Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging

Jia Hao Syu; Shih Hsuan Cho; Sheng-Jyh Wang; Li-Chun Wang

doi:10.1007/978-3-319-94211-7_28

Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging

Jia Hao Syu^*, Shih Hsuan Cho, Sheng-Jyh Wang, Li-Chun Wang

^*Corresponding author for this work

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

Abstract

In this paper, we propose an iterative contraction and merging framework (ICM) for semantic segmentation in indoor scenes. Given an input image and a raw depth image, we first derive the dense prediction map from a convolutional neural network (CNN) and a normal vector map from the depth image. By combining the RGB-D image with these two maps, we can guide the ICM process to produce a more accurate hierarchical segmentation tree in a bottom-up manner. After that, based on the hierarchical segmentation tree, we design a decision process which uses the dense prediction map as a reference to make the final decision of semantic segmentation. Experimental results show that the proposed method can generate much more accurate object boundaries if compared to the state-of-the-art methods.

Original language	English
Title of host publication	Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings
Editors	Driss Mammass, Fathallah Nouboud, Alamin Mansouri, Abderrahim El Moataz
Publisher	Springer Verlag
Pages	252-261
Number of pages	10
ISBN (Print)	9783319942100
DOIs	https://doi.org/10.1007/978-3-319-94211-7_28
State	Published - 1 Jan 2018
Event	8th International Conference on Image and Signal Processing, ICISP 2018 - Cherbourg, France Duration: 2 Jul 2018 → 4 Jul 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10884 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	8th International Conference on Image and Signal Processing, ICISP 2018
Country/Territory	France
City	Cherbourg
Period	2/07/18 → 4/07/18

Keywords

Convolutional neural network
Iterative contraction and merging
RGB-D image
Semantic segmentation

Access to Document

10.1007/978-3-319-94211-7_28

Cite this

Syu, J. H., Cho, S. H., Wang, S-J., & Wang, L-C. (2018). Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging. In D. Mammass, F. Nouboud, A. Mansouri, & A. El Moataz (Eds.), Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings (pp. 252-261). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10884 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-94211-7_28

Syu, Jia Hao ; Cho, Shih Hsuan ; Wang, Sheng-Jyh et al. / Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging. Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings. editor / Driss Mammass ; Fathallah Nouboud ; Alamin Mansouri ; Abderrahim El Moataz. Springer Verlag, 2018. pp. 252-261 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{55bc196140ad4848866fa036c7fd02d0,

title = "Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging",

abstract = "In this paper, we propose an iterative contraction and merging framework (ICM) for semantic segmentation in indoor scenes. Given an input image and a raw depth image, we first derive the dense prediction map from a convolutional neural network (CNN) and a normal vector map from the depth image. By combining the RGB-D image with these two maps, we can guide the ICM process to produce a more accurate hierarchical segmentation tree in a bottom-up manner. After that, based on the hierarchical segmentation tree, we design a decision process which uses the dense prediction map as a reference to make the final decision of semantic segmentation. Experimental results show that the proposed method can generate much more accurate object boundaries if compared to the state-of-the-art methods.",

keywords = "Convolutional neural network, Iterative contraction and merging, RGB-D image, Semantic segmentation",

author = "Syu, {Jia Hao} and Cho, {Shih Hsuan} and Sheng-Jyh Wang and Li-Chun Wang",

year = "2018",

month = jan,

day = "1",

doi = "10.1007/978-3-319-94211-7_28",

language = "English",

isbn = "9783319942100",

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

publisher = "Springer Verlag",

pages = "252--261",

editor = "Driss Mammass and Fathallah Nouboud and Alamin Mansouri and {El Moataz}, Abderrahim",

booktitle = "Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings",

address = "Germany",

note = "null ; Conference date: 02-07-2018 Through 04-07-2018",

}

Syu, JH, Cho, SH, Wang, S-J & Wang, L-C 2018, Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging. in D Mammass, F Nouboud, A Mansouri & A El Moataz (eds), Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10884 LNCS, Springer Verlag, pp. 252-261, 8th International Conference on Image and Signal Processing, ICISP 2018, Cherbourg, France, 2/07/18. https://doi.org/10.1007/978-3-319-94211-7_28

Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging. / Syu, Jia Hao; Cho, Shih Hsuan; Wang, Sheng-Jyh et al.

Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings. ed. / Driss Mammass; Fathallah Nouboud; Alamin Mansouri; Abderrahim El Moataz. Springer Verlag, 2018. p. 252-261 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10884 LNCS).

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

TY - GEN

T1 - Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging

AU - Syu, Jia Hao

AU - Cho, Shih Hsuan

AU - Wang, Sheng-Jyh

AU - Wang, Li-Chun

PY - 2018/1/1

Y1 - 2018/1/1

N2 - In this paper, we propose an iterative contraction and merging framework (ICM) for semantic segmentation in indoor scenes. Given an input image and a raw depth image, we first derive the dense prediction map from a convolutional neural network (CNN) and a normal vector map from the depth image. By combining the RGB-D image with these two maps, we can guide the ICM process to produce a more accurate hierarchical segmentation tree in a bottom-up manner. After that, based on the hierarchical segmentation tree, we design a decision process which uses the dense prediction map as a reference to make the final decision of semantic segmentation. Experimental results show that the proposed method can generate much more accurate object boundaries if compared to the state-of-the-art methods.

AB - In this paper, we propose an iterative contraction and merging framework (ICM) for semantic segmentation in indoor scenes. Given an input image and a raw depth image, we first derive the dense prediction map from a convolutional neural network (CNN) and a normal vector map from the depth image. By combining the RGB-D image with these two maps, we can guide the ICM process to produce a more accurate hierarchical segmentation tree in a bottom-up manner. After that, based on the hierarchical segmentation tree, we design a decision process which uses the dense prediction map as a reference to make the final decision of semantic segmentation. Experimental results show that the proposed method can generate much more accurate object boundaries if compared to the state-of-the-art methods.

KW - Convolutional neural network

KW - Iterative contraction and merging

KW - RGB-D image

KW - Semantic segmentation

UR - http://www.scopus.com/inward/record.url?scp=85049666545&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-94211-7_28

DO - 10.1007/978-3-319-94211-7_28

M3 - Conference contribution

AN - SCOPUS:85049666545

SN - 9783319942100

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

SP - 252

EP - 261

BT - Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings

A2 - Mammass, Driss

A2 - Nouboud, Fathallah

A2 - Mansouri, Alamin

A2 - El Moataz, Abderrahim

PB - Springer Verlag

Y2 - 2 July 2018 through 4 July 2018

ER -

Syu JH, Cho SH, Wang S-J , Wang L-C. Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging. In Mammass D, Nouboud F, Mansouri A, El Moataz A, editors, Image and Signal Processing - 8th International Conference, ICISP 2018, Proceedings. Springer Verlag. 2018. p. 252-261. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-94211-7_28

Semantic segmentation of indoor-scene RGB-D images based on iterative contraction and merging

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this