Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation

Chaewon Lee; Won Dong Jang; Chang Su Kim

doi:10.1109/APSIPAASC58517.2023.10317169

Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation

Chaewon Lee, Won Dong Jang, Chang Su Kim

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

Abstract

BRS is the first CNN-based interactive image segmentation algorithm to refine segmentation results based on a backpropagation scheme. In this paper, we give a detailed description of how BRS operates and demonstrate how to implement the algorithm. In BRS, user-provided clicks are first converted into interaction maps, which are then concatenated with the RGB image and provided as input to a segmentation network. In the test phase, performing the forward pass in the network generates an initial segmentation map. However, the user-annotated pixels may be mislabeled in the initial result. BRS refines this result by correcting the mislabeled pixels. We implement this BRS algorithm in PyTorch and publish the source codes. Moreover, we first show that BRS can reach the perfect IoU ratio of 1.0 in most cases and delineate objects more accurately than a variant of BRS, called f-BRS.

Original language	English
Title of host publication	2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	696-702
Number of pages	7
ISBN (Electronic)	9798350300673
DOIs	https://doi.org/10.1109/APSIPAASC58517.2023.10317169
Publication status	Published - 2023
Event	2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023 - Taipei, Taiwan, Province of China Duration: 2023 Oct 31 → 2023 Nov 3

Publication series

Name	2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023

Conference

Conference	2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023
Country/Territory	Taiwan, Province of China
City	Taipei
Period	23/10/31 → 23/11/3

Bibliographical note

Publisher Copyright:
© 2023 IEEE.

ASJC Scopus subject areas

Hardware and Architecture
Signal Processing
Artificial Intelligence
Computer Science Applications

Access to Document

10.1109/APSIPAASC58517.2023.10317169

Cite this

Lee, C., Jang, W. D., & Kim, C. S. (2023). Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation. In 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023 (pp. 696-702). (2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/APSIPAASC58517.2023.10317169

Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation. / Lee, Chaewon; Jang, Won Dong; Kim, Chang Su.
2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023. Institute of Electrical and Electronics Engineers Inc., 2023. p. 696-702 (2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023).

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

Lee, C, Jang, WD & Kim, CS 2023, Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation. in 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023. 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023, Institute of Electrical and Electronics Engineers Inc., pp. 696-702, 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023, Taipei, Taiwan, Province of China, 23/10/31. https://doi.org/10.1109/APSIPAASC58517.2023.10317169

Lee C, Jang WD, Kim CS. Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation. In 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023. Institute of Electrical and Electronics Engineers Inc. 2023. p. 696-702. (2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023). doi: 10.1109/APSIPAASC58517.2023.10317169

Lee, Chaewon ; Jang, Won Dong ; Kim, Chang Su. / Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation. 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023. Institute of Electrical and Electronics Engineers Inc., 2023. pp. 696-702 (2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023).

@inproceedings{941452b28597422ca5f9ee91c6dc4e79,

title = "Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation",

abstract = "BRS is the first CNN-based interactive image segmentation algorithm to refine segmentation results based on a backpropagation scheme. In this paper, we give a detailed description of how BRS operates and demonstrate how to implement the algorithm. In BRS, user-provided clicks are first converted into interaction maps, which are then concatenated with the RGB image and provided as input to a segmentation network. In the test phase, performing the forward pass in the network generates an initial segmentation map. However, the user-annotated pixels may be mislabeled in the initial result. BRS refines this result by correcting the mislabeled pixels. We implement this BRS algorithm in PyTorch and publish the source codes. Moreover, we first show that BRS can reach the perfect IoU ratio of 1.0 in most cases and delineate objects more accurately than a variant of BRS, called f-BRS.",

author = "Chaewon Lee and Jang, {Won Dong} and Kim, {Chang Su}",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023 ; Conference date: 31-10-2023 Through 03-11-2023",

year = "2023",

doi = "10.1109/APSIPAASC58517.2023.10317169",

language = "English",

series = "2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023",

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

pages = "696--702",

booktitle = "2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023",

}

TY - GEN

T1 - Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation

AU - Lee, Chaewon

AU - Jang, Won Dong

AU - Kim, Chang Su

PY - 2023

Y1 - 2023

N2 - BRS is the first CNN-based interactive image segmentation algorithm to refine segmentation results based on a backpropagation scheme. In this paper, we give a detailed description of how BRS operates and demonstrate how to implement the algorithm. In BRS, user-provided clicks are first converted into interaction maps, which are then concatenated with the RGB image and provided as input to a segmentation network. In the test phase, performing the forward pass in the network generates an initial segmentation map. However, the user-annotated pixels may be mislabeled in the initial result. BRS refines this result by correcting the mislabeled pixels. We implement this BRS algorithm in PyTorch and publish the source codes. Moreover, we first show that BRS can reach the perfect IoU ratio of 1.0 in most cases and delineate objects more accurately than a variant of BRS, called f-BRS.

AB - BRS is the first CNN-based interactive image segmentation algorithm to refine segmentation results based on a backpropagation scheme. In this paper, we give a detailed description of how BRS operates and demonstrate how to implement the algorithm. In BRS, user-provided clicks are first converted into interaction maps, which are then concatenated with the RGB image and provided as input to a segmentation network. In the test phase, performing the forward pass in the network generates an initial segmentation map. However, the user-annotated pixels may be mislabeled in the initial result. BRS refines this result by correcting the mislabeled pixels. We implement this BRS algorithm in PyTorch and publish the source codes. Moreover, we first show that BRS can reach the perfect IoU ratio of 1.0 in most cases and delineate objects more accurately than a variant of BRS, called f-BRS.

UR - http://www.scopus.com/inward/record.url?scp=85180014219&partnerID=8YFLogxK

U2 - 10.1109/APSIPAASC58517.2023.10317169

DO - 10.1109/APSIPAASC58517.2023.10317169

M3 - Conference contribution

AN - SCOPUS:85180014219

T3 - 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023

SP - 696

EP - 702

BT - 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA ASC 2023

Y2 - 31 October 2023 through 3 November 2023

ER -

Implementation and Analysis on Backpropagating Refinement Scheme for Interactive Image Segmentation

Abstract

Publication series

Conference

Bibliographical note

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this