Transfer Learning using Transformation: Is Large Unlabeled Data Helpful at Segmentation?

Heejeong Lim; Seongwook Yoon; Sanghoon Sull

doi:10.1109/ICTC49870.2020.9289267

Transfer Learning using Transformation: Is Large Unlabeled Data Helpful at Segmentation?

Heejeong Lim, Seongwook Yoon, Sanghoon Sull

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We propose a simple method of transfer learning for image segmentation. Creating labeled data for deep neural network training in image segmentation is particularly expensive than other tasks. Hence, practically, the labeled data is much less than the unlabeled data. So, we introduce a method that is helpful for segmentation by using unlabeled data. Our key is the RGB-to-HSV transformation and we use it in two ways. The first way is to pre-train a network to work as an RGB-to-HSV transformer which can extract useful features, and transfer the pre-trained weights to another network for segmentation, which is one of the most common transfer learning method. The second way is to provide additional information to the segmented network by providing HSV, the output of the pre-trained network, as additional input. We performed several experiments about our proposal using Cityscapes dataset.

Original language	English
Title of host publication	ICTC 2020 - 11th International Conference on ICT Convergence
Subtitle of host publication	Data, Network, and AI in the Age of Untact
Publisher	IEEE Computer Society
Pages	387-390
Number of pages	4
ISBN (Electronic)	9781728167589
DOIs	https://doi.org/10.1109/ICTC49870.2020.9289267
Publication status	Published - 2020 Oct 21
Event	11th International Conference on Information and Communication Technology Convergence, ICTC 2020 - Jeju Island, Korea, Republic of Duration: 2020 Oct 21 → 2020 Oct 23

Publication series

Name	International Conference on ICT Convergence
Volume	2020-October
ISSN (Print)	2162-1233
ISSN (Electronic)	2162-1241

Conference

Conference	11th International Conference on Information and Communication Technology Convergence, ICTC 2020
Country/Territory	Korea, Republic of
City	Jeju Island
Period	20/10/21 → 20/10/23

Bibliographical note

Funding Information:
”This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2020-2016-0-00464) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation)“

Publisher Copyright:
© 2020 IEEE.

Keywords

HSV transformation
Image segmentation
Transfer learning

ASJC Scopus subject areas

Information Systems
Computer Networks and Communications

Access to Document

10.1109/ICTC49870.2020.9289267

Cite this

Lim, H., Yoon, S., & Sull, S. (2020). Transfer Learning using Transformation: Is Large Unlabeled Data Helpful at Segmentation? In ICTC 2020 - 11th International Conference on ICT Convergence: Data, Network, and AI in the Age of Untact (pp. 387-390). Article 9289267 (International Conference on ICT Convergence; Vol. 2020-October). IEEE Computer Society. https://doi.org/10.1109/ICTC49870.2020.9289267

Transfer Learning using Transformation: Is Large Unlabeled Data Helpful at Segmentation? / Lim, Heejeong; Yoon, Seongwook; Sull, Sanghoon.
ICTC 2020 - 11th International Conference on ICT Convergence: Data, Network, and AI in the Age of Untact. IEEE Computer Society, 2020. p. 387-390 9289267 (International Conference on ICT Convergence; Vol. 2020-October).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Lim, H, Yoon, S & Sull, S 2020, Transfer Learning using Transformation: Is Large Unlabeled Data Helpful at Segmentation? in ICTC 2020 - 11th International Conference on ICT Convergence: Data, Network, and AI in the Age of Untact., 9289267, International Conference on ICT Convergence, vol. 2020-October, IEEE Computer Society, pp. 387-390, 11th International Conference on Information and Communication Technology Convergence, ICTC 2020, Jeju Island, Korea, Republic of, 20/10/21. https://doi.org/10.1109/ICTC49870.2020.9289267

@inproceedings{84281c14d1a54d62bde81de575684831,

title = "Transfer Learning using Transformation: Is Large Unlabeled Data Helpful at Segmentation?",

abstract = "We propose a simple method of transfer learning for image segmentation. Creating labeled data for deep neural network training in image segmentation is particularly expensive than other tasks. Hence, practically, the labeled data is much less than the unlabeled data. So, we introduce a method that is helpful for segmentation by using unlabeled data. Our key is the RGB-to-HSV transformation and we use it in two ways. The first way is to pre-train a network to work as an RGB-to-HSV transformer which can extract useful features, and transfer the pre-trained weights to another network for segmentation, which is one of the most common transfer learning method. The second way is to provide additional information to the segmented network by providing HSV, the output of the pre-trained network, as additional input. We performed several experiments about our proposal using Cityscapes dataset. ",

keywords = "HSV transformation, Image segmentation, Transfer learning",

author = "Heejeong Lim and Seongwook Yoon and Sanghoon Sull",

note = "Funding Information: ”This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2020-2016-0-00464) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation)“ Publisher Copyright: {\textcopyright} 2020 IEEE.; 11th International Conference on Information and Communication Technology Convergence, ICTC 2020 ; Conference date: 21-10-2020 Through 23-10-2020",

year = "2020",

month = oct,

day = "21",

doi = "10.1109/ICTC49870.2020.9289267",

language = "English",

series = "International Conference on ICT Convergence",

publisher = "IEEE Computer Society",

pages = "387--390",

booktitle = "ICTC 2020 - 11th International Conference on ICT Convergence",

}

TY - GEN

T1 - Transfer Learning using Transformation

T2 - 11th International Conference on Information and Communication Technology Convergence, ICTC 2020

AU - Lim, Heejeong

AU - Yoon, Seongwook

AU - Sull, Sanghoon

N1 - Funding Information: ”This research was supported by the MSIT(Ministry of Science and ICT), Korea, under the ITRC(Information Technology Research Center) support program(IITP-2020-2016-0-00464) supervised by the IITP(Institute for Information & communications Technology Planning & Evaluation)“ Publisher Copyright: © 2020 IEEE.

PY - 2020/10/21

Y1 - 2020/10/21

N2 - We propose a simple method of transfer learning for image segmentation. Creating labeled data for deep neural network training in image segmentation is particularly expensive than other tasks. Hence, practically, the labeled data is much less than the unlabeled data. So, we introduce a method that is helpful for segmentation by using unlabeled data. Our key is the RGB-to-HSV transformation and we use it in two ways. The first way is to pre-train a network to work as an RGB-to-HSV transformer which can extract useful features, and transfer the pre-trained weights to another network for segmentation, which is one of the most common transfer learning method. The second way is to provide additional information to the segmented network by providing HSV, the output of the pre-trained network, as additional input. We performed several experiments about our proposal using Cityscapes dataset.

AB - We propose a simple method of transfer learning for image segmentation. Creating labeled data for deep neural network training in image segmentation is particularly expensive than other tasks. Hence, practically, the labeled data is much less than the unlabeled data. So, we introduce a method that is helpful for segmentation by using unlabeled data. Our key is the RGB-to-HSV transformation and we use it in two ways. The first way is to pre-train a network to work as an RGB-to-HSV transformer which can extract useful features, and transfer the pre-trained weights to another network for segmentation, which is one of the most common transfer learning method. The second way is to provide additional information to the segmented network by providing HSV, the output of the pre-trained network, as additional input. We performed several experiments about our proposal using Cityscapes dataset.

KW - HSV transformation

KW - Image segmentation

KW - Transfer learning

UR - http://www.scopus.com/inward/record.url?scp=85098958122&partnerID=8YFLogxK

U2 - 10.1109/ICTC49870.2020.9289267

DO - 10.1109/ICTC49870.2020.9289267

M3 - Conference contribution

AN - SCOPUS:85098958122

T3 - International Conference on ICT Convergence

SP - 387

EP - 390

BT - ICTC 2020 - 11th International Conference on ICT Convergence

PB - IEEE Computer Society

Y2 - 21 October 2020 through 23 October 2020

ER -

Transfer Learning using Transformation: Is Large Unlabeled Data Helpful at Segmentation?

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this