基于多阶段特征提取的鱼类识别研究

TP18%TP391.41%S951.2; 鱼类自动识别在海洋生态学、水产养殖等领域应用广泛.受光照变化、目标相似、遮挡及类别分布不均衡等因素影响,鱼类精准自动识别极具挑战性.提出了一种基于多阶段特征提取网络(Multi-stage Feature Extraction Network,MF-Net)模型进行鱼类识别.该模型首先对图片作弱增强预处理,以提高模型的计算效率;然后采用多阶段卷积特征提取策略,提升模型对鱼类细粒度特征的提取能力;最后通过标签平滑损失计算以缓解数据的不平衡性.为验证模型的性能,构建了一个500类、含32 768张图片的鱼类数据集,所建模型在该数据集上的准确率达到86.8...

Full description

Saved in:
Bibliographic Details
Published in南方水产科学 Vol. 20; no. 1; pp. 99 - 109
Main Authors 吕俊霖, 陈作志, 李碧龙, 蔡润基, 高月芳
Format Journal Article
LanguageChinese
Published 中国水产科学研究院南海水产研究所,广东广州 510300%华南农业大学,广东广州 510642 01.02.2024
Subjects
标签平滑
鱼类识别
长尾识别
Long-tailed recognition
特征提取网络模型
Feature extraction Network model
Label smoothing
Fish recognition
Online AccessGet full text
ISSN2095-0780
DOI10.12131/20230197

Cover

Abstract TP18%TP391.41%S951.2; 鱼类自动识别在海洋生态学、水产养殖等领域应用广泛.受光照变化、目标相似、遮挡及类别分布不均衡等因素影响,鱼类精准自动识别极具挑战性.提出了一种基于多阶段特征提取网络(Multi-stage Feature Extraction Network,MF-Net)模型进行鱼类识别.该模型首先对图片作弱增强预处理,以提高模型的计算效率;然后采用多阶段卷积特征提取策略,提升模型对鱼类细粒度特征的提取能力;最后通过标签平滑损失计算以缓解数据的不平衡性.为验证模型的性能,构建了一个500类、含32 768张图片的鱼类数据集,所建模型在该数据集上的准确率达到86.8％,优于现有的主流目标识别方法.利用公开的蝴蝶数据集对该模型进行泛化性能验证,多组消融实验进一步验证了所提算法的有效性.
AbstractList TP18%TP391.41%S951.2; 鱼类自动识别在海洋生态学、水产养殖等领域应用广泛.受光照变化、目标相似、遮挡及类别分布不均衡等因素影响,鱼类精准自动识别极具挑战性.提出了一种基于多阶段特征提取网络(Multi-stage Feature Extraction Network,MF-Net)模型进行鱼类识别.该模型首先对图片作弱增强预处理,以提高模型的计算效率;然后采用多阶段卷积特征提取策略,提升模型对鱼类细粒度特征的提取能力;最后通过标签平滑损失计算以缓解数据的不平衡性.为验证模型的性能,构建了一个500类、含32 768张图片的鱼类数据集,所建模型在该数据集上的准确率达到86.8％,优于现有的主流目标识别方法.利用公开的蝴蝶数据集对该模型进行泛化性能验证,多组消融实验进一步验证了所提算法的有效性.
Abstract_FL Automatic fish recognition is widely used in the fields of marine ecology and aquaculture.Due to factors such as fluctuating illumination,overlapping instances and occlusion,accurate automatic identification of fish is extremely challenging.In order to solve these problems,this paper introduces an innovative Multi-stage Feature Extraction Network(MF-Net)model,which is predicated upon a multi-stage feature extraction paradigm for the domain of automatic fish recognition.The architec-ture of MF-Net commences with a subtle image enhancement preprocessing step,judiciously designed to augment the computa-tional efficiency of the model.Then the deployment of a multi-stage convolutional feature extraction strategy is applied to im-prove the model's sensitivity towards the granular features of fish species.In an effort to mitigate issues arising from data imba-lance,the model incorporates a long-tail loss computation strategy.To evaluate the efficacy of the proposed MF-Net,the study collects a comprehensive fish dataset encompassing 500 categories including 32 768 images.The proposed MF-Net demon-strated a remarkable accuracy of 86.8％on this dataset,thereby outperforming the recognition performance of the existing state-of-the-art target recognition algorithms.Furthermore,the model is tested on a publicly butterfly dataset to verify its generaliza-tion performance,and multiple ablation experiments further validate the effectiveness of the proposed algorithm.
Author 高月芳
李碧龙
吕俊霖
蔡润基
陈作志
AuthorAffiliation 中国水产科学研究院南海水产研究所,广东广州 510300%华南农业大学,广东广州 510642
AuthorAffiliation_xml – name: 中国水产科学研究院南海水产研究所,广东广州 510300%华南农业大学,广东广州 510642
Author_FL LYU Junlin
CHEN Zuozhi
CAI Runji
GAO Yuefang
LI Bilong
Author_FL_xml – sequence: 1
  fullname: LYU Junlin
– sequence: 2
  fullname: CHEN Zuozhi
– sequence: 3
  fullname: LI Bilong
– sequence: 4
  fullname: CAI Runji
– sequence: 5
  fullname: GAO Yuefang
Author_xml – sequence: 1
  fullname: 吕俊霖
– sequence: 2
  fullname: 陈作志
– sequence: 3
  fullname: 李碧龙
– sequence: 4
  fullname: 蔡润基
– sequence: 5
  fullname: 高月芳
BookMark eNotjz1LQzEYhTNUsNYO_gjHq--bxORmlOIXFFzqXG5uElEkBYM4CyK6WAcR6aKDSxGUDhas4q8xjf0XXtDpcODhPJwFUvM9bwlZQlhBigxXKVAGqGSN1CmotQxkDvOkGcKBBpRcSiVUnUB8mHxPruPTYHY_nr68pav3-HU27d_E_l0anM9Gn2n08fN6ES-f0-NtGo4XyZwrjoJt_meD7G1udFrbWXt3a6e13s4CAmCmuOZWSJdjWXIqeE4Zk0KUjBstkeVoqLaWU1MBmipdFdCMOiMLI7jTrEGW_3ZPC-8Kv9897J0c-8rY9S6U1TkOCIjsFwPHVcI
ClassificationCodes TP18%TP391.41%S951.2
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 10.12131/20230197
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 Agriculture
DocumentTitle_FL Research on fish recognition based on multi-stage feature extraction learning
EndPage 109
ExternalDocumentID nfsc202401011
GrantInformation_xml – fundername: (农业农村部财政专项); (广东省重点领域研发计划项目)
  funderid: (农业农村部财政专项); (广东省重点领域研发计划项目)
GroupedDBID -04
2B.
4A8
5VS
92G
92I
93N
ALMA_UNASSIGNED_HOLDINGS
CCEZO
CHDYS
ECGQY
KQ8
PSX
TCJ
TGD
U1G
U5N
UY8
ID FETCH-LOGICAL-s1001-94b4e67f81cc42648233766c34db71381d2bee42d81cb29bbee0b32fd7ad64fb3
ISSN 2095-0780
IngestDate Thu May 29 04:08:32 EDT 2025
IsPeerReviewed true
IsScholarly true
Issue 1
Keywords 标签平滑
鱼类识别
长尾识别
Long-tailed recognition
特征提取网络模型
Feature extraction Network model
Label smoothing
Fish recognition
Language Chinese
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-s1001-94b4e67f81cc42648233766c34db71381d2bee42d81cb29bbee0b32fd7ad64fb3
PageCount 11
ParticipantIDs wanfang_journals_nfsc202401011
PublicationCentury 2000
PublicationDate 2024-02-01
PublicationDateYYYYMMDD 2024-02-01
PublicationDate_xml – month: 02
  year: 2024
  text: 2024-02-01
  day: 01
PublicationDecade 2020
PublicationTitle 南方水产科学
PublicationTitle_FL South China Fisheries Science
PublicationYear 2024
Publisher 中国水产科学研究院南海水产研究所,广东广州 510300%华南农业大学,广东广州 510642
Publisher_xml – name: 中国水产科学研究院南海水产研究所,广东广州 510300%华南农业大学,广东广州 510642
SSID ssib017477969
ssib051370567
ssib007891391
ssib038075195
ssib036438877
ssj0001560132
Score 2.3433495
Snippet TP18%TP391.41%S951.2; 鱼类自动识别在海洋生态学、水产养殖等领域应用广泛.受光照变化、目标相似、遮挡及类别分布不均衡等因素影响,鱼类精准自动识别极具挑战性.提出了一种基于多阶段特征提取网络(Multi-stage Feature Extraction...
SourceID wanfang
SourceType Aggregation Database
StartPage 99
Title 基于多阶段特征提取的鱼类识别研究
URI https://d.wanfangdata.com.cn/periodical/nfsc202401011
Volume 20
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwnV29b9QwFLfassCA-BSfVQc8oYPEdvwxOr2cKhBISK3UrUouSZmuUj-WbkgIwUIZEEJdYGBBSKAOVKIg_hquR_8GFt5zfJeUA1FYLOfl2e89_5L42bGfCbkm8kwyDDnLi4K1RC5UK2WqbPFUZFrz1OQupNCdu3JuQdxajBYnJn80Vi1trGc3upu_3VfyP6gCDXDFXbL_gOyoUiBAHvCFFBCG9EgY0ySipkNjSxOBqU6QYgU1QDHUaBpLmkhqExpHNFFUGxob5IkTcCHxlu5QEyAFMxJ5oKwWWDwOaTyLFMzENNHUdqiWjhnyMd6yATXCZYBfNj1dx9amRqEUqBnlQhrQWHhtrXIFQWLo1G5TO5omdJZB3ZHjBbGVQbPUNFiAYFAVZGnjTTQMzFA1C4huu2YBQcxJNGi7MTWLRgusawxoLStHjdqcEGFiuIbaP8JOqEalkT1GBY5g3niTeStAuWZ7AVqx-kOFvxSXCCou4KzMN9gClW6mSXEVmuQ6hjbEg7kjL696ZLzgCOHFYlV5658mq2oDji5GisaMMgtwi76qjtMadoksGHv1q_7Nw1N5SqELbDHeCbOQu41GOLwNPeaHY5r3wL9D6DDAYThJjjGl3BqL2_fqzkDhD_Q6kBOMnJUycuR7c3CloYMc-c54cEJ11mR1HYVcgXOv6glXnHtwBxaOTPaRxlDfm0Nt3ea9Xpn2lht-5vwpctIPEGds9bafJhOb98-QE3Z51QfJKc6SoP9q79ve0_6b7YOXu_vvPw6efOp_fbC_9ay_9WKw_fBg58tg5_P3D4_6j98NXj8fvN09RxY6yfzsXMuffNJaw5hoLSMyUUhV6rDbxSGLZhwcAdnl8GlVITjZOcuKQrAcGDJmMrgIMs7KXKW5FGXGz5Op3kqvuEBmcvDfc15GWRiUQpoiTYtC80yaKNdBwflFMu3tXfJftrWlQ_Bc-hvDZXK8fg2vkKn11Y3iKnjq69m0Q_QnbP2hcQ
linkProvider Colorado Alliance of Research Libraries
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%E5%A4%9A%E9%98%B6%E6%AE%B5%E7%89%B9%E5%BE%81%E6%8F%90%E5%8F%96%E7%9A%84%E9%B1%BC%E7%B1%BB%E8%AF%86%E5%88%AB%E7%A0%94%E7%A9%B6&rft.jtitle=%E5%8D%97%E6%96%B9%E6%B0%B4%E4%BA%A7%E7%A7%91%E5%AD%A6&rft.au=%E5%90%95%E4%BF%8A%E9%9C%96&rft.au=%E9%99%88%E4%BD%9C%E5%BF%97&rft.au=%E6%9D%8E%E7%A2%A7%E9%BE%99&rft.au=%E8%94%A1%E6%B6%A6%E5%9F%BA&rft.date=2024-02-01&rft.pub=%E4%B8%AD%E5%9B%BD%E6%B0%B4%E4%BA%A7%E7%A7%91%E5%AD%A6%E7%A0%94%E7%A9%B6%E9%99%A2%E5%8D%97%E6%B5%B7%E6%B0%B4%E4%BA%A7%E7%A0%94%E7%A9%B6%E6%89%80%2C%E5%B9%BF%E4%B8%9C%E5%B9%BF%E5%B7%9E+510300%25%E5%8D%8E%E5%8D%97%E5%86%9C%E4%B8%9A%E5%A4%A7%E5%AD%A6%2C%E5%B9%BF%E4%B8%9C%E5%B9%BF%E5%B7%9E+510642&rft.issn=2095-0780&rft.volume=20&rft.issue=1&rft.spage=99&rft.epage=109&rft_id=info:doi/10.12131%2F20230197&rft.externalDocID=nfsc202401011
thumbnail_s http://utb.summon.serialssolutions.com/2.0.0/image/custom?url=http%3A%2F%2Fwww.wanfangdata.com.cn%2Fimages%2FPeriodicalImages%2Fnfsc%2Fnfsc.jpg