Two-step transfer learning for semantic plant segmentation

Shunsuke Sakurai, Hideaki Uchiyama, Atsushi Shimada, Daisaku Arita, Rin-Ichiro Taniguchi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

We discuss the applicability of a fully convolutional network (FCN), which provides promising performance in semantic segmentation tasks, to plant segmentation tasks. The challenge lies in training the network with a small dataset because there are not many samples in plant image datasets, as compared to object image datasets such as ImageNet and PASCAL VOC datasets. The proposed method is inspired by transfer learning, but involves a two-step adaptation. In the first step, we apply transfer learning from a source domain that contains many objects with a large amount of labeled data to a major category in the plant domain. Then, in the second step, category adaptation is performed from the major category to a minor category with a few samples within the plant domain. With leaf segmentation challenge (LSC) dataset, the experimental results confirm the effectiveness of the proposed method such that F-measure criterion was, for instance, 0.953 for the A2 dataset, which was 0.355 higher than that of direct adaptation, and 0.527 higher than that of non-adaptation.

Original languageEnglish
Title of host publicationICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods
EditorsAna Fred, Gabriella Sanniti di Baja, Maria De Marsico
PublisherSciTePress
Pages332-339
Number of pages8
ISBN (Electronic)9789897582769
Publication statusPublished - Jan 1 2018
Event7th International Conference on Pattern Recognition Applications and Methods, ICPRAM 2018 - Funchal, Madeira, Portugal
Duration: Jan 16 2018Jan 18 2018

Publication series

NameICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods
Volume2018-January

Other

Other7th International Conference on Pattern Recognition Applications and Methods, ICPRAM 2018
CountryPortugal
CityFunchal, Madeira
Period1/16/181/18/18

Fingerprint

Volatile organic compounds
Semantics

All Science Journal Classification (ASJC) codes

  • Computer Vision and Pattern Recognition

Cite this

Sakurai, S., Uchiyama, H., Shimada, A., Arita, D., & Taniguchi, R-I. (2018). Two-step transfer learning for semantic plant segmentation. In A. Fred, G. S. di Baja, & M. De Marsico (Eds.), ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods (pp. 332-339). (ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods; Vol. 2018-January). SciTePress.

Two-step transfer learning for semantic plant segmentation. / Sakurai, Shunsuke; Uchiyama, Hideaki; Shimada, Atsushi; Arita, Daisaku; Taniguchi, Rin-Ichiro.

ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods. ed. / Ana Fred; Gabriella Sanniti di Baja; Maria De Marsico. SciTePress, 2018. p. 332-339 (ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods; Vol. 2018-January).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sakurai, S, Uchiyama, H, Shimada, A, Arita, D & Taniguchi, R-I 2018, Two-step transfer learning for semantic plant segmentation. in A Fred, GS di Baja & M De Marsico (eds), ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods. ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods, vol. 2018-January, SciTePress, pp. 332-339, 7th International Conference on Pattern Recognition Applications and Methods, ICPRAM 2018, Funchal, Madeira, Portugal, 1/16/18.
Sakurai S, Uchiyama H, Shimada A, Arita D, Taniguchi R-I. Two-step transfer learning for semantic plant segmentation. In Fred A, di Baja GS, De Marsico M, editors, ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods. SciTePress. 2018. p. 332-339. (ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods).
Sakurai, Shunsuke ; Uchiyama, Hideaki ; Shimada, Atsushi ; Arita, Daisaku ; Taniguchi, Rin-Ichiro. / Two-step transfer learning for semantic plant segmentation. ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods. editor / Ana Fred ; Gabriella Sanniti di Baja ; Maria De Marsico. SciTePress, 2018. pp. 332-339 (ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods).
@inproceedings{47027446b45d42379fb253c6e03bd20c,
title = "Two-step transfer learning for semantic plant segmentation",
abstract = "We discuss the applicability of a fully convolutional network (FCN), which provides promising performance in semantic segmentation tasks, to plant segmentation tasks. The challenge lies in training the network with a small dataset because there are not many samples in plant image datasets, as compared to object image datasets such as ImageNet and PASCAL VOC datasets. The proposed method is inspired by transfer learning, but involves a two-step adaptation. In the first step, we apply transfer learning from a source domain that contains many objects with a large amount of labeled data to a major category in the plant domain. Then, in the second step, category adaptation is performed from the major category to a minor category with a few samples within the plant domain. With leaf segmentation challenge (LSC) dataset, the experimental results confirm the effectiveness of the proposed method such that F-measure criterion was, for instance, 0.953 for the A2 dataset, which was 0.355 higher than that of direct adaptation, and 0.527 higher than that of non-adaptation.",
author = "Shunsuke Sakurai and Hideaki Uchiyama and Atsushi Shimada and Daisaku Arita and Rin-Ichiro Taniguchi",
year = "2018",
month = "1",
day = "1",
language = "English",
series = "ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods",
publisher = "SciTePress",
pages = "332--339",
editor = "Ana Fred and {di Baja}, {Gabriella Sanniti} and {De Marsico}, Maria",
booktitle = "ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods",

}

TY - GEN

T1 - Two-step transfer learning for semantic plant segmentation

AU - Sakurai, Shunsuke

AU - Uchiyama, Hideaki

AU - Shimada, Atsushi

AU - Arita, Daisaku

AU - Taniguchi, Rin-Ichiro

PY - 2018/1/1

Y1 - 2018/1/1

N2 - We discuss the applicability of a fully convolutional network (FCN), which provides promising performance in semantic segmentation tasks, to plant segmentation tasks. The challenge lies in training the network with a small dataset because there are not many samples in plant image datasets, as compared to object image datasets such as ImageNet and PASCAL VOC datasets. The proposed method is inspired by transfer learning, but involves a two-step adaptation. In the first step, we apply transfer learning from a source domain that contains many objects with a large amount of labeled data to a major category in the plant domain. Then, in the second step, category adaptation is performed from the major category to a minor category with a few samples within the plant domain. With leaf segmentation challenge (LSC) dataset, the experimental results confirm the effectiveness of the proposed method such that F-measure criterion was, for instance, 0.953 for the A2 dataset, which was 0.355 higher than that of direct adaptation, and 0.527 higher than that of non-adaptation.

AB - We discuss the applicability of a fully convolutional network (FCN), which provides promising performance in semantic segmentation tasks, to plant segmentation tasks. The challenge lies in training the network with a small dataset because there are not many samples in plant image datasets, as compared to object image datasets such as ImageNet and PASCAL VOC datasets. The proposed method is inspired by transfer learning, but involves a two-step adaptation. In the first step, we apply transfer learning from a source domain that contains many objects with a large amount of labeled data to a major category in the plant domain. Then, in the second step, category adaptation is performed from the major category to a minor category with a few samples within the plant domain. With leaf segmentation challenge (LSC) dataset, the experimental results confirm the effectiveness of the proposed method such that F-measure criterion was, for instance, 0.953 for the A2 dataset, which was 0.355 higher than that of direct adaptation, and 0.527 higher than that of non-adaptation.

UR - http://www.scopus.com/inward/record.url?scp=85052017436&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85052017436&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85052017436

T3 - ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods

SP - 332

EP - 339

BT - ICPRAM 2018 - Proceedings of the 7th International Conference on Pattern Recognition Applications and Methods

A2 - Fred, Ana

A2 - di Baja, Gabriella Sanniti

A2 - De Marsico, Maria

PB - SciTePress

ER -