR. pp. 1–8. IEEE (2007)
3. Borji, A., Cheng, M.M., Jiang, H., Li, J.: Salient object detection: A benchmark.
IEEE TIP 24(12), 5706–5722 (2015)
4. Borji, A., Itti, L.: State-of-the-art in visual attention modeling. IEEE TPAMI
35(1), 185–207 (2013)
5. Borji, A., Sihite, D.N., Itti, L.: Salient object detection: a benchmark. In: ECCV.
pp. 414–429. Springer (2012)
6. Caesar, H., Uijlings, J., Ferrari, V.: COCO-Stuff: Thing and Stuff Classes in Con-
text. In: CVPR. IEEE (2018)
7. Chen, T., Cheng, M.M., Tan, P., Shamir, A., Hu, S.M.: Sketch2photo: Internet
image montage. ACM Transactions on Graphics (TOG) 28(5), 124 (2009)
8. Chen, T., Lin, L., Liu, L., Luo, X., Li, X.: DISC: Deep image saliency computing
via progressive representation learning. IEEE transactions on neural networks and
learning systems 27(6), 1135–1149 (2016)
9. Cheng, M.M., Hou, Q.B., Zhang, S.H., Rosin, P.L.: Intelligent Visual Media Pro-
cessing: When Graphics Meets Vision. Journal of Computer Science and Technol-
ogy 32(1), 110–121 (2017)
10. Cheng, M.M., Mitra, N.J., Huang, X., Hu, S.M.: Salientshape: group saliency in
image collections. The Visual Computer 30(4), 443–453 (2014)
11. Cheng, M.M., Mitra, N.J., Huang, X., Torr, P.H.S., Hu, S.M.: Global contrast
based salient region detection. IEEE TPAMI 37(3), 569–582 (2015)
12. Everingham, M., Van Gool, L., Williams, C.K.I., Winn, J., Zisserman, A.: The
PASCAL Visual Object Classes Challenge 2010 (VOC2010) Results
13. Fan, D.P., Cheng, M.M., Liu, Y., Li, T., Borji, A.: Structure-measure: A New Way
to Evaluate Foreground Maps. In: ICCV. pp. 4548–4557. IEEE (2017)
14. Fan, D.P., Gong, C., Cao, Y., Ren, B., Cheng, M.M., Borji, A.: Enhanced-alignment
Measure for Binary Foreground Map Evaluation. In: International Joint Conference
on Artificial Intelligence (IJCAI). pp. 698–704 (2018)
15. Gayoung, L., Yu-Wing, T., Junmo, K.: Deep Saliency with Encoded Low level
Distance Map and High Level Features. In: CVPR. IEEE (2016)
16. He, J., Feng, J., Liu, X., Cheng, T., Lin, T.H., Chung, H., Chang, S.F.: Mobile
product search with bag of hash bits and boundary reranking. In: CVPR. pp.
3005–3012. IEEE (2012)
16 Deng-Ping Fan et al.
17. Hou, Q., Cheng, M.M., Hu, X., Borji, A., Tu, Z., Torr, P.: Deeply supervised salient
object detection with short connections. IEEE TPAMI (2018)
18. Hou, Q., Dokania, P.K., Massiceti, D., Wei, Y., Cheng, M.M., Torr, P.H.S.: Bottom-
Up Top-Down Cues for Weakly Supervised Semantic Segmentation. In: EMM-
CVPR. IEEE (2017)
19. Jiang, H., Cheng, M.M., Li, S.J., Borji, A., Wang, J.: Joint Salient Object Detection
and Existence Prediction. Front. Comput. Sci (2018)
20. Judd, T., Durand, F., Torralba, A.: A benchmark of computational models of
saliency to predict human fixations. In: MIT Technical Report (2012)
21. Lazebnik, S., Schmid, C., Ponce, J.: A sparse texture representation using local
affine regions. IEEE TPAMI 27(8), 1265–1278 (2005)
22. Li, G., Xie, Y., Lin, L., Yu, Y.: Instance-Level Salient Object Segmentation. In:
CVPR. pp. 247–256. IEEE (2017)
23. Li, G., Yu, Y.: Visual saliency based on multiscale deep features. In: CVPR. pp.
5455–5463. IEEE (2015)
24. Li, G., Yu, Y.: Deep Contrast Learning for Salient Object Detection. In: CVPR.
pp. 478–487. IEEE (2016)
25. Li, X., Zhao, L., Wei, L., Yang, M.H., Wu, F., Zhuang, Y., Ling, H., Wang, J.:
DeepSaliency: Multi-task deep neural network model for salient object detection.
IEEE TIP 25(8), 3919–3930 (2016)
26. Li, Y., Hou, X., Koch, C., Rehg, J.M., Yuille, A.L.: The secrets of salient object
segmentation. In: CVPR. pp. 280–287. IEEE (2014)
27. Lin, T.Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollar,
P., Zitnick, C.L.: Microsoft COCO: Common objects in context. In: ECCV. pp.
740–755. Springer (2014)
28. Liu, N., Han, J.: DHSNet: Deep Hierarchical Saliency Network for Salient Object
Detection. In: CVPR. pp. 678–686. IEEE (2016)
29. Liu, T., Sun, J., Zheng, N., Tang, X., Shum, H.Y.: Learning to Detect A Salient
Object. In: CVPR. pp. 1–8. IEEE (2007)
30. Long, J., Shelhamer, E., Darrell, T.: Fully convolutional networks for semantic
segmentation. In: CVPR. pp. 3431–3440. IEEE (2015)
31. Luo, Z., Mishra, A.K., Achkar, A., Eichel, J.A., Li, S., Jodoin, P.M.: Non-local
deep features for salient object detection. In: CVPR. vol. 2, p. 7 (2017)
32. Martin, D., Fowlkes, C., Tal, D., Malik, J.: A database of human segmented natural
images and its application to evaluating segmentation algorithms and measuring
ecological statistics. In: ICCV. vol. 2, pp. 416–423. IEEE (2001)
33. Perazzi, F., Pont-Tuset, J., McWilliams, B., Van Gool, L., Gross, M., Sorkine-
Hornung, A.: A benchmark dataset and evaluation methodology for video object
segmentation. In: CVPR. pp. 724–732. IEEE (2016)
òwÍ5ÔNuÿí�#pÝ 17
34. Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z.,
Karpathy, A., Khosla, A., Bernstein, M., et al.: Imagenet large scale visual recog-
nition challenge. IJCV 115(3), 211–252 (2015)
35. Wang, J., Jiang, H., Yuan, Z., Cheng, M.M., Hu, X., Zheng, N.: Salient Object
Detection: A Discriminative Regional Feature Integration Approach. IJCV 123(2),
251–268 (2017)
36. Wang, L., Lu, H., Ruan, X., Yang, M.H.: Deep networks for saliency detection via
local estimation and global search. In: CVPR. pp. 3183–3192. IEEE (2015)
37. Wang, L., Lu, H., Wang, Y., Feng, M., Wang, D., Yin, B., Ruan, X.: Learning to
detect salient objects with image-level supervision. In: CVPR. pp. 136–145. IEEE
(2017)
38. Wang, L., Wang, L., Lu, H., Zhang, P., Ruan, X.: Saliency detection with recurrent
fully convolutional networks. In: ECCV. pp. 825–841. Springer (2016)
39. Wei, Y., Feng, J., Liang, X., Cheng, M.M., Zhao, Y., Yan, S.: Object Region Min-
ing with Adversarial Erasing: A Simple Classification to Semantic Segmentation
Approach. In: CVPR. IEEE (2017)
40. Wei, Y., Liang, X., Chen, Y., Shen, X., Cheng, M.M., Feng, J., Zhao, Y., Yan, S.:
STC: A simple to complex framework for weakly-supervised semantic segmenta-
tion. IEEE TPAMI (2017)
41. Xia, C., Li, J., Chen, X., Zheng, A., Zhang, Y.: What is and what is not a salient
object? learning salient object detector by ensembling linear exemplar regressors.
In: CVPR. IEEE (2017)
42. Xie, S., Tu, Z.: Holistically-nested edge detection. In: ICCV. pp. 1395–1403. IEEE
(2015)
43. Yan, Q., Xu, L., Shi, J., Jia, J.: Hierarchical saliency detection. In: CVPR. pp.
1155–1162. IEEE (2013)
44. Yang, C., Zhang, L., Lu, H., Ruan, X., Yang, M.H.: Saliency detection via graph-
based manifold ranking. In: CVPR. pp. 3166–3173. IEEE (2013)
45. Zhang, G.X., Cheng, M.M., Hu, S.M., Martin, R.R.: A Shape-Preserving Approach
to Image Resizing. Computer Graphics Forum 28(7), 1897–1906 (2009)
46. Zhang, J., Ma, S., Sameki, M., Sclaroff, S., Betke, M., Lin, Z., Shen, X., Price, B.,
Mech, R.: Salient object subitizing. In: CVPR. pp. 4045–4054. IEEE (2015)
47. Zhang, J., Sclaroff, S., Lin, Z., Shen, X., Price, B., Mech, R.: Unconstrained salient
object detection via proposal subset optimization. In: CVPR. pp. 5733–5742. IEEE
(2016)
48. Zhang, J., Dai, Y., Porikli, F.: Deep Salient Object Detection by Integrating Multi-
level Cues. In: Winter Conference on Applications of Computer Vision (WACV).
pp. 1–10. IEEE (2017)
18 Deng-Ping Fan et al.
49. Zhang, P., Wang, D., Lu, H., Wang, H., Ruan, X.: Amulet: Aggregating multi-level
convolutional features for salient object detection. In: ICCV. pp. 202–211 (2017)
50. Zhang, P., Wang, D., Lu, H., Wang, H., Yin, B.: Learning Uncertain Convolutional
Features for Accurate Saliency Detection. In: ICCV. pp. 212–221 (2017)
51. Zhao, R., Ouyang, W., Li, H., Wang, X.: Saliency detection by multi-context deep
learning. In: CVPR. pp. 1265–1274. IEEE (2015)