Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning

Ci Yin Zhang; Wei Ta Chu

doi:10.1145/3595916.3626401

Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning

Ci Yin Zhang, Wei Ta Chu

Department of Computer Science and Information Engineering

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

Abstract

Existing methods for manga character re-identification primarily rely on facial information, overlooking the unique characteristics of characters’ bodies and failing to address common challenges like occlusion by speech balloons and incomplete body parts. To tackle these issues, we propose a method called Occlusion-Aware Manga Character Re-identification (OAM-ReID) with self-paced contrastive learning, which leverages annotated body data from the Manga109 dataset for training. By synthesizing data with occluded speech balloons and incomplete bodies, we empower the framework to be aware of occlusion, so that more effective feature representations are learnt. Experimental results show that this approach outperforms the state-of-the-art person ReID method.

Original language	English
Title of host publication	Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023
Publisher	Association for Computing Machinery, Inc
ISBN (Electronic)	9798400702051
DOIs	https://doi.org/10.1145/3595916.3626401
Publication status	Published - 2023 Dec 6
Event	5th ACM International Conference on Multimedia in Asia, MMAsia 2023 - Hybrid, Tainan, Taiwan Duration: 2023 Dec 6 → 2023 Dec 8

Publication series

Name	Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023

Conference

Conference	5th ACM International Conference on Multimedia in Asia, MMAsia 2023
Country/Territory	Taiwan
City	Hybrid, Tainan
Period	23-12-06 → 23-12-08

All Science Journal Classification (ASJC) codes

Computer Graphics and Computer-Aided Design
Human-Computer Interaction

Access to Document

10.1145/3595916.3626401

Cite this

Zhang, C. Y., & Chu, W. T. (2023). Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning. In Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023 Article 29 (Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023). Association for Computing Machinery, Inc. https://doi.org/10.1145/3595916.3626401

@inproceedings{e8ad675a17904c20a05ad32b5a44d048,

title = "Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning",

abstract = "Existing methods for manga character re-identification primarily rely on facial information, overlooking the unique characteristics of characters{\textquoteright} bodies and failing to address common challenges like occlusion by speech balloons and incomplete body parts. To tackle these issues, we propose a method called Occlusion-Aware Manga Character Re-identification (OAM-ReID) with self-paced contrastive learning, which leverages annotated body data from the Manga109 dataset for training. By synthesizing data with occluded speech balloons and incomplete bodies, we empower the framework to be aware of occlusion, so that more effective feature representations are learnt. Experimental results show that this approach outperforms the state-of-the-art person ReID method.",

author = "Zhang, {Ci Yin} and Chu, {Wei Ta}",

note = "Publisher Copyright: {\textcopyright} 2023 Copyright held by the owner/author(s).; 5th ACM International Conference on Multimedia in Asia, MMAsia 2023 ; Conference date: 06-12-2023 Through 08-12-2023",

year = "2023",

month = dec,

day = "6",

doi = "10.1145/3595916.3626401",

language = "English",

series = "Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023",

publisher = "Association for Computing Machinery, Inc",

booktitle = "Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023",

}

Zhang, CY & Chu, WT 2023, Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning. in Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023., 29, Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023, Association for Computing Machinery, Inc, 5th ACM International Conference on Multimedia in Asia, MMAsia 2023, Hybrid, Tainan, Taiwan, 23-12-06. https://doi.org/10.1145/3595916.3626401

Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning. / Zhang, Ci Yin; Chu, Wei Ta.
Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023. Association for Computing Machinery, Inc, 2023. 29 (Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023).

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

TY - GEN

T1 - Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning

AU - Zhang, Ci Yin

AU - Chu, Wei Ta

PY - 2023/12/6

Y1 - 2023/12/6

N2 - Existing methods for manga character re-identification primarily rely on facial information, overlooking the unique characteristics of characters’ bodies and failing to address common challenges like occlusion by speech balloons and incomplete body parts. To tackle these issues, we propose a method called Occlusion-Aware Manga Character Re-identification (OAM-ReID) with self-paced contrastive learning, which leverages annotated body data from the Manga109 dataset for training. By synthesizing data with occluded speech balloons and incomplete bodies, we empower the framework to be aware of occlusion, so that more effective feature representations are learnt. Experimental results show that this approach outperforms the state-of-the-art person ReID method.

AB - Existing methods for manga character re-identification primarily rely on facial information, overlooking the unique characteristics of characters’ bodies and failing to address common challenges like occlusion by speech balloons and incomplete body parts. To tackle these issues, we propose a method called Occlusion-Aware Manga Character Re-identification (OAM-ReID) with self-paced contrastive learning, which leverages annotated body data from the Manga109 dataset for training. By synthesizing data with occluded speech balloons and incomplete bodies, we empower the framework to be aware of occlusion, so that more effective feature representations are learnt. Experimental results show that this approach outperforms the state-of-the-art person ReID method.

UR - http://www.scopus.com/inward/record.url?scp=85182949998&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85182949998&partnerID=8YFLogxK

U2 - 10.1145/3595916.3626401

DO - 10.1145/3595916.3626401

M3 - Conference contribution

AN - SCOPUS:85182949998

T3 - Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023

BT - Proceedings of the 5th ACM International Conference on Multimedia in Asia, MMAsia 2023

PB - Association for Computing Machinery, Inc

T2 - 5th ACM International Conference on Multimedia in Asia, MMAsia 2023

Y2 - 6 December 2023 through 8 December 2023

ER -

Occlusion-Aware Manga Character Re-identification with Self-Paced Contrastive Learning

Abstract

Publication series

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this