Attention Deeplabv3+: Multi-level Context Attention Mechanism for Skin Lesion Segmentation

Reza Azad; Maryam Asadi-Aghbolaghi; Mahmood Fathy; Sergio Escalera

doi:10.1007/978-3-030-66415-2_16

Attention Deeplabv3+: Multi-level Context Attention Mechanism for Skin Lesion Segmentation

Reza Azad, Maryam Asadi-Aghbolaghi^*, Mahmood Fathy, Sergio Escalera

^*Corresponding author for this work

Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review

68 Citations (Scopus)

Abstract

Skin lesion segmentation is a challenging task due to the large variation of anatomy across different cases. In the last few years, deep learning frameworks have shown high performance in image segmentation. In this paper, we propose Attention Deeplabv3+, an extended version of Deeplabv3+ for skin lesion segmentation by employing the idea of attention mechanism in two stages. We first capture the relationship between the channels of a set of feature maps by assigning a weight for each channel (i.e., channels attention). Channel attention allows the network to emphasize more on the informative and meaningful channels by a context gating mechanism. We also exploit the second level attention strategy to integrate different layers of the atrous convolution. It helps the network to focus on the more relevant field of view to the target. The proposed model is evaluated on three datasets ISIC 2017, ISIC 2018, and PH², achieving state-of-the-art performance.

Original language	English
Title of host publication	Computer Vision – ECCV 2020 Workshops, Proceedings
Editors	Adrien Bartoli, Andrea Fusiello
Number of pages	16
Publisher	Springer
Publication date	2020
Pages	251-266
ISBN (Print)	9783030664145
DOIs	https://doi.org/10.1007/978-3-030-66415-2_16
Publication status	Published - 2020
Event	Workshops held at the 16th European Conference on Computer Vision, ECCV 2020 - Glasgow, United Kingdom Duration: 23 Aug 2020 → 28 Aug 2020

Conference

Conference	Workshops held at the 16th European Conference on Computer Vision, ECCV 2020
Country/Territory	United Kingdom
City	Glasgow
Period	23/08/2020 → 28/08/2020

Series	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	12535 LNCS
ISSN	0302-9743

Bibliographical note

Funding Information:
This work has been partially supported by the Spanish project PID2019-105093GB-I00 (MINECO/FEDER, UE) and CERCA Pro-gramme/Generalitat de Catalunya, and ICREA under the ICREA Academia programme. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPU used for this research.

Funding Information:
Acknowledgment. This work has been partially supported by the Spanish project PID2019-105093GB-I00 (MINECO/FEDER, UE) and CERCA Programme/Generalitat de Catalunya, and ICREA under the ICREA Academia programme. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPU used for this research.

Publisher Copyright:
© 2020, Springer Nature Switzerland AG.

Keywords

Attention mechanism
Deeplabv3+
Medical image segmentation

Access to Document

10.1007/978-3-030-66415-2_16

AUB Link

Search for the material in Aalborg University Library's search engine

Cite this

Azad, Reza ; Asadi-Aghbolaghi, Maryam ; Fathy, Mahmood et al. / Attention Deeplabv3+ : Multi-level Context Attention Mechanism for Skin Lesion Segmentation. Computer Vision – ECCV 2020 Workshops, Proceedings. editor / Adrien Bartoli ; Andrea Fusiello. Springer, 2020. pp. 251-266 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12535 LNCS).

@inproceedings{8104ee869c4842e29fa7c9528633c6bf,

title = "Attention Deeplabv3+: Multi-level Context Attention Mechanism for Skin Lesion Segmentation",

abstract = "Skin lesion segmentation is a challenging task due to the large variation of anatomy across different cases. In the last few years, deep learning frameworks have shown high performance in image segmentation. In this paper, we propose Attention Deeplabv3+, an extended version of Deeplabv3+ for skin lesion segmentation by employing the idea of attention mechanism in two stages. We first capture the relationship between the channels of a set of feature maps by assigning a weight for each channel (i.e., channels attention). Channel attention allows the network to emphasize more on the informative and meaningful channels by a context gating mechanism. We also exploit the second level attention strategy to integrate different layers of the atrous convolution. It helps the network to focus on the more relevant field of view to the target. The proposed model is evaluated on three datasets ISIC 2017, ISIC 2018, and PH2, achieving state-of-the-art performance.",

keywords = "Attention mechanism, Deeplabv3+, Medical image segmentation",

author = "Reza Azad and Maryam Asadi-Aghbolaghi and Mahmood Fathy and Sergio Escalera",

note = "Funding Information: This work has been partially supported by the Spanish project PID2019-105093GB-I00 (MINECO/FEDER, UE) and CERCA Pro-gramme/Generalitat de Catalunya, and ICREA under the ICREA Academia programme. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPU used for this research. Funding Information: Acknowledgment. This work has been partially supported by the Spanish project PID2019-105093GB-I00 (MINECO/FEDER, UE) and CERCA Programme/Generalitat de Catalunya, and ICREA under the ICREA Academia programme. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPU used for this research. Publisher Copyright: {\textcopyright} 2020, Springer Nature Switzerland AG.; Workshops held at the 16th European Conference on Computer Vision, ECCV 2020 ; Conference date: 23-08-2020 Through 28-08-2020",

year = "2020",

doi = "10.1007/978-3-030-66415-2_16",

language = "English",

isbn = "9783030664145",

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

publisher = "Springer",

pages = "251--266",

editor = "Adrien Bartoli and Andrea Fusiello",

booktitle = "Computer Vision – ECCV 2020 Workshops, Proceedings",

address = "Germany",

}

Azad, R, Asadi-Aghbolaghi, M, Fathy, M & Escalera, S 2020, Attention Deeplabv3+: Multi-level Context Attention Mechanism for Skin Lesion Segmentation. in A Bartoli & A Fusiello (eds), Computer Vision – ECCV 2020 Workshops, Proceedings. Springer, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12535 LNCS, pp. 251-266, Workshops held at the 16th European Conference on Computer Vision, ECCV 2020, Glasgow, United Kingdom, 23/08/2020. https://doi.org/10.1007/978-3-030-66415-2_16

Attention Deeplabv3+: Multi-level Context Attention Mechanism for Skin Lesion Segmentation. / Azad, Reza; Asadi-Aghbolaghi, Maryam; Fathy, Mahmood et al.
Computer Vision – ECCV 2020 Workshops, Proceedings. ed. / Adrien Bartoli; Andrea Fusiello. Springer, 2020. p. 251-266 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12535 LNCS).

Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review

TY - GEN

T1 - Attention Deeplabv3+

T2 - Workshops held at the 16th European Conference on Computer Vision, ECCV 2020

AU - Azad, Reza

AU - Asadi-Aghbolaghi, Maryam

AU - Fathy, Mahmood

AU - Escalera, Sergio

N1 - Funding Information: This work has been partially supported by the Spanish project PID2019-105093GB-I00 (MINECO/FEDER, UE) and CERCA Pro-gramme/Generalitat de Catalunya, and ICREA under the ICREA Academia programme. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPU used for this research. Funding Information: Acknowledgment. This work has been partially supported by the Spanish project PID2019-105093GB-I00 (MINECO/FEDER, UE) and CERCA Programme/Generalitat de Catalunya, and ICREA under the ICREA Academia programme. We gratefully acknowledge the support of NVIDIA Corporation with the donation of the GPU used for this research. Publisher Copyright: © 2020, Springer Nature Switzerland AG.

PY - 2020

Y1 - 2020

N2 - Skin lesion segmentation is a challenging task due to the large variation of anatomy across different cases. In the last few years, deep learning frameworks have shown high performance in image segmentation. In this paper, we propose Attention Deeplabv3+, an extended version of Deeplabv3+ for skin lesion segmentation by employing the idea of attention mechanism in two stages. We first capture the relationship between the channels of a set of feature maps by assigning a weight for each channel (i.e., channels attention). Channel attention allows the network to emphasize more on the informative and meaningful channels by a context gating mechanism. We also exploit the second level attention strategy to integrate different layers of the atrous convolution. It helps the network to focus on the more relevant field of view to the target. The proposed model is evaluated on three datasets ISIC 2017, ISIC 2018, and PH2, achieving state-of-the-art performance.

AB - Skin lesion segmentation is a challenging task due to the large variation of anatomy across different cases. In the last few years, deep learning frameworks have shown high performance in image segmentation. In this paper, we propose Attention Deeplabv3+, an extended version of Deeplabv3+ for skin lesion segmentation by employing the idea of attention mechanism in two stages. We first capture the relationship between the channels of a set of feature maps by assigning a weight for each channel (i.e., channels attention). Channel attention allows the network to emphasize more on the informative and meaningful channels by a context gating mechanism. We also exploit the second level attention strategy to integrate different layers of the atrous convolution. It helps the network to focus on the more relevant field of view to the target. The proposed model is evaluated on three datasets ISIC 2017, ISIC 2018, and PH2, achieving state-of-the-art performance.

KW - Attention mechanism

KW - Deeplabv3+

KW - Medical image segmentation

UR - http://www.scopus.com/inward/record.url?scp=85101325812&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-66415-2_16

DO - 10.1007/978-3-030-66415-2_16

M3 - Article in proceeding

AN - SCOPUS:85101325812

SN - 9783030664145

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

SP - 251

EP - 266

BT - Computer Vision – ECCV 2020 Workshops, Proceedings

A2 - Bartoli, Adrien

A2 - Fusiello, Andrea

PB - Springer

Y2 - 23 August 2020 through 28 August 2020

ER -

Azad R, Asadi-Aghbolaghi M, Fathy M, Escalera S. Attention Deeplabv3+: Multi-level Context Attention Mechanism for Skin Lesion Segmentation. In Bartoli A, Fusiello A, editors, Computer Vision – ECCV 2020 Workshops, Proceedings. Springer. 2020. p. 251-266. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol. 12535 LNCS). doi: 10.1007/978-3-030-66415-2_16

Attention Deeplabv3+: Multi-level Context Attention Mechanism for Skin Lesion Segmentation

Abstract

Conference

Bibliographical note

Keywords

Access to Document

AUB Link

Other files and links

Fingerprint

Cite this