FDCNet: filtering deep convolutional network for marine organism classification

Huimin Lu, Yujie Li, Tomoki Uemura, Zongyuan Ge, Xing Xu, Li He, Seiichi Serikawa, Hyoungseop Kim

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Convolutional networks are currently the most popular computer vision methods for a wide variety of applications in multimedia research fields. Most recent methods have focused on solving problems with natural images and usually use a training database, such as Imagenet or Openimage, to detect the characteristics of the objects. However, in practical applications, training samples are difficult to acquire. In this study, we develop a powerful approach that can accurately learn marine organisms. The proposed filtering deep convolutional network (FDCNet) classifies deep-sea objects better than state-of-the-art classification methods, such as AlexNet, GoogLeNet, ResNet50, and ResNet101. The classification accuracy of the proposed FDCNet method is 1.8%, 2.9%, 2.0%, and 1.0% better than AlexNet, GooLeNet, ResNet50, and ResNet101, respectively. In addition, we have built the first marine organism database, Kyutech10K, with seven categories (i.e., shrimp, squid, crab, shark, sea urchin, manganese, and sand).

Original languageEnglish
Pages (from-to)21847-21860
Number of pages14
JournalMultimedia Tools and Applications
Volume77
Issue number17
DOIs
Publication statusPublished - 1 Sep 2018
Externally publishedYes

Keywords

  • Artificial intelligence
  • Deep learning
  • Filtering deep convolutional network
  • Marine organism classification

Cite this

Lu, Huimin ; Li, Yujie ; Uemura, Tomoki ; Ge, Zongyuan ; Xu, Xing ; He, Li ; Serikawa, Seiichi ; Kim, Hyoungseop. / FDCNet : filtering deep convolutional network for marine organism classification. In: Multimedia Tools and Applications. 2018 ; Vol. 77, No. 17. pp. 21847-21860.
@article{a705d76f715f44e38e9d2857b6794bc0,
title = "FDCNet: filtering deep convolutional network for marine organism classification",
abstract = "Convolutional networks are currently the most popular computer vision methods for a wide variety of applications in multimedia research fields. Most recent methods have focused on solving problems with natural images and usually use a training database, such as Imagenet or Openimage, to detect the characteristics of the objects. However, in practical applications, training samples are difficult to acquire. In this study, we develop a powerful approach that can accurately learn marine organisms. The proposed filtering deep convolutional network (FDCNet) classifies deep-sea objects better than state-of-the-art classification methods, such as AlexNet, GoogLeNet, ResNet50, and ResNet101. The classification accuracy of the proposed FDCNet method is 1.8{\%}, 2.9{\%}, 2.0{\%}, and 1.0{\%} better than AlexNet, GooLeNet, ResNet50, and ResNet101, respectively. In addition, we have built the first marine organism database, Kyutech10K, with seven categories (i.e., shrimp, squid, crab, shark, sea urchin, manganese, and sand).",
keywords = "Artificial intelligence, Deep learning, Filtering deep convolutional network, Marine organism classification",
author = "Huimin Lu and Yujie Li and Tomoki Uemura and Zongyuan Ge and Xing Xu and Li He and Seiichi Serikawa and Hyoungseop Kim",
year = "2018",
month = "9",
day = "1",
doi = "10.1007/s11042-017-4585-1",
language = "English",
volume = "77",
pages = "21847--21860",
journal = "Multimedia Tools and Applications",
issn = "1380-7501",
publisher = "Springer-Verlag London Ltd.",
number = "17",

}

FDCNet : filtering deep convolutional network for marine organism classification. / Lu, Huimin; Li, Yujie; Uemura, Tomoki; Ge, Zongyuan; Xu, Xing; He, Li; Serikawa, Seiichi; Kim, Hyoungseop.

In: Multimedia Tools and Applications, Vol. 77, No. 17, 01.09.2018, p. 21847-21860.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - FDCNet

T2 - filtering deep convolutional network for marine organism classification

AU - Lu, Huimin

AU - Li, Yujie

AU - Uemura, Tomoki

AU - Ge, Zongyuan

AU - Xu, Xing

AU - He, Li

AU - Serikawa, Seiichi

AU - Kim, Hyoungseop

PY - 2018/9/1

Y1 - 2018/9/1

N2 - Convolutional networks are currently the most popular computer vision methods for a wide variety of applications in multimedia research fields. Most recent methods have focused on solving problems with natural images and usually use a training database, such as Imagenet or Openimage, to detect the characteristics of the objects. However, in practical applications, training samples are difficult to acquire. In this study, we develop a powerful approach that can accurately learn marine organisms. The proposed filtering deep convolutional network (FDCNet) classifies deep-sea objects better than state-of-the-art classification methods, such as AlexNet, GoogLeNet, ResNet50, and ResNet101. The classification accuracy of the proposed FDCNet method is 1.8%, 2.9%, 2.0%, and 1.0% better than AlexNet, GooLeNet, ResNet50, and ResNet101, respectively. In addition, we have built the first marine organism database, Kyutech10K, with seven categories (i.e., shrimp, squid, crab, shark, sea urchin, manganese, and sand).

AB - Convolutional networks are currently the most popular computer vision methods for a wide variety of applications in multimedia research fields. Most recent methods have focused on solving problems with natural images and usually use a training database, such as Imagenet or Openimage, to detect the characteristics of the objects. However, in practical applications, training samples are difficult to acquire. In this study, we develop a powerful approach that can accurately learn marine organisms. The proposed filtering deep convolutional network (FDCNet) classifies deep-sea objects better than state-of-the-art classification methods, such as AlexNet, GoogLeNet, ResNet50, and ResNet101. The classification accuracy of the proposed FDCNet method is 1.8%, 2.9%, 2.0%, and 1.0% better than AlexNet, GooLeNet, ResNet50, and ResNet101, respectively. In addition, we have built the first marine organism database, Kyutech10K, with seven categories (i.e., shrimp, squid, crab, shark, sea urchin, manganese, and sand).

KW - Artificial intelligence

KW - Deep learning

KW - Filtering deep convolutional network

KW - Marine organism classification

UR - http://www.scopus.com/inward/record.url?scp=85015699675&partnerID=8YFLogxK

U2 - 10.1007/s11042-017-4585-1

DO - 10.1007/s11042-017-4585-1

M3 - Article

VL - 77

SP - 21847

EP - 21860

JO - Multimedia Tools and Applications

JF - Multimedia Tools and Applications

SN - 1380-7501

IS - 17

ER -