Graph-FCN for Image Semantic Segmentation

Yi Lu; Yaran Chen; Dongbin Zhao; Jianxin Chen

doi:10.1007/978-3-030-22796-8_11

Graph-FCN for Image Semantic Segmentation

Yi Lu
, Yaran Chen^*
, Dongbin Zhao
, Jianxin Chen

^*Corresponding author for this work

CAS - Institute of Automation
University of Chinese Academy of Sciences
Beijing University of Chinese Medicine

Research output: Chapter in Book or Report/Conference proceeding › Conference Proceeding › peer-review

169 Citations (Scopus)

Abstract

Semantic segmentation with deep learning has achieved great progress in classifying the pixels in the image. However, the local location information is usually ignored in the high-level feature extraction by the deep learning, which is important for image semantic segmentation. To avoid this problem, we propose a graph model initialized by a fully convolutional network (FCN) named Graph-FCN for image semantic segmentation. Firstly, the image grid data is extended to graph structure data by a convolutional network, which transforms the semantic segmentation problem into a graph node classification problem. Then we apply graph convolutional network to solve this graph node classification problem. As far as we know, it is the first time that we apply the graph convolutional network in image semantic segmentation. Our method achieves competitive performance in mean intersection over union (mIOU) on the VOC dataset (about 1.34% improvement), compared to the original FCN model.

Original language	English
Title of host publication	Advances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings
Editors	Huchuan Lu, Huajin Tang, Zhanshan Wang
Publisher	Springer Verlag
Pages	97-105
Number of pages	9
ISBN (Print)	9783030227951
DOIs	https://doi.org/10.1007/978-3-030-22796-8_11
Publication status	Published - 2019
Event	16th International Symposium on Neural Networks, ISNN 2019 - Moscow, Russian Federation Duration: 10 Jul 2019 → 12 Jul 2019

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11554 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th International Symposium on Neural Networks, ISNN 2019
Country/Territory	Russian Federation
City	Moscow
Period	10/07/19 → 12/07/19

Keywords

Graph convolutional network
Graph neural network
Semantic segmentation

Access to Document

10.1007/978-3-030-22796-8_11

Cite this

Lu, Y., Chen, Y., Zhao, D., & Chen, J. (2019). Graph-FCN for Image Semantic Segmentation. In H. Lu, H. Tang, & Z. Wang (Eds.), Advances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings (pp. 97-105). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11554 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-22796-8_11

Lu, Yi ; Chen, Yaran ; Zhao, Dongbin et al. / Graph-FCN for Image Semantic Segmentation. Advances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings. editor / Huchuan Lu ; Huajin Tang ; Zhanshan Wang. Springer Verlag, 2019. pp. 97-105 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{794e676b49fb447fab6db331bb2b676e,

title = "Graph-FCN for Image Semantic Segmentation",

abstract = "Semantic segmentation with deep learning has achieved great progress in classifying the pixels in the image. However, the local location information is usually ignored in the high-level feature extraction by the deep learning, which is important for image semantic segmentation. To avoid this problem, we propose a graph model initialized by a fully convolutional network (FCN) named Graph-FCN for image semantic segmentation. Firstly, the image grid data is extended to graph structure data by a convolutional network, which transforms the semantic segmentation problem into a graph node classification problem. Then we apply graph convolutional network to solve this graph node classification problem. As far as we know, it is the first time that we apply the graph convolutional network in image semantic segmentation. Our method achieves competitive performance in mean intersection over union (mIOU) on the VOC dataset (about 1.34\% improvement), compared to the original FCN model.",

keywords = "Graph convolutional network, Graph neural network, Semantic segmentation",

author = "Yi Lu and Yaran Chen and Dongbin Zhao and Jianxin Chen",

note = "Publisher Copyright: {\textcopyright} 2019, Springer Nature Switzerland AG.; 16th International Symposium on Neural Networks, ISNN 2019 ; Conference date: 10-07-2019 Through 12-07-2019",

year = "2019",

doi = "10.1007/978-3-030-22796-8\_11",

language = "English",

isbn = "9783030227951",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "97--105",

editor = "Huchuan Lu and Huajin Tang and Zhanshan Wang",

booktitle = "Advances in Neural Networks {\textendash} ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings",

}

Lu, Y, Chen, Y, Zhao, D & Chen, J 2019, Graph-FCN for Image Semantic Segmentation. in H Lu, H Tang & Z Wang (eds), Advances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11554 LNCS, Springer Verlag, pp. 97-105, 16th International Symposium on Neural Networks, ISNN 2019, Moscow, Russian Federation, 10/07/19. https://doi.org/10.1007/978-3-030-22796-8_11

Graph-FCN for Image Semantic Segmentation. / Lu, Yi; Chen, Yaran; Zhao, Dongbin et al.
Advances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings. ed. / Huchuan Lu; Huajin Tang; Zhanshan Wang. Springer Verlag, 2019. p. 97-105 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11554 LNCS).

Research output: Chapter in Book or Report/Conference proceeding › Conference Proceeding › peer-review

TY - GEN

T1 - Graph-FCN for Image Semantic Segmentation

AU - Lu, Yi

AU - Chen, Yaran

AU - Zhao, Dongbin

AU - Chen, Jianxin

PY - 2019

Y1 - 2019

N2 - Semantic segmentation with deep learning has achieved great progress in classifying the pixels in the image. However, the local location information is usually ignored in the high-level feature extraction by the deep learning, which is important for image semantic segmentation. To avoid this problem, we propose a graph model initialized by a fully convolutional network (FCN) named Graph-FCN for image semantic segmentation. Firstly, the image grid data is extended to graph structure data by a convolutional network, which transforms the semantic segmentation problem into a graph node classification problem. Then we apply graph convolutional network to solve this graph node classification problem. As far as we know, it is the first time that we apply the graph convolutional network in image semantic segmentation. Our method achieves competitive performance in mean intersection over union (mIOU) on the VOC dataset (about 1.34% improvement), compared to the original FCN model.

AB - Semantic segmentation with deep learning has achieved great progress in classifying the pixels in the image. However, the local location information is usually ignored in the high-level feature extraction by the deep learning, which is important for image semantic segmentation. To avoid this problem, we propose a graph model initialized by a fully convolutional network (FCN) named Graph-FCN for image semantic segmentation. Firstly, the image grid data is extended to graph structure data by a convolutional network, which transforms the semantic segmentation problem into a graph node classification problem. Then we apply graph convolutional network to solve this graph node classification problem. As far as we know, it is the first time that we apply the graph convolutional network in image semantic segmentation. Our method achieves competitive performance in mean intersection over union (mIOU) on the VOC dataset (about 1.34% improvement), compared to the original FCN model.

KW - Graph convolutional network

KW - Graph neural network

KW - Semantic segmentation

UR - https://www.scopus.com/pages/publications/85068614822

U2 - 10.1007/978-3-030-22796-8_11

DO - 10.1007/978-3-030-22796-8_11

M3 - Conference Proceeding

AN - SCOPUS:85068614822

SN - 9783030227951

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 97

EP - 105

BT - Advances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings

A2 - Lu, Huchuan

A2 - Tang, Huajin

A2 - Wang, Zhanshan

PB - Springer Verlag

T2 - 16th International Symposium on Neural Networks, ISNN 2019

Y2 - 10 July 2019 through 12 July 2019

ER -

Lu Y, Chen Y, Zhao D, Chen J. Graph-FCN for Image Semantic Segmentation. In Lu H, Tang H, Wang Z, editors, Advances in Neural Networks – ISNN 2019 - 16th International Symposium on Neural Networks, ISNN 2019, Proceedings. Springer Verlag. 2019. p. 97-105. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-22796-8_11

Graph-FCN for Image Semantic Segmentation

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this