Object Detection Algorithm Based on Alternate-Attention

The one-stage object detection algorithm, YOLOv3, has a fast detection speed and can meet real-time requirements. But giving the same attention weight to all grids during detection will result in the inability to highlight the detection subject, so the positioning accuracy of the bounding box is sti...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2021 2nd International Conference on Computer Science and Management Technology (ICCSMT) S. 376 - 381
Hauptverfasser: He, Xuejie, Bai, Chenyan, Qi, Haoru, Liu, Honghong
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.11.2021
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract The one-stage object detection algorithm, YOLOv3, has a fast detection speed and can meet real-time requirements. But giving the same attention weight to all grids during detection will result in the inability to highlight the detection subject, so the positioning accuracy of the bounding box is still room for improvement. In order to improve the detection accuracy, this paper proposes an Alternate-Attention mechanism, using the global pooled attention mechanism to highlight the overall characteristics, and the self-attention mechanism to reflect the self-weight relationship between features. The two attention mechanisms are alternated and applied to the two dimensions of channel and space, and finally enhance the features extracted by Darknet-53. Experiments on the PASCAL VOC2007 dataset shows that this algorithm can effectively improve the detection accuracy. Compared with Faster RCNN, YOLO series and SSD series algorithms, the mAPlouo.5 value of this algorithm is higher, up to 80.24.
AbstractList The one-stage object detection algorithm, YOLOv3, has a fast detection speed and can meet real-time requirements. But giving the same attention weight to all grids during detection will result in the inability to highlight the detection subject, so the positioning accuracy of the bounding box is still room for improvement. In order to improve the detection accuracy, this paper proposes an Alternate-Attention mechanism, using the global pooled attention mechanism to highlight the overall characteristics, and the self-attention mechanism to reflect the self-weight relationship between features. The two attention mechanisms are alternated and applied to the two dimensions of channel and space, and finally enhance the features extracted by Darknet-53. Experiments on the PASCAL VOC2007 dataset shows that this algorithm can effectively improve the detection accuracy. Compared with Faster RCNN, YOLO series and SSD series algorithms, the mAPlouo.5 value of this algorithm is higher, up to 80.24.
Author Qi, Haoru
He, Xuejie
Bai, Chenyan
Liu, Honghong
Author_xml – sequence: 1
  givenname: Xuejie
  surname: He
  fullname: He, Xuejie
  email: 219100201O@cnu.edu.cn
  organization: Capital Normal University, College of Information Engineering,Beijing,China
– sequence: 2
  givenname: Chenyan
  surname: Bai
  fullname: Bai, Chenyan
  email: cybai@cnu.edu.cn
  organization: Capital Normal University, College of Information Engineering,Beijing,China
– sequence: 3
  givenname: Haoru
  surname: Qi
  fullname: Qi, Haoru
  organization: Capital Normal University, College of Information Engineering,Beijing,China
– sequence: 4
  givenname: Honghong
  surname: Liu
  fullname: Liu, Honghong
  organization: Capital Normal University, College of Information Engineering,Beijing,China
BookMark eNotj8lKA0EURUvQhcZ8gSD9A93W9GpYtq3GQCQLex9qeKUlSbV0auPfG4fVgcvhwrki52UqSMgtox1j1N6th-H1ZQQJHDpOOesopRrOyNJqw5QCyakS7JLorf_AUJsHrCfkqTT9_m2ac30_NPfuiLH5nSrOxVVs-1qx_GjX5CK5_RGX_1yQ8elxHJ7bzXa1HvpNm6WBVvAQETSohEHJwLlKEaMXOgqrqEqSCu-TsY55qYAJFAl4MEZy1MY5LRbk5u82I-Luc84HN3_tTg3KAhXfvqNECA
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/ICCSMT54525.2021.00075
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Computer Science
EISBN 9781665420631
1665420634
EndPage 381
ExternalDocumentID 9786950
Genre orig-research
GroupedDBID 6IE
6IL
CBEJK
RIE
RIL
ID FETCH-LOGICAL-i485-32cde5756fec64c226fdedb37d39606f403bbf89a1b46513e3f52c8842e78aa73
IEDL.DBID RIE
IngestDate Thu Jun 29 18:36:44 EDT 2023
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i485-32cde5756fec64c226fdedb37d39606f403bbf89a1b46513e3f52c8842e78aa73
PageCount 6
ParticipantIDs ieee_primary_9786950
PublicationCentury 2000
PublicationDate 2021-Nov.
PublicationDateYYYYMMDD 2021-11-01
PublicationDate_xml – month: 11
  year: 2021
  text: 2021-Nov.
PublicationDecade 2020
PublicationTitle 2021 2nd International Conference on Computer Science and Management Technology (ICCSMT)
PublicationTitleAbbrev ICCSMT
PublicationYear 2021
Publisher IEEE
Publisher_xml – name: IEEE
Score 1.7749896
Snippet The one-stage object detection algorithm, YOLOv3, has a fast detection speed and can meet real-time requirements. But giving the same attention weight to all...
SourceID ieee
SourceType Publisher
StartPage 376
SubjectTerms channel attention
Computational modeling
Computer science
Feature extraction
Object detection
Real-time systems
spatial attention
YOLOv3 algorithm
Title Object Detection Algorithm Based on Alternate-Attention
URI https://ieeexplore.ieee.org/document/9786950
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1NSwMxEA21ePBUtRW_ycGjqd0k3STHWi0KWgsW6a3kY6IF3ZW69fe7ya4VwYu3MJfAC-HNTOblIXTmpaK-ZylRnmvCBUuJSrUjzFHtUwupiQ23pzsxHsvZTE0a6HythQGAOHwG3bCMb_kut6vQKgu_waYqFOgbQohKq1WLfpOeurgdDh_vp8Ezu1_WfTTpRj785ZoSSWPU-t9226jzo77DkzWv7KAGZLuo9W2_gOvb2EbiwYQmCr6CIs5TZXjw-pyXxf7LG74sycnhGKo6fkAGRVGNNnbQdHQ9Hd6Q2geBLLjsE0atgzKrSj3YlNsyX_IOnGHCsVB-eN5jxpSY68QEY3MGzPeplZJTEFJrwfZQM8sz2EfYSMvBCm4Sn5TspaUxjnpurLeaOwUHqB1gmL9XP13MawQO_w4foa2Ac6XMO0bNYrmCE7RpP4vFx_I0Hs8XNa6TzQ
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1NSwMxEA1FBT1VbcVv9-DR1G6S3WSPtVpabGvBRXor-ZhoQbdSt_5-N9m1InjxFuYSeCG8mcm8PIQurUiIbWuCE8skZpzGOImlwdQQaWMNsfINt6chH4_FdJpMauhqrYUBAD98Bi239G_5ZqFXrlXmfoONE1egb0aMkbBUa1Wy37CdXA-63cdR6lyzo6LyI2HLM-Iv3xRPG736_zbcRc0f_V0wWTPLHqpBto_q3wYMQXUfG4g_KNdGCW4h9xNVWdB5fV4U5f7LW3BT0JMJfKjs-QHu5Hk53NhEae8u7fZx5YSA50xEmBJtoMirYgs6ZrrImKwBoyg31BUglrWpUgXqMlTO2pwCtRHRQjACXEjJ6QHayBYZHKJACc1Ac6ZCGxb8JYVShlimtNWSmQSOUMPBMHsv_7qYVQgc_x2-QNv9dDScDQfj-xO04zAvdXqnaCNfruAMbenPfP6xPPdH9QUVm5cU
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=proceeding&rft.title=2021+2nd+International+Conference+on+Computer+Science+and+Management+Technology+%28ICCSMT%29&rft.atitle=Object+Detection+Algorithm+Based+on+Alternate-Attention&rft.au=He%2C+Xuejie&rft.au=Bai%2C+Chenyan&rft.au=Qi%2C+Haoru&rft.au=Liu%2C+Honghong&rft.date=2021-11-01&rft.pub=IEEE&rft.spage=376&rft.epage=381&rft_id=info:doi/10.1109%2FICCSMT54525.2021.00075&rft.externalDocID=9786950