DANF: Dual-Flow Aggregation Network for Curtain Wall Frame Real-Time Segmentation

Xiaoyu Xu; Decheng Wu; Rui Li; Xin Huang; Chul Hee Lee; Sheng Liu

doi:10.1109/CAC59555.2023.10449973

DANF: Dual-Flow Aggregation Network for Curtain Wall Frame Real-Time Segmentation

Xiaoyu Xu, Decheng Wu, Rui Li, Xin Huang, Chul Hee Lee, Sheng Liu

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

1 Scopus citations

Abstract

The precise positioning of curtain wall frames is a crucial step for the automated installation of curtain walls. This paper proposes a network (DANF) with a dual-flow aggregation structure to achieve semantic segmentation of curtain wall frames. We constructed a dataset (CWF) for the segmentation of curtain wall frames and analyzed the characteristics of both the curtain wall frames and the background environment. Our proposed network architecture combines the strengths of transformers for semantic information extraction and CNNs for high-resolution feature extraction. Moreover, we introduced a dual-flow aggregation module to effectively fuse the features derived from transformers and CNNs. The experimental results on the CWF dataset validate the powerful performance of our method.

Original language	English
Title of host publication	Proceedings - 2023 China Automation Congress, CAC 2023
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	8108-8113
Number of pages	6
ISBN (Electronic)	9798350303759
DOIs	https://doi.org/10.1109/CAC59555.2023.10449973
State	Published - 2023
Event	2023 China Automation Congress, CAC 2023 - Chongqing, China Duration: 17 Nov 2023 → 19 Nov 2023

Publication series

Name	Proceedings - 2023 China Automation Congress, CAC 2023

Conference

Conference	2023 China Automation Congress, CAC 2023
Country/Territory	China
City	Chongqing
Period	17/11/23 → 19/11/23

Bibliographical note

Publisher Copyright:
© 2023 IEEE.

Keywords

CNN
curtain wall frame
feature fusion
semantic segmentation
transformer

Access to Document

10.1109/CAC59555.2023.10449973

Cite this

Xu, X., Wu, D., Li, R., Huang, X., Lee, C. H., & Liu, S. (2023). DANF: Dual-Flow Aggregation Network for Curtain Wall Frame Real-Time Segmentation. In Proceedings - 2023 China Automation Congress, CAC 2023 (pp. 8108-8113). (Proceedings - 2023 China Automation Congress, CAC 2023). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CAC59555.2023.10449973

@inproceedings{26b548d73baa4ad18d6f2588ad61a436,

title = "DANF: Dual-Flow Aggregation Network for Curtain Wall Frame Real-Time Segmentation",

abstract = "The precise positioning of curtain wall frames is a crucial step for the automated installation of curtain walls. This paper proposes a network (DANF) with a dual-flow aggregation structure to achieve semantic segmentation of curtain wall frames. We constructed a dataset (CWF) for the segmentation of curtain wall frames and analyzed the characteristics of both the curtain wall frames and the background environment. Our proposed network architecture combines the strengths of transformers for semantic information extraction and CNNs for high-resolution feature extraction. Moreover, we introduced a dual-flow aggregation module to effectively fuse the features derived from transformers and CNNs. The experimental results on the CWF dataset validate the powerful performance of our method.",

keywords = "CNN, curtain wall frame, feature fusion, semantic segmentation, transformer",

author = "Xiaoyu Xu and Decheng Wu and Rui Li and Xin Huang and Lee, \{Chul Hee\} and Sheng Liu",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 China Automation Congress, CAC 2023 ; Conference date: 17-11-2023 Through 19-11-2023",

year = "2023",

doi = "10.1109/CAC59555.2023.10449973",

language = "English",

series = "Proceedings - 2023 China Automation Congress, CAC 2023",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "8108--8113",

booktitle = "Proceedings - 2023 China Automation Congress, CAC 2023",

address = "United States",

}

Xu, X, Wu, D, Li, R, Huang, X, Lee, CH & Liu, S 2023, DANF: Dual-Flow Aggregation Network for Curtain Wall Frame Real-Time Segmentation. in Proceedings - 2023 China Automation Congress, CAC 2023. Proceedings - 2023 China Automation Congress, CAC 2023, Institute of Electrical and Electronics Engineers Inc., pp. 8108-8113, 2023 China Automation Congress, CAC 2023, Chongqing, China, 17/11/23. https://doi.org/10.1109/CAC59555.2023.10449973

DANF: Dual-Flow Aggregation Network for Curtain Wall Frame Real-Time Segmentation. / Xu, Xiaoyu; Wu, Decheng; Li, Rui et al.
Proceedings - 2023 China Automation Congress, CAC 2023. Institute of Electrical and Electronics Engineers Inc., 2023. p. 8108-8113 (Proceedings - 2023 China Automation Congress, CAC 2023).

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

TY - GEN

T1 - DANF

T2 - 2023 China Automation Congress, CAC 2023

AU - Xu, Xiaoyu

AU - Wu, Decheng

AU - Li, Rui

AU - Huang, Xin

AU - Lee, Chul Hee

AU - Liu, Sheng

PY - 2023

Y1 - 2023

N2 - The precise positioning of curtain wall frames is a crucial step for the automated installation of curtain walls. This paper proposes a network (DANF) with a dual-flow aggregation structure to achieve semantic segmentation of curtain wall frames. We constructed a dataset (CWF) for the segmentation of curtain wall frames and analyzed the characteristics of both the curtain wall frames and the background environment. Our proposed network architecture combines the strengths of transformers for semantic information extraction and CNNs for high-resolution feature extraction. Moreover, we introduced a dual-flow aggregation module to effectively fuse the features derived from transformers and CNNs. The experimental results on the CWF dataset validate the powerful performance of our method.

AB - The precise positioning of curtain wall frames is a crucial step for the automated installation of curtain walls. This paper proposes a network (DANF) with a dual-flow aggregation structure to achieve semantic segmentation of curtain wall frames. We constructed a dataset (CWF) for the segmentation of curtain wall frames and analyzed the characteristics of both the curtain wall frames and the background environment. Our proposed network architecture combines the strengths of transformers for semantic information extraction and CNNs for high-resolution feature extraction. Moreover, we introduced a dual-flow aggregation module to effectively fuse the features derived from transformers and CNNs. The experimental results on the CWF dataset validate the powerful performance of our method.

KW - CNN

KW - curtain wall frame

KW - feature fusion

KW - semantic segmentation

KW - transformer

UR - http://www.scopus.com/inward/record.url?scp=85189355114&partnerID=8YFLogxK

U2 - 10.1109/CAC59555.2023.10449973

DO - 10.1109/CAC59555.2023.10449973

M3 - Conference contribution

AN - SCOPUS:85189355114

T3 - Proceedings - 2023 China Automation Congress, CAC 2023

SP - 8108

EP - 8113

BT - Proceedings - 2023 China Automation Congress, CAC 2023

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 17 November 2023 through 19 November 2023

ER -

DANF: Dual-Flow Aggregation Network for Curtain Wall Frame Real-Time Segmentation

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this