基于结构感知和全局上下文信息的小目标检测

TP391.4; 在小目标检测任务中,由于小目标像素值少、特征不丰富和难提取等局限性,容易导致模型漏检、误检以及精度低等问题,提出了一种基于多尺度结构感知和全局上下文信息的小目标检测算法.针对复杂场景设计了多尺度结构感知模块,可以更好地捕获小目标的细节特征,以此增强模型识别不同尺寸物体的检测能力.为了获取更多的全局特征,借助Transformer捕获长距离依赖特征的优势设计了全局上下文信息模块,有效地建立起不同区域像素点之间的联系.针对模型训练时的梯度爆炸现象,设计了一种新的带权重损失函数W-CIoU,使得训练时的收敛速度有明显改善.大量的实验结果表明,提出的方法相较于其他经典的轻量级方法取得...

Full description

Saved in:
Bibliographic Details
Published in计算机工程与应用 Vol. 60; no. 9; pp. 292 - 298
Main Authors 李钟华, 林初俊, 朱恒亮, 廖诗宇, 白云起
Format Journal Article
LanguageChinese
Published 福建理工大学 计算机科学与数学学院,福州 350118 01.05.2024
Subjects
损失函数
注意力机制
small object detection
loss function
上下文信息
attention mechanism
context information
小目标检测
Online AccessGet full text
ISSN1002-8331
DOI10.3778/j.issn.1002-8331.2302-0275

Cover

Abstract TP391.4; 在小目标检测任务中,由于小目标像素值少、特征不丰富和难提取等局限性,容易导致模型漏检、误检以及精度低等问题,提出了一种基于多尺度结构感知和全局上下文信息的小目标检测算法.针对复杂场景设计了多尺度结构感知模块,可以更好地捕获小目标的细节特征,以此增强模型识别不同尺寸物体的检测能力.为了获取更多的全局特征,借助Transformer捕获长距离依赖特征的优势设计了全局上下文信息模块,有效地建立起不同区域像素点之间的联系.针对模型训练时的梯度爆炸现象,设计了一种新的带权重损失函数W-CIoU,使得训练时的收敛速度有明显改善.大量的实验结果表明,提出的方法相较于其他经典的轻量级方法取得了较好的检测效果.与基准模型相比,提出的模型在VisDrone数据集上mAP50和mAP50:95分别提高了6.4和4.6个百分点,同时在TinyPerson数据集上也有着不错的表现.
AbstractList TP391.4; 在小目标检测任务中,由于小目标像素值少、特征不丰富和难提取等局限性,容易导致模型漏检、误检以及精度低等问题,提出了一种基于多尺度结构感知和全局上下文信息的小目标检测算法.针对复杂场景设计了多尺度结构感知模块,可以更好地捕获小目标的细节特征,以此增强模型识别不同尺寸物体的检测能力.为了获取更多的全局特征,借助Transformer捕获长距离依赖特征的优势设计了全局上下文信息模块,有效地建立起不同区域像素点之间的联系.针对模型训练时的梯度爆炸现象,设计了一种新的带权重损失函数W-CIoU,使得训练时的收敛速度有明显改善.大量的实验结果表明,提出的方法相较于其他经典的轻量级方法取得了较好的检测效果.与基准模型相比,提出的模型在VisDrone数据集上mAP50和mAP50:95分别提高了6.4和4.6个百分点,同时在TinyPerson数据集上也有着不错的表现.
Abstract_FL In the small targets detection task,it's difficult to detect the small objects with limited pixels and unabundant characteristics.So it is easy to cause many problems,such as missed or failed detection,and low accuracy of the model.To address these issues,this paper proposes a novel multi-scale structure perception and global contextual information net-work for small object detection.Firstly,a multi-scale structure perception module(MSSP)is proposed to capture the detailed features of small targets,it can enhance the ability of model to identify objects with different sizes.Secondly,in order to obtain more global features,a global context module(GCM)is introduced to extract global information and effectively establish the relationship between different pixels.Finally,a new IoU loss function,namely W-CIoU,is designed for the tiny objects detection,which can relieve the gradient explosion phenomenon caused by too few small target pixels during model training.Extensive experiments show that the proposed approach achieves higher accuracy than other classic light-weight methods.Compared with the baseline,the proposed model obtains over 6.4 percentage points mAP50 gain and 4.6 percentage points mAP50:95 gain on the VisDrone dataset,and also has a good performance on the TinyPerson dataset.
Author 朱恒亮
廖诗宇
李钟华
林初俊
白云起
AuthorAffiliation 福建理工大学 计算机科学与数学学院,福州 350118
AuthorAffiliation_xml – name: 福建理工大学 计算机科学与数学学院,福州 350118
Author_FL LI Zhonghua
ZHU Hengliang
LIN Chujun
BAI Yunqi
LIAO Shiyu
Author_FL_xml – sequence: 1
  fullname: LI Zhonghua
– sequence: 2
  fullname: LIN Chujun
– sequence: 3
  fullname: ZHU Hengliang
– sequence: 4
  fullname: LIAO Shiyu
– sequence: 5
  fullname: BAI Yunqi
Author_xml – sequence: 1
  fullname: 李钟华
– sequence: 2
  fullname: 林初俊
– sequence: 3
  fullname: 朱恒亮
– sequence: 4
  fullname: 廖诗宇
– sequence: 5
  fullname: 白云起
BookMark eNo9jT1Lw0Ach2-oYK39Em4Oif-7f3KXAxcpvkHBReeSXJLSICl4iGSrUClYsaDgoGAUEZx8GYRSB79McvkaFhSn38MzPL8lUkv7aUTICgUbhfDWErundWpTAGZ5iNRmOCdgwq2R-r9dJE2tewG4FIUrUNbJepnPitll9XVt7odmmFf5c3l1UZ69lB-DYnpeTMfmZlR8P5rTt-p2WL5PqrtX8zAyTwPzOV4mC7F_qKPm3zbIwdbmfmvHau9t77Y22pamwMEKkAO6sRcJBz3uUi4Zou9TGiiumKJRLFXMQmRCsMDDQDqBiljkcwFShSHHBln97Z74aeyn3U7SPz5K54-dRCddlWUZA-aABAT8AcAWY2A
ClassificationCodes TP391.4
ContentType Journal Article
Copyright Copyright © Wanfang Data Co. Ltd. All Rights Reserved.
Copyright_xml – notice: Copyright © Wanfang Data Co. Ltd. All Rights Reserved.
DBID 2B.
4A8
92I
93N
PSX
TCJ
DOI 10.3778/j.issn.1002-8331.2302-0275
DatabaseName Wanfang Data Journals - Hong Kong
WANFANG Data Centre
Wanfang Data Journals
万方数据期刊 - 香港版
China Online Journals (COJ)
China Online Journals (COJ)
DatabaseTitleList
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
DocumentTitle_FL Small Object Detection Based on Structure Perception and Global Context Information
EndPage 298
ExternalDocumentID jsjgcyyy202409030
GrantInformation_xml – fundername: (福建省自然科学基金项目); (福建省教育厅中青年教师教育科研项目); (福建省教育厅中青年教师教育科研项目); (福建工程学院科技项目)
  funderid: (福建省自然科学基金项目); (福建省教育厅中青年教师教育科研项目); (福建省教育厅中青年教师教育科研项目); (福建工程学院科技项目)
GroupedDBID -0Y
2B.
4A8
5XA
5XJ
92H
92I
93N
ABJNI
ACGFS
ALMA_UNASSIGNED_HOLDINGS
CCEZO
CUBFJ
CW9
PSX
TCJ
TGT
U1G
U5S
ID FETCH-LOGICAL-s1060-b36035f8e743865169233aa11bc6c2c1ef9cf2d32772b83b94bce2ea6709cdd63
ISSN 1002-8331
IngestDate Thu May 29 04:10:55 EDT 2025
IsPeerReviewed false
IsScholarly false
Issue 9
Keywords 损失函数
注意力机制
small object detection
loss function
上下文信息
attention mechanism
context information
小目标检测
Language Chinese
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-s1060-b36035f8e743865169233aa11bc6c2c1ef9cf2d32772b83b94bce2ea6709cdd63
PageCount 7
ParticipantIDs wanfang_journals_jsjgcyyy202409030
PublicationCentury 2000
PublicationDate 2024-05-01
PublicationDateYYYYMMDD 2024-05-01
PublicationDate_xml – month: 05
  year: 2024
  text: 2024-05-01
  day: 01
PublicationDecade 2020
PublicationTitle 计算机工程与应用
PublicationTitle_FL Computer Engineering and Applications
PublicationYear 2024
Publisher 福建理工大学 计算机科学与数学学院,福州 350118
Publisher_xml – name: 福建理工大学 计算机科学与数学学院,福州 350118
SSID ssib051375739
ssib001102935
ssj0000561668
ssib023646291
ssib057620132
Score 1.9751627
Snippet TP391.4;...
SourceID wanfang
SourceType Aggregation Database
StartPage 292
Title 基于结构感知和全局上下文信息的小目标检测
URI https://d.wanfangdata.com.cn/periodical/jsjgcyyy202409030
Volume 60
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV1JaxRBFG6yXPQgrrgTxDqFid1V3bWAl-pJhyDGiwnkFnqN5DCCkxySU4RIwIiCggcFo4jgyeUghHjwJ_gnZnr-hu9VV2Z63IheikfVq6_e0tX1qrsWx7nKFJMyLmBakvG04QdF1kgy6jcyxlXux9xVGW5wnrvFZxf8G4vB4sjo99qqpbXVZCrd-O2-kv_xKuSBX3GX7D94tg8KGUCDfyEFD0N6KB-TKCBqhoSaRD6mMiKRIGFIFCMRJyoi0kcCUmCDIkh1YGpRIptIyIBoiUTo4aIHxJFE6gMiNDiAIEwOVPcMoEd0BahNE1DdJbLKCYmOkEe7phYQzCBzEkJzYT0aJpFEZsQUSCjDr5pGI8AURlqBEqIklUiRKdJE-aY5H0oPnhpTe9qwKFQRlYY2MWeIpWoJSiSyV4pJPcTSRINUmgJQZV5dQwmMmbnRYMbCgQZ2WLefUeBp7C9aNA--UYYbSxkA1BN0AEvxmsKABFpVkNPAP_lXOwlkVV6fu2Yn4DGeGRT1CbCPIppCDxuWCaCiSfwNPDRemQHtYNObHdCqCxpsx1X10am6dtAGOrS6_vvnMZQJIc0Yig1M9RvAHQMU__gHg8ihv55zpb2ynK6vr6Nh8cOfO-qMUyFw3cS4np67eXsQoUNAqwYROl5fwOnguKbAYyIQg4NqYTZMXXu4qD2un3vcbnC1klVHCaPY1_4stNmT1yri1nItfJw_7hyz874JXXXiE87Ixp2TztHaaaCnnOvd3f3O_uPe12flq61ya7e3-6779FH3wfvu583O3sPO3k75fLvz7U15_2PvxVb305Peyw_l6-3y7Wb5Zee0szATzTdnG_Zyk0bbc7nbSBh3WVDIHEJ4yfFvNWUsjj0vSXlKUy8vVFrQjIEZaSJZovwkzWke43mLaZZxdsYZa91t5WediTQuBKMJz3yY3GQwAxGZxxOpck_6hZuKc84Vq_uSfXm1l37x2PnDMF1wjgz6zkVnbPXeWn4JgvLV5LJ19A-xn6dE
linkProvider EBSCOhost
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%3Ajournal&rft.genre=article&rft.atitle=%E5%9F%BA%E4%BA%8E%E7%BB%93%E6%9E%84%E6%84%9F%E7%9F%A5%E5%92%8C%E5%85%A8%E5%B1%80%E4%B8%8A%E4%B8%8B%E6%96%87%E4%BF%A1%E6%81%AF%E7%9A%84%E5%B0%8F%E7%9B%AE%E6%A0%87%E6%A3%80%E6%B5%8B&rft.jtitle=%E8%AE%A1%E7%AE%97%E6%9C%BA%E5%B7%A5%E7%A8%8B%E4%B8%8E%E5%BA%94%E7%94%A8&rft.au=%E6%9D%8E%E9%92%9F%E5%8D%8E&rft.au=%E6%9E%97%E5%88%9D%E4%BF%8A&rft.au=%E6%9C%B1%E6%81%92%E4%BA%AE&rft.au=%E5%BB%96%E8%AF%97%E5%AE%87&rft.date=2024-05-01&rft.pub=%E7%A6%8F%E5%BB%BA%E7%90%86%E5%B7%A5%E5%A4%A7%E5%AD%A6+%E8%AE%A1%E7%AE%97%E6%9C%BA%E7%A7%91%E5%AD%A6%E4%B8%8E%E6%95%B0%E5%AD%A6%E5%AD%A6%E9%99%A2%2C%E7%A6%8F%E5%B7%9E+350118&rft.issn=1002-8331&rft.volume=60&rft.issue=9&rft.spage=292&rft.epage=298&rft_id=info:doi/10.3778%2Fj.issn.1002-8331.2302-0275&rft.externalDocID=jsjgcyyy202409030
thumbnail_s http://utb.summon.serialssolutions.com/2.0.0/image/custom?url=http%3A%2F%2Fwww.wanfangdata.com.cn%2Fimages%2FPeriodicalImages%2Fjsjgcyyy%2Fjsjgcyyy.jpg